For version 12.2 release of WASD VMS Web Services.
Published January 2024
Document generated using wasDOC version 2.0.0
This document introduces the WASD Web Services package and provides detailed installation, update and configuration instructions.
For installation and update details see WASD Web Services - Installation
For detailed configuration information see WASD Web Services - Configuration
For the more significant features and facilities available see WASD Web Services - Features
And for a description of WASD Web document, SSI and directory listing behaviours and options, WASD Web Services - Environment
WASD VMS Web Services – Copyright © 1996-2024 Mark G. Daniel
Licensed under the Apache License, Version 2.0 (the "License");
https://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.
Mark.Daniel@wasd.vsm.com.au
A pox on the houses of all spamers. Make that two poxes.
All copyright and trademarks within this document belong to their rightful owners. See 13. Attribution and Acknowledgement.
This is a static (file), single document.
Alternative multi-part static and
dynamic documents.
Links followed by ⤤ open in a new page.
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
This document is not a general tutorial on authoring scripts, CGI or any other. A large number of references in the popular computing press covers all aspects of this technology, usually quite comprehensively. The information here is about the specifics of scripting in the WASD environment, which is generally very much like any other implementation, VMS or otherwise (although there are always annoying idiosyncracies, see 2.4 CGI Function Library for a partial solution to smoothing out some of these wrinkles for VMS environments).
Scripts are mechanisms for creating Web applications and services, sending data to (and often receiving data from) a client, extending the capabilities of the basic server. Scripts execute in processes and accounts separate from the actual HTTP server but under its control and interacting with it.
WASD manages a script's execution environment as an independent detached process created and managed by the HTTP server. By configuration a script also can be executed in a process created using DECnet. Originally, WASD scripted in subprocesses but this is obsolete. There is no analogue of the Apache loadable module executed within the server process itself.
WASD scripting can be deployed in a number of environments. Other sections cover the specifics of these. Don't become bewildered or be put off by all these apparent options, they are basically variations on a CGI theme.
It is strongly recommended to execute scripts in an account distinct from that executing the server. This minimises the risk of both unintentional and malicious interference with server operation through either Inter-Process Communication (IPC) or scripts manipulating files used by the server.
The default WASD installation creates two such accounts, with distinct UICs, usernames and default directory space. The UICs and home areas can be specified differently to the displayed defaults. Nothing should be assumed or read into the scripting account username – it's just a username.
Username | UIC | Default | Description |
---|---|---|---|
HTTP$SERVER | [077,001] | WASD_ROOT:[HTTP$SERVER] | Server Account |
HTTP$NOBODY | [076,001] | WASD_ROOT:[HTTP$NOBODY] | Scripting Account |
During startup the server checks for the existence of the default scripting account and automatically configures itself to use this for scripting. If it is not present it falls-back to using the server account. Other account names can be used if the startup procedures are modified accordingly. The default scripting username may be overridden using the /SCRIPT=AS=<username> qualifier (see Server Startup in WASD Installation). The default scripting account cannot be a member of the SYSTEM group and cannot have any privilege other than NETMBX and TMPMBX (‘Privileged User Scripting’ in 1.2.3.1 Persona Scripting describes how to configure to allow this).
Scripting under a separate account is not available with subprocess scripting and is distinct from PERSONA scripting (even though it uses the same mechanism, see below).
Process creation under the VMS operating system is notoriously slow and expensive. This is an inescapable overhead when scripting via child processes. An obvious strategy is to avoid, at least as much as possible, the creation of these processes. The only way to do this is to share processes between multiple scripts/requests, addressing the attendant complications of isolating potential interactions between requests. These could occur through changes made by any script to the process' enviroment. For VMS this involves symbol and logical name creation, and files opened at the DCL level. In reality few scripts need to make logical name changes and symbols are easily removed between uses. DCL-opened files are a little more problematic, but again, in reality most scripts doing file manipulation will be images.
A reasonable assumption is that for almost all environments scripts can quite safely share processes with great benefit to response latency and system impact (see Server Performance of WASD Features) for a table with some comparative performances). If the local environment requires absolute script isolation for some reason then this process-persistance may easily be disabled with a consequent trade-off on performance.
The term zombie is used to describe processes when persisting between uses (the reason should be obvious, they are neither "alive" (processing a request) nor are they "dead" (deleted :^) Zombie processes have a finite time to exist (non-life-time?) before they are automatically run-down (see below). This keeps process clutter on the system to a minimum.
Scripting processes are created on-demand, within configuration limits and timeout periods. There are no arbitrary limits, only system resource limits, on the number of scripting processes. WASD_CONFIG_GLOBAL directives control the configuration limits of these (see Global Configuration of WASD Configuration).
Scripting processes of all kinds (CGI, CGIplus and RTE) are created on-demand up until [DclHardLimit] is reached. If all scripting processes are busy with requests at that limit then the server provides a 503 (too busy) response.
If there are more than [DclSoftLimit] scripting processes then the least-recently-used of any idle processes (those not currently processing a request) are proactively run-down until the soft-limit is reached. This provides head-room for the immediate creation of additional scripting processes for new requests that cannot be satisfied from currently instantiated processes. Soft-limit should of course be configured less than hard-limit (and if not WASD makes it that way).
Idle scripting processes (those not having been given a request to process) are proactively run-down (see 1.2.6 Script Process Run-Down) after configured periods.
[DclZombieLifeTime] specifies the period in minutes a CGI scripting process can remain idle.
[DclCgiPlusLifeTime] specifies the period a CGIplus script (inside a CGIplus scripting process) or a RTE process can remain idle.
If requests being serviced by scripts drop to zero for a period (governed by the above lifetimes) then eventually all scripting processes should be run-down leaving only the server process.
Script process naming occurs with the intention of providing some obvious indication of the activity of the process.
During script process creation the name momentarily is seen using the schema WASD:80-1 ... WASD:80-999 but quickly moves to WASD:80-pid where pid are four hex digits of the LSW of the process PID (essentially to ensure uniqueness).
Historically, WASD scripting processes have been named sequentially WASD:80-1 ... WASD:80-999. A script could change that and it would remain even if the process was reused for another script. Now at conclusion of script processing v12 reverts the name.
The v12 schema may be disabled by defining the logical name WASD_DCL_PRCNAM_PRE12 prior to server start.
The default is for WASD to execute scripts in detached processes created and managed completely independently of the server process itself. This offers a significant number of advantages
Creation of a process is expensive in terms of system resources and initial invocation response latency (particularly if extensive login procedures are required), but this quickly becomes negligable as most script processes are used multiple times for successive scripts and/or requests.
There is no WASD analogue of the Apache loadable module executed within the server process itself.
With detached processes the server must explicitly ensure that each scripting process is removed from the system during server shutdown (cf. subprocesses where the VMS executive provides that automatically). This is performed by the server exit handler. With VMS it is possible to bypass the exit handler (using a $DELPRC or the equivalent $STOP/ID= for instance), making it possible for "orphaned" scripting processes to remain – and potentially accumulate on the system!
To address this possibility the server scans the system for candidate processes during each startup. These are identified by a terminal mailbox (SYS$COMMAND device), and then further that the mailbox has an ACL with two entries; the first identifying itself as a WASD HTTPd mailbox and the second allowing access to the account the script is being executed under. Such a device ACL looks like the following example.
This rights identifier is generated from the server process name and is therefore system-unique (so multiple autonomous servers will not accidentally cleanup the script processes of others), and is created during server startup if it does not already exist. For example, if the process name was "HTTPd:80" (the default for a standard service) the rights identifier name would be "WASD_HTTPD_80" (as shown in the example above).
Scripting processes are created through the full "LOGINOUT" life-cycle and execute all system and account LOGIN procedures. Although immune to the effects of most actions within these procedures, and absorbing any output generated during this phase of the process life-cycle, some consideration should be given to minimising the LOGIN procedure paths. This can noticably reduce initial script latency on less powerful platforms.
The usual recommendations for non-interactive LOGIN procedures apply for script environments as well. Avoid interactive-only commands and reduce unnecessary interactive process environment setup. This is usually accomplished though code structures such as the following
WASD scripting processes can be specifically detected using DCL tests similar to the following. This checks the mode, that standard output is a mailbox, and the process name. These are fairly reliable (but not absolutely infallible) indicators.
There are advantages in running a script under a non-server account. The most obvious of these is the security isolation it offers with respect to the rest of the Web and server environment. It also means that the server account does not need to be resourced especially for any particularly demanding application.
The $PERSONA functionality must be explicitly enabled at server startup using the /PERSONA qualifier (Server Account and Environment of WASD Installation). The ability for the server to be able to execute scripts under any user account is a very powerful (and potentially dangerous) capability, and so is designed that the site administrator must explicitly and deliberately enable the functionality. Configuration files need to be rigorously protected against unauthorized modification.
A specific script or directory of scripts can be designated for execution under a specified account using the WASD_CONFIG_MAP configuration file set script=as= mapping rule. The following example illustrates the essentials.
Access to package scripting directories (e.g. WASD_ROOT:[CGI-BIN]) is controlled by ACLs and possession of the rights identifier WASD_HTTP_NOBODY. If a non-server account requires access to these areas it too will need to be granted this identifier.
In some situations it may be desirable to allow the average Web user to experiment with or implement scripts. If the set script=as= mapping rule specifies a tilde character then for a user request the mapped SYSUAF username is substituted. Note that this requires the script to be colocated with the user account Web location and that the script is run under that account.
The following example shows the essentials of setting up a user environment where access to a subdirectory in the user's home directory, [.WWW] with script's located in a subdirectory of that, [.WWW.CGI-BIN].
Where the site administrator has less than full control of the scripting environment it may be prudent to put some constraints on the quantity of resource that potentially can be consumed by non-core or errant scripting. The following WASD_CONFIG_MAP rule allows the "maximum" CPU time consumed by a single script to be constrained.
Note that this is on a per-script basis, contrasted to the sort of limit a CPULM-type constraint would place on a scripting process.
The following WASD_CONFIG_GLOBAL rule specifies at which priority the scripting process executes. This can be used to provide the server and its infrastructure an advantage over user scripts.
If the set script=as= mapping rule specifies a dollar then a request that has been SYSUAF authenticated has the SYSUAF username substituted. Note that the script itself can be located anywhere provided the user account has read and/or execute access to the area and file.
If the script has not been subject to SYSUAF authorization then this causes the script activation to fail. To allow authenticated requests to be executed under the corresponding VMS account and non-authenticated requests to script as the usual server/scripting account use the following variant.
If the server startup included /PERSONA=AUTHORIZED then only requests that have been subject to HTTP authorization and authentication are allowed to script under non-server accounts.
By default a privileged account cannot be used for scripting. This is done to reduce the chance of unintended capabilities when executing scripts. With additional configuration it is possible to use such accounts. Great care should be exercised when undertaking this.
To allow the server to activate a script using a privileged account the keyword /PERSONA=RELAXED must be used with the persona startup qualifier. If the keywords /PERSONA=RELAXED=AUTHORIZED are used then privileged accounts are allowed for scripting but only if the request has been subject to HTTP authorization and authentication.
By default, activating the /PERSONA server startup qualifier allows all the modes described above to be deployed using appropriate mapping rules. Of course there may be circumstances where such broad capabilities are inappropriate or otherwise undesirable. It is possible to control which user accounts are able to be used in this fashion with a rights identifier. Only those accounts granted the identifier can have scripts activated under them. This means all accounts … including the server account!
This is enabled by specifying the name of a rights identifier as a parameter to the /PERSONA qualifier. This may be any identifier but the one shown in the following example is probably as good as any.
This identifier could be created using the following commands
Meaningful combinations of startup parameters are possible:
When detached processes are created they can be assigned differing priorities depending on the origin and purpose. The objective is to give the server process a slight advantage when competing with scripts for system resources. This allows the server to respond to new requests more quickly (reducing latency) even if a script may then take some time to complete the request.
The allocation of base process priorities is determined from the WASD_CONFIG_GLOBAL [DclDetachProcessPriority] configuration directive, which takes one or two (comma-separated) integers that determine how many priorities lower than the server scripting processes are created. The first integer determines server processes. A second, if supplied, determines user scripts. User scripts may never be a higher priority that server scripts. The following provides example directives.
Scripts executed under the server account, or those created using a mapped username (i.e. "script=as=username"), have a process priority set by the first/only integer.
Scripts activated from user mappings (i.e. "script=as=~" or "script=as=$") have a process priority set by any second integer, or fall back to the priority of the first/only integer.
For standard CGI and CGIplus script the script process' default device and directory is established using a SET DEFAULT command immediately before activating the script. This default is derived from the script file specification.
An alternative default location may be specified using the mapping rule shown in the following example.
The default may be specified in VMS or Unix file system syntax as appropriate. If in Unix syntax (beginning with a forward-slash) no SET DEFAULT is performed using DCL. The script itself must access this value using the SCRIPT_DEFAULT CGI variable and perform a chdir().
On platforms where the Extended File Specification (EFS) is supported a SET PROCESS /PARSE=EXTENDED or SET PROCESS /PARSE=TRADITIONAL is executed by the scripting process before script activation depending on whether the script path is located on an ODS-2 or ODS-5 volume.
The server can stop a script process at any point, although this is generally done at a time and in such a way as to eliminate any disruption to request processing. Reasons for the server running-down a script process.
In running down a script process the server must both update its own internal data structures as well as manage the run-down of the script process environment and script process itself. These are the steps.
Generally CGIplus processes delete themselves immediately. With standard CGI scripts executing an image it may take from zero to a few seconds for the image run-down to be detected by the server. A script is allowed approximately one minute to complete the image run-down.
If a client disconnects from a running script (by hitting the browser Stop button, or selecting another hyperlink) the loss of network connectivity is detected by the server at the next output write.
Generally it is necessary for there to be some mechanism for a client to stop long-running (and presumably resource consuming) scripts. Network disconnection is the only viable one. Experience would indicate however that most scripts are short running and most disconnections are due to clients changing their minds about waiting for a page to build or having seen the page superstructure moving on to something else.
With these considerations in mind there is significiant benefit in not running-down a script immediately the client disconnection is detected. A short wait will result in most scripts completing their output elegantly (the script itself unaware the output is not being transmitted on to the client), and in the case of persistent scripts remaining available for the next request, or for standard CGI the process remaining for use in the next CGI script.
The period allowing the script to complete its processing may be set using the WASD_CONFIG_GLOBAL configuration directive [DclBitBucketTimeout]. It should be set to say fifteen seconds, or whatever is appropriate to the local site.
NB. "Bit-bucket" is a common term for the place discarded data is stored. :^)
Script proctoring proactively creates and maintains the specified minimum number of scripting processes, configured persistent scripts, and scripting environments (RTEs). It is primarily intended for those environments that have significant startup latency but can also be used to maintain idle scripting processes ready for immediate use.
The script proctor initially instantiates configured items during server startup and before enabling request acceptance and processing.
Then during subsequent request processing, at each scripting process run-down it scans current DCL task list counting the number of instances of each configured item. The proctor facility can differentiate between idle and active instances of the script/RTE and will optionally maintain a specified number of idle processes in addition to any currently active. If fewer than the configured requirement(s) one or more new processes are instantiated.
It is possible (and probably likely) that a proctored script specification will at some stage fail to activate the script (activation specification error, script unavailable for some reason, etc.) which would lead to a runaway series of attempts to proctor with each process exit. To help avoid this situation proctored processes that exit before successfully completing initial startup are quickly suppressed from further proctoring action. This suppression then more slowly times out, again allowing proctoring for that item.
Proctored scripts and RTEs contain nothing of the usual request-related environment. No CGI variables to speak of, no service, no request method, nothing! This means that rules used for proctor activations must be outside all virtual service conditionals (i.e. outside of any specific [[service:port]] in the rules, can be inside [[*:*]]) and anything else that may be dependent on a request characteristic.
The easiest way for a script to detect if its been proctored into existence is to look for the absence of this or these. No REQUEST_METHOD is a fair indicator as it should exist with all "real" requests. Of course a proctored script is really just there to instantiate itself, not to do anything directly productive, and so a script/RTE can just recognise this and conclude with perhaps a 204 HTTP status (no content) and then remain quiescent (awaiting its first actual request). Any and all output from a proctored script goes to the bit-bucket.
Once proctored into existance the script process is then subject to the normal scripting process management and (for example) if idle for a period exceeding a lifetime value will be procactively removed. Of course, during that process rundown the proctor facility will effectively replace it with a new instance, maintaining the overall requirement.
The Server Admin, DCL Report includes a Proctor List with the currently configured proctor items and associated statistics.
Proctored script activation can be WATCHed just like any other script activation using the [x]CGI and [x]DCL items. To explicitly trigger such an event merely $STOP/ID=pid a proctored scripting process.
Proctor global configuration is introduced with the WASD_CONFIG_GLOBAL [DclScriptProctor] item with each following line representing one script/RTE to be proctored. Each line contains three mandatory and one optional, space-separated components.
The zombie form is
The minimum plus any idle requirement cannot exceed the [DclSoftLimit] configuration value (in order to minimise potential process thrashing).
The proctor facility works by matching the identification string to the script paths as present in the DCL task list (and as presented in the Server Admin, DCL Report). So it needs to contain something unique to that script or environment and often contains a wildcard specification.
The activation path used to activate the script/RTE is the same as if it was activated via a scripting request.
For an RTE the activation script specification does not actually need to exist. It must contain a minimum path to activate the desired environment but the script itself is not checked to exist by WASD and does not need to exist. If it does then it should do whatever is required to instantiate its required environment and return a 204 (no content) response. If it does not exist then the RTE itself should detect it's a proctored activation and return a default 204 response itself, instantiating only the RTE environment.
Proctored scripts can be detected during mapping using
Specific information can also be passed to the proctored script during mapping using such conditional processing in concert with the SET
The combination of these allows some control of proctored scripting.
A proctor item with a minimum (and optionally idle) value of zero can be specified as a place-marker; the facility ignores zero valued items.
This example illustrates a number of non-trivial proctoring scenarios. Only configuration items directly involved in the proctoring are shown; others would be involved in the general web-server infrastructure.
The [DclScriptProctor] contains five items. The first two specify that two scripts each be maintained, the third specifies four, the final two maintain zombie processes. The mapping rules (below) contain a conditional detecting the absence of a REQUEST_METHOD and processing the proctored scripts inside that decision structure. Proctor-specific mapping rules tend to be used only to supplement otherwise fundamental (but in this case proctored) scripting.
The WASD cache was originally provided to reduce file-system access (a somewhat expensive activity under VMS). With the expansion in the use of dynamically generated page content (e.g. PHP, Perl, Python) there is an obvious need to reduce the system impact of some of these activities. While many such responses have content specific to the individual request a large number are also generated as general site pages, perhaps with simple time or date components, or other periodic information. Non-file caching is intended for this type of dynamic content.
Revalidation of non-file content is difficult to implement for a number of reasons, both by the server and by the scripts, and so is not provided. Instead the cache entry is flushed on expiry of the [CacheValidateSeconds], or as otherwise specified by path mapping, and the request is serviced by the content source (script, PHP, Perl, etc.) with the generated response being freshly cached. Browser requests specifying no-caching are honoured (within server configuration parameters) and will flush the entry, resulting in the content being reloaded.
Determining which script content is to be cached and which not, and how long before flushing, is done using mapping rules (described in detail in Request Processing Configuration of WASD Configuration). The source of script cache content is specified using one or a combination of the following SET rules against general or specific paths in WASD_CONFIG_MAP. All mapping rules (script and non-script) are described here to put the script oriented ones into context. Those specific to script output caching are noted.
A good understanding of site requirements and dynamic content sources, along with considerable care in specifying cache path SETings, is required to cache dynamic content effectively. It is especially important to get the content revalidation period appropriate to the content of the pages. This is specified using the following path SETings.
To cache the content of PHP-generated home pages that contain a time-of-day clock, resolving down to the minute, would require a mapping rule similar to the following.
To prevent requests from flushing a particular scripts output (say the main page of a site) using no-cache fields until the server determines that it needs reloading use the cache guard period.
By default the server accesses scripts using the search list logical name CGI-BIN, although this can be significantly changed using mapping rules. CGI-BIN is defined to first search WASD_ROOT:[CGI-BIN] and then WASD_ROOT:[AXP-BIN], WASD_ROOT:[IA64-BIN], or WASD_ROOT:[X86_64-BIN] depending on the platform. [CGI-BIN] is intended for architecture-neutral script files (.CLASS., COM, .PL, .PY, etc.) and the architecture specific directories for executables (.EXE, .DLL, etc.)
These directories are delivered empty and it is up to the site to populate them with the desired scripts. A script is made available by copying its file(s) into the appropriate directory. By default ACLs will be propagated to allow access by the default scripting account. Scripts can be made unavailable by deleting them from these directories.
WASD script executables are built into the WASD_ROOT:[AXP], WASD_ROOT:[IA64] or WASD_ROOT:[X86_64] directories depending on the architecture. Other script files, such as DCL procedures, Perl examples, Java class examples, etc. are located in other directories in the WASD_ROOT:[SRC] tree. The procedure WASD_ROOT:[INSTALL]SCRIPTS.COM assists in the installation or deinstallation of groups of WASD scripts.
Scripts are enabled using the exec/uxec or script rules in the mapping file (also see Request Processing Configuration of WASD Configuration). The script portion of the result must be a URL equivalent of the physical VMS procedure or executable specification.
All files in a directory may be mapped as scripts using the exec rule. For instance, in the WASD_CONFIG_MAP configuration file can be found a rule
Multiple such paths may be designated as executable, with their contents expected to be scripts, either directly executable by VMS (e.g. .EXEs and .COMs) or processable by a designated interpreter, etc., (e.g. .PLs, .CLASSes) (4. Run-Time Environments).
In addition individual files may be specified as scripts. This is done using the script rule. In the following example the request path "/help" activates the "Conan The Librarian" script.
Of course, multiple such rules may be used to map such abbreviated or self-explanatory script paths to the actual script providing the application.
It is not necessary to move/copy scripts into the server directory structure to make them accessible. In fact there are probably good reasons for not doing so! For instance, it keeps a package together so that at the next upgrade there is no possibility of the "server-instance" of that application being overlooked.
To make scripts provided by third party packages available for server activation three requirements must be met.
Most packages having such an interface for Web server access would provide details on mapping into the package directory. For illustration the following mapping rules provide access to a package's scripts (assuming it provides more than one) and also into a documentation area.
The hypothetical "Application X" directory locations are
The required mapping rules would be
Access to X's scripts would be using a path such as
Sometimes it may be necessary to provide a particular non-WASD, local, or third-party script with a particular environment in which to execute. This can be provided by wrapping the script executable or interpreted script in a DCL procedure (of course, if the local or third-party script is already activated by a DCL procedure, then that may need to be directly modified). Simply create a DCL procedure, in the same directory as the script executable, containing the required environmental commands.
For example, the following DCL procedure defines a scratch directory and provides the location of the configuration file. It is assumed the script executable is APPLICATIONX_ROOT:[CGI-BIN]APPX.EXE and the script wrapper APPLICATIONX_ROOT:[CGI-BIN]APPX.COM.
A script is merely an executed or interpreted file. Although by default VMS executables and DCL procedures can be used as scripts, other environments may also be configured. For example, scripts written for the Perl language may be transparently given to the Perl interpreter in a script process. This type of script activation is based on a unique file type (extension following the file name), for the Perl example this is most commonly ".PL", or sometimes ".CGI". Both of these may be configured to automatically invoke the site's Perl interpreter, or any other for that matter.
This configuration is performed using the WASD_CONFIG_GLOBAL [DclScriptRunTime] directive, where a file type is associated with a run-time interpreter. This parameter takes two components, the file extension and the run-time verb. The verb may be specified as a simple, globally-accessible verb (e.g. one embedded in the CLI tables), or in the format to construct a foreign-verb, providing reasonable versatility. Run-time parameters may also be appended to the verb if desired. The server ensures the verb is foreign-assigned if necessary, then used on a command line with the script file name as the final parameter to it.
The following is an example showing a Perl interpreter being specified. The first line assumes the "Perl" verb is globally accessible on the system (e.g. perhaps provided by the DCL$PATH logical) while the second (for the sake of illustration) shows the same Perl interpreter being configured for a different file type using the foreign verb syntax.
A file contain a Perl script then may be activated merely by specifying a path such as the following
To add any required parameters just append them to the verb specified.
If a more complex run-time interpreter is required it may be necessary to wrap the script's execution in a DCL procedure.
The WASD server does not require a file type (extension) to be explicitly provided when activating a script. This can help hide the implementation detail of any script. If the script path does not contain a file type the server searches the script location for a file with one of the known file types, first ".COM" for a DCL procedure, then ".EXE" for an executable, then any file types specified using script run-time configuration directive, in the order specified.
For instance, the script activated in the Perl example above could have been specified as below and (provided there was no "EXAMPLE.COM" or "EXAMPLE.EXE" in the search) the same script would have been executed.
CGI environment variables SCRIPT_FILENAME and PATH_TRANSLATED can be provided to any script (CGI, CGIplus, RTE) in Unix file-system syntax should that script require or prefer it using this format.
The path mapping rule "SET script=syntax=unix" changes the default syntax from VMS to Unix file-system. For example; by default using the URL
If the script path had been specifically mapped using
Note that the CGI or CGIplus script file is still activated using VMS file-system syntax, it is just the CGI representation that is changed. This can be particularly useful for environments ported from Unix expecting to manipulate paths using Unix syntax. This would most commonly occur with RTE engines such as PHP, Perl, etc.
As of WASD v12.1, a default CSP is applied to all CGI responses. CSP is an added layer of security that helps to detect and mitigate certain types of attacks, including Cross Site Scripting (XSS) and data injection attacks.
Content Security Policy (CSP) in WASD Configuration).
https://en.wikipedia.org/wiki/Content_Security_Policy
https://developer.mozilla.org/en-US/docs/Web/HTTP/CSP
https://cheatsheetseries.owasp.org/cheatsheets/Content_Security_Policy_Cheat_Sheet.html
In the absence of a document-embedded CSP, a set path CSP, or a script generated CSP, WASD itself generates a response header field
The server generates a single-use, dynamically generated nonce for each request, provided in CGI variable
Any WASD generated CSP may be disabled on a per-script or whole-of-server using
Two logicals provide some control of and input to the DCL process scripting environment (which includes standard CGI, CGIplus and ISAPI, DECnet-based CGI, but excludes DECnet-based OSU).
Note that each layer of execution added to the scripting environment increases both system overhead and response latency.
If the logical name value is a dotted-decimal specified IP address the verify is only applied to scripts associated with requests originating from that address. This is useful when trying to trouble-shoot scripts on a live server.
Note that most WASD scripts also contain logical names that can be set for debugging purposes. These are generally in the format script_name$DBUG and if exist activate debugging statements throughout the script.
Scripts often require temporary file space during execution. Of course this can be located anywhere the scripting account (most often HTTP$SERVER) has appropriate access. The WASD package does provide a default area for such purposes with permissions set during startup to allow the server account full access. The default area is located in
The server provides for the routine clean-up of old files in WASD_SCRATCH: left behind by aborted or misbehaving scripts (although as a matter of design all scripts should attempt to clean up after themselves). The WASD_CONFIG_GLOBAL directives
Of course there is always the potential for interaction between scripts using a common area for such purposes. At the most elemetary, care must be taken to ensure unique file name are generated. At worst there is the potential for malicious interaction and information leakage. Use such common areas with discretion.
The "UNIQUE_ID" CGI variable provides a unique 19 character alpha-numeric string (‘UNIQUE_ID Symbol’ in 2.1 CGI Environment Variables) suitable for many uses including the type extension of temporary files. The following DCL illustrates the essentials of generating a script-unqiue file name. For mutliple file names add further text to the type, as shown below.
A similar approach can be used for script coded using the C language, with the useful capacity to mark the file for delete-on-close (of course this is only really useful if it is, say, only to be written, rewound and then re-read without closing first – but I'm sure you get the idea).
DCL is the native scripting environment for VMS and provides a rich set of constructs and capabilities for ad hoc and low usage scripting, and as a glue when several processing steps need to be undertaken for a particular script. In common with many interpreted environments care must be taken with effective exception handling and data validation. To assist with the processing of request content and response generation from within DCL procedures the CGIUTL utility is available in WASD_ROOT:[SRC.MISC]
Functionality includes
Most usefully it can read the request body, decoding form-URL-encoded contents into DCL symbols and/or a scratch file, allowing a DCL procedure to easily and effectively process this form of request.
A source code collection of C language functions useful for processing the more vexing aspects of CGI and general script programming is available in CGILIB. This and an example implementation is available in WASD_ROOT:[SRC.MISC]
Functionality includes
The WASD scripts use this library extensively and may serve as example applications.
Of course a script can generate any output it requires including non-success (non-200) pages (e.g. 400, 401, 302, etc.) For error pages a certain consistency results from making these substantially the same layout and content as those generated by the server itself. To this end, script response header output can contain one or more of several extension fields to indicate to the server that instead of sending the script response to the client it should internally generate an error response using the script-supplied information. These fields are listed in ‘Script-Control:’ in 2.2.1 CGI Compliant Output section of 2.2.1 CGI Compliant Output and are available in any scripting environment.
If a "Script-Control: X-error-text="text of error message"" field occurs in the script response header the server stops processing further output and generates an error message. Other error fields can be used to provide additional or message-modifying information. A significant example is the "Script-Control: X-error-vms-status=integer" field which supplies a VMS status value for a more detailed, status-related error message explanation.
Essentially the script just generates a standard CGI "Status: nnn" response and includes at least the "X-error-text=" field before the header-terminating empty record (blank line). Some variations are shown in the following DCL examples.
Interestingly, because CGI environments should ignore response fields unknown to them, for scripts deployed across multiple server platforms it should be possible to have these WASD-specific elements in every header for WASD uses followed by other explicitly error page content for use in those other environments.
An example implemented using DCL is available
and if currently enabled for scripting↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
The information in this chapter merely outlines the WASD implementation details, which are in general very much vanilla CGI and NCSA CGI (Common Gateway Interface) compliant, originally based the INTERNET-DRAFT authored by D.Robinson (drtr@ast.cam.ac.uk), 8 January 1996, confirmed against the final RFC 3875, authored by David Robinson (drtr@apache.org) and Ken A.L.Coar (coar@apache.org), October 2004.
With the standard CGI environment variables are provided to the script via DCL global symbols. Each CGI variable symbol name is prefixed with "WWW_" (by default, although this can be changed using the "/CGI_PREFIX" qualifier and the SET CGIPREFIX mapping rule, see "Features and Facilities", this is not recommended if the WASD VMS scripts are to be used, as they expect CGI variable symbols to be prefixed in this manner).
There are a number of non-"standard" CGI variables to assist in tailoring scripts for the WASD environment. Do not make your scripts dependent on any of these if portability is a goal.
CGI variable capacity now varies significantly with VMS version.
The total size of all CGI variable names and values is determined by the value of [BufferSizeDclCommand] configuration directive, which determines the total buffer space of a mailbox providing the script's SYS$COMMAND. The default value of 4096 bytes will be ample for the typical CGI script request, however if it contains very large individual variables or a large number of form fields, etc., it may be possible to exhaust this quantity.
CGI variables may contain values in excess of 8000 characters (the full 8192 symbol capacity cannot be realized due to the way the symbols are created via the CLI). This is a significant increase on earlier capacities. Mailbox buffer [BufferSizeDclCommand] may need to be increased if this capacity is to be fully utilized.
Values may contain approximately 1000 characters minus the size of the variable name. This should still be sufficient for most circumstances (if not consider using CGIplus or ISAPI, extensions to CGI programming which remove this limitation). Why such an odd number and why a little rubbery? A DCL command line with these versions is limited to 255 characters so the symbols for larger variables are built up over successive DCL commands with the limit determined by CLI behaviour.
On VMS V7.3-2 and later symbol capacity should never be an issue (well, perhaps only with the most extraordinarily poorly designed script). With VMS V7.3-1 and earlier, with a symbol value that is too large, the server by default aborts the request, generating and returning a 500 HTTP status. Experience has shown that this occurs very rarely. If it does occur it is possible to instruct the server to instead truncate the CGI variable value and continue processing. Any CGI variable that is truncated in such a manner has its name placed in CGI variable SERVER_TRUNCATE, so that a script can check for, and take appropriate action on, any such truncation. To have the server truncate such variables instead of aborting processing SET the path using the script=symbol=truncate mapping rule. For example
Remember, by default all variables are prefixed by "WWW_" (though this may be modified using the set CGIprefix= mapping rule), and not all variables will be present for all requests. These CGI environment variables reflect a combination of HTTP/1.1 and HTTP/1.0 request parameters.
Name | Description | Origin |
---|---|---|
AUTH_ACCESS | "READ" or "READ+WRITE" | WASD |
AUTH_AGENT | used by an authorization agent (specialized use) | WASD |
AUTH_GROUP | authentication group | WASD |
AUTH_PASSWORD | plain-text password, only if EXTERNAL realm | WASD |
AUTH_REALM | authentication realm | WASD |
AUTH_REALM_DESCRIPTION | browser displayed string | WASD |
AUTH_TYPE | authentication type (BASIC or DIGEST) | CGI |
AUTH_USER | details of authenticated user | WASD |
CONTENT_LENGTH | "Content-Length:" from request header | CGI |
CONTENT_TYPE | "Content-Type:" from request header | CGI |
CSP_NONCE | a single-use CSP nonce for each request | WASD |
DOCUMENT_ROOT | generally empty, configurable path setting | Apache |
FORM_field-name | query string "&" separated form elements | WASD |
GATEWAY_BG | device name of raw client socket (specialized use) | WASD |
GATEWAY_EOF | End of request sentinal (specialized use) | WASD |
GATEWAY_EOT | End of callout sentinal (specialized use) | WASD |
GATEWAY_ESC | Callout escape sentinal (specialized use) | WASD |
GATEWAY_INTERFACE | "CGI/1.1" | CGI |
GATEWAY_MRS | maximum record size of script SYS$OUTPUT | WASD |
GATEWAY_SYMBOLS | comma-separated list of all CGI variable names | WASD |
HTTP2_PING | HTTP/2 server-client RTT ping in (real number) milliseconds | WASD |
HTTP_ACCEPT | any list of browser-accepted content types | CGI |
HTTP_ACCEPT_CHARSET | any list of browser-accepted character sets | CGI |
HTTP_ACCEPT_LANGUAGE | any list of browser-accepted languages | CGI |
HTTP_AUTHORIZATION | any from request header (specialized use) | CGI |
HTTP_CACHE_CONTROL | cache control directive | CGI |
HTTP_COOKIE | any cookie sent by the client | CGI |
HTTP_CONNECTION | connection persistence request field | CGI |
HTTP_FORWARDED | any proxy/gateway hosts that forwarded the request | CGI |
HTTP_HOST | host and port request was sent to | CGI |
HTTP_IF_MATCH | if-match request field | CGI |
HTTP_IF_NONE_MATCH | if-none-match request field | CGI |
HTTP_IF_MODIFIED_SINCE | any last modified GMT time string | CGI |
HTTP_IF_UNMODIFIED_SINCE | request field | CGI |
HTTP_IF_RANGE | if-range request field | CGI |
HTTP_KEEP_ALIVE | connection persistence request field | CGI |
HTTP_PRAGMA | any pragma directive of request header | CGI |
HTTP_REFERER | any source document URL for this request | CGI |
HTTP_USER_AGENT | client/browser identification string | CGI |
HTTP_X_FORWARDED_FOR | proxied client host name or address | Squid |
HTTP_field-name | any other request header field | WASD |
KEY_n | query string "+" separated elements | WASD |
KEY_COUNT | number of "+" separated elements | WASD |
PATH_INFO | virtual path of data requested in URL | CGI |
PATH_TRANSLATED | VMS file path of data requested in URL | CGI |
QUERY_STRING | un-URL-decoded string following "?" in URL | CGI |
REMOTE_ADDR | IP host address of HTTP client | CGI |
REMOTE_HOST | IP host name of HTTP client | CGI |
REMOTE_USER | authenticated remote user name (or empty) | CGI |
REQUEST_CHARSET | any server-determined request character set | WASD |
REQUEST_METHOD | "GET", "PUT", etc. | CGI |
REQUEST_PROTOCOL | "HTTP/2", "HTTP/1.1" or "HTTP/1.0" | WASD |
REQUEST_SCHEME | "http:" or "https:" | WASD |
REQUEST_TIME_GMT | GMT time request received | WASD |
REQUEST_TIME_LOCAL | Local time request received | WASD |
REQUEST_URI | full, unescaped request string | Apache |
SCRIPT_DEFAULT | mapped default directory for script | WASD |
SCRIPT_FILENAME | script file name (e.g. CGI-BIN:[000000]QUERY.COM) | Apache |
SCRIPT_NAME | script being executed (e.g. "/query") | CGI |
SERVER_ADDR | IP host name of server system | WASD |
SERVER_ADMIN | email address for server administration | Apache |
SERVER_CHARSET | server default character set | WASD |
SERVER_GMT | offset from GMT (e.g. "+09:30") | WASD |
SERVER_NAME | IP host name of server | CGI |
SERVER_PROTOCOL | HTTP protocol version (generally "HTTP/1.1") | CGI |
SERVER_PORT | IP port request was received on | CGI |
SERVER_SIGNATURE | server ID, host name and port | Apache |
SERVER_SOFTWARE | software ID of HTTP server | CGI |
SERVER_TRUNCATE | CGI variable value the server was forced to truncate | WASD |
UPSTREAM_ADDR | address of transparent proxy when client mapped | WASD |
UNIQUE_ID | unique 19 character string | Apache |
WATCH_SCRIPT | only present when script being WATCHed | WASD |
If the request path is set to provide them, there are also be variables providing information about a Secure Sockets Layer transported request's SSL environment.
In line with other CGI implementations, additional, non-compliant variables are provided to ease CGI interfacing. These provide the various components of any query string. A keyword query string and a form query string are parsed into
Variables named WWW_KEY_number will be generated if the query string contains one or more plus ("+") and no equate symbols ("=").
Variables named WWW_FORM_form-element-name will be generated if the query string contains one or more equate symbols. Generally such a query string is used to encode form-URL-encoded (MIME type x-www-form-urlencoded) requests. By default the server will report an incorrect encoding with a 400 error response. However some scripts use malformed encodings and so this behaviour may be suppressed using the set script=query=relaxed mapping rule.
To suppress this decoding completely (and save a few CPU cycles) use the following rule.
The GATEWAY_SYMBOLS variable contains a comma-separated list of CGI variable names for each vanilla CGI request. With scripting based on CGIplus (see 3. CGIplus) variable names and values may be parsed for the CGI data. This is not possible with vanilla CGI and so GATEWAY_SYMBOLS is provided, allowing each name to be simply parsed.
Example DCL code:
Example C code using CgiLibVar() (see 1.13 Scripting Function Library):
The UNIQUE_ID variable for each request generates a globally and temporally unique 19 character string that can be used where such a identifier might be needed. This string contains only alphanumeric characters and is "guaranteed" to be unique in (Internet) space and time. For example use see 1.11 Scripting Scratch Space.
WASD v7.0 had its CGI environment tailored slightly to ease portability between VMS Apache (CSWS) and WASD. This included the provision of an APACHE$INPUT: stream and several Apache-specific CGI variables (see the table below). The CGILIB C function library (1.13 Scripting Function Library) has also been made CSWS V1.0-1 and later (Apache 1.3.12 and higher) compliant.
The basic CGI symbol names are demonstrated here with a call to a script that simply executes the following DCL code:
/cgi-bin/cgi_symbols |
/cgi-bin/cgi_symbols/wasd_root/wasdoc/ |
/cgi-bin/cgi_symbols/wasd_root/wasdoc/*.*?string1+string2 |
/cgi-bin/cgi_symbols/wasd_root/wasdoc/?FirstField=for&SecondField=this |
The source code for this simple script is:
WASD_ROOT:[SRC.OTHER]CGI_SYMBOLS.COM
This information applies to all non-DECnet based scripting, CGI, CGIplus, RTE, ISAPI. WASD uses mailboxes for script inter-process communication (IPC). These are efficient, versatile and allow direct output from all VMS environments and utilities. Like many VMS record-oriented devices however there are some things to consider when using them (also see ‘IPC Tickler’ in 2.2.1 CGI Compliant Output).
The mailboxes are created record, not stream oriented. This means records output by standard VMS means (e.g. DCL, utilities, programming languages) are discretely identified and may be processed appropriately by the server as text or binary depending on the content-type.
Being record oriented there is a maximum record size (MRS) that can be output. Records larger than this result in SYSTEM-F-MBTOOSML errors. The WASD default is 4096 bytes. This may be changed using the [BufferSizeDclOutput] configuration directive. This allocation consumes process BYTLM with each mailbox created so the account must be dimensioned sufficiently to supply demands for this quota. The maximum possible size for this is a VMS-limit of 60,000 bytes.
When created the mailbox has its buffer space set. With WASD IPC mailboxes this is the same as the MRS. The total data buffered may not exceed this without the script entering a wait state (for the mailbox contents to be cleared by the server). As mailboxes use a little of the buffer space to delimit records stored in it the amount of data is actually less than the total buffer space.
To determine the maximum record size and total capacity of the mailbox buffer between server and script WASD provides a CGI environment variable, GATEWAY_MRS, containing an integer with this value.
Script response may be CGI or NPH compliant (2.2.2 Non-Parsed-Header Output). CGI compliance means the script's response must begin with a line containing one of the following fields.
Other HTTP-compliant response fields may follow, with the response header terminated and the response body begun by a single empty line. The following are examples of CGI-compliant responses.
Strict CGI output compliance can be enabled and disabled using the [CgiStrictOutput] configuration directive. With it disabled the server will accept any output from the script, if not CGI or NPH compliant then it automatically generates plain-text header. When enabled, if not a CGI or NPH header the server returns a "502 Bad Gateway" error. For debugging scripts generating this error introduce a plain-text debug mode and header, or use the WATCH facility's CGI item (see Features and Facilities).
With HTTP/1.1 it is generally better to use CGI than NPH responses. A CGI response allows the server to parse the response header and from that make decisions about connection persistence and content-encoding. These can contribute significantly to reducing response latency and content transfer efficiency. It allows any policy established by server configuration for such characteristics to be employed.
This section describes how WASD deals with some particular output issues (also see ‘IPC Tickler’ in 2.2.1 CGI Compliant Output).
If the script response content-type is "text/…" (text document) WASD assumes that output will be line-oriented and requiring HTTP carriage-control (each record/line terminated by a line-feed), and will ensure each record it receives is correctly terminated before passing it to the client. In this way DCL procedure output (and the VMS environment in general) is supported transparently. Any other content-type is assumed to be binary and no carriage-control is enforced. This default behaviour may be modified as described below.
Carriage-control behaviour for any content-type may be explicitly set using either of two additional response header fields. The term stream is used to describe the server just transfering records, without additional processing, as they were received from the script. This is obviously necessary for binary/raw content such as images, octet-streams, etc. The term record describes the server ensuring each record it receives has correct carriage-control - a trailing newline. If not present one is added. This mode is useful for VMS textual streams (e.g. output from DCL and VMS utilities).
Using the Apache Group's proposed CGI/1.2 "Script-Control:" field. The WASD extension-directives X-record-mode and X-stream-mode sets the script output into each of the respective modes (‘Script-Control:’ in 2.2.1 CGI Compliant Output).
Examples of usage this field:
By default WASD writes each record received from the script to the client as it is received. This can range from a single byte to a complete mailbox buffer full. WASD leaves it up to the script to determine the rate at which output flows back to the client.
While this allows a certain flexibility it can be inefficient. There will be many instances where a script will be providing just a body of data to the client, and wish to do it as quickly and efficiently as possible. Using the proposed CGI/1.2 "Script-Control:" field with the WASD extension directive X-buffer-records a script can direct the server to buffer as many script output records as possible before transfering it to the client. The following should be added to the CGI response header.
While the above offers some significant improvements to efficiency and perceived throughput the best approach is for the script to provide records the same size as the mailbox (2.2 Script Output) for detail on determining this size if required). The can be done explicitly by the script programming or if using the C language simply by changing stdout to a binary stream. With this environment the C-RTL will control output, automatically buffering as much as possible before writing it to the server.
Also see the section describing ‘NPH C Script’ in 2.2.2 Non-Parsed-Header Output.
Non-C Runtime Libraries (C-RTL) do not contend with records delimitted by embedded characters (the newlines and nulls, etc., of the C environment). They use VMS' and RMS' default record-oriented I/O. The C-RTL needs to accomodate the C environment's bag-o'-bytes paradigm for file content against RMS' record structures, and it's embedded terminator, stream-oriented I/O with unterminated, record-oriented I/O. Often this results in a number of issues particularly with code ported from *x environments.
The C-RTL behaviour can be modified in all sorts of ways, including some file and other I/O characteristics. The features available to such modification are incrementally increasing with each release of the C-RTL and/or C compiler. It is well advised to consult the latest release (or as appropriate for the local environment) of the Run-Time Library Reference Manual for OpenVMS Systems for the current set.
Behaviours are modified by setting various flags, either from within the program itself using thef using the decc$feature_set() and allied group of functions, or by defining an equivalent logical name, usually externally to and before executing the image. See C-RTL Reference Manual section Enabling C RTL Features Using Feature Logical Names. This is particularly useful if the source is unavailable or just as a simpler approach to modifying code.
An example of a useful feature and associated logical name is DECC$STDIO_CTX_EOL which when enabled "writing to stdout and stderr for stream access is deferred until a terminator is seen or the buffer is full" in contrast to the default behaviour of "each fwrite generates a separate write, which for mailbox and record files generates a separate record". For an application performing write()s or fwrite()s with a record-oriented <stdio> and generating inappropriate record boundaries the application could be wrapped as follows (a real-world example).
The interactions between VMS' record-oriented I/O, various run-time libraries (in particular the C-RTL), the streaming character-oriented Web, and of course WASD, can be quite complex and result in unintended output or formatting. The CGI script Inter-Process Communication (IPC) tickler WASD_ROOT:[SRC.MISC]IPCTICKLER.C is designed to allow a script programmer to gain an appreciation of how these elements interact, how WASD attempts to accomodate them, what mechanisms a script can use to explicitly convey exact requirements to WASD ... and finally, how these affect output (in particular the carriage-control) delivered to the client. If installed use /cgi-bin/IPCtickler to obtain an HTML form allowing control of several parameters into the script.
The Apache Group has proposed a CGI/1.2 that includes a Script-Control: CGI response header field. WASD implements the one proposed directive, along with a number of WASD extensions (those beginning with the "X-"). Note that by convention extensions unknown by an agent should be ignored, meaning that they can be freely included, only being meaningful to WASD and not significant to other implementations.
The following is a simple example response where the server is instructed not to delete the script process under any circumstances, and that the body does not require any carriage-control changes.
A simple script to provide the system time might be:
A script to provide the system time more elaborately (using HTML):
A script does not have to output a CGI-compliant data stream. If it begins with a HTTP header status line WASD assumes it will supply a raw HTTP data stream, containing all the HTTP requirements. This is the same as or equivalent to the non-parsed-header, or "nph…" scripts of many environments. This is an example of such a script response.
Any such script must observe the HyperText Transfer Protocol, supplying a full response header and body, including correct carriage-control. Once the server detects the HTTP status header line it pays no more attention to any response header fields or body records, just transfering everything directly to the client. This can be very efficient, the server just a conduit between script and client, but does transfer the responsibility for a correct HTTP response onto the script.
The following example shows a DCL script. Note the full HTTP header and each line explicitly terminated with a carriage-return and line-feed pair.
When scripting using the C programming language there can be considerable efficiencies to be gained by providing a binary output stream from the script. This results in the C Run-Time Library (C-RTL) buffering output up to the maximum supported by the IPC mailbox. This may be enabled using a code construct similar to following to reopen stdout in binary mode.
This is used consistently in WASD scripts. Carriage-control must be supplied as part of the C standard output (no differently to any other C program). Output can be be explicitly sent to the client at any stage using the fflush() standard library function. Note that if the fwrite() function is used the current contents of the C-RTL buffer are automatically flushed along the the content of the fwrite().
As described above, 2.2 Script Output, the default script<->server IPC uses a mailbox. While versatile and sufficiently efficient for general use, when megabytes, tens of megabytes, and hundreds of megabytes need to be transferred, using a memory buffer shared between script and server can yield transfer improvements of up to 500%.
The script requests a memory-buffer using a CGI callout (6. CGI Callouts). Buffer size is constrained by the usual VMS 32bit memory considerations, along with available process and system resources. The server creates and maps a non-permanent global section. If this is successful the script is advised of the global section name using the callout response. The script uses this to map the section name and can then populate the buffer. When the buffer is full or otherwise ready, the script issues a callout with the number of bytes to write, and then stalls. The complete memory buffer may be written at once or any subsection of that buffer. The write is accomplished asynchronously and may comprise multiple network $QIOs or TLS/SSL blocks. When complete, a callout response to the script is issued and the script can continue processing. Standard script mailbox I/O (SYS$OUTPUT, <stdout>) and memory-buffer I/O may be interleaved as required.
The callouts are as follows:
Create a temporary global section to act as a memory buffer shared between a script process and the server. The default is Megabytes.
Dispose of the shared memory buffer created by callout BUFFER-BEGIN.
Instruct the server to write <integer> bytes from the shared memory buffer to the client.
See working examples in WASD_ROOT:[SRC.MISC].
Actual data comparing standard mailbox IPC with memory-buffer generated using [SRC.MISC]MEMBUFDEMO.C on a HP rx2660 (1.40GHz/6.0MB) with 4 CPUs and 16383MB running VSI VMS V8.4-2L1 with Multinet UCX$IPC_SHR V55A-B147, OpenSSL 1.0.2k and WASD v11.2.0, with [BufferSizeDclOutput] 16384. In each case 250MB ("?250") is transfered via a either a 16.4kB mailbox (default) or 16.4kB memory buffer ("+b"). Significantly larger memory buffer may well improve throughput further.
It is obvious that memory-buffer provides significantly greater throughput than mailbox (from the http:// test) and that with TLS/SSL network transport the encryption becomes a significant overhead and choke-point. Nevertheless, there is still an approximate 15% dividend, plus the more efficient interface the script->memory-buffer->server provides. The VMS TLS/SSL implementation may improve with time, especially if TLS/SSL hardware engines become available with the port to x86_64.
For POST and PUT HTTP methods (e.g. a POSTed HTML form) the body of the request may be read from the HTTP$INPUT stream. For executable image scripts requiring the body to be present on SYS$INPUT (the C language stdin stream) a user-mode logical may be defined immediately before invoking the image, as in the example.
The HTTP$INPUT stream may be explicitly opened and read. Note that this is a raw stream, and HTTP lines (carriage-return/line-feed terminated sequences of characters) may have been blocked together for network transport. These would need to be explicity parsed by the program.
When scripting using the C programming language there is a tendency for the C-RTL to check for and/or add newline (0x10, <LF>) carriage-control on receipt of record (single write). While this can be useful in converting from VMS to C conventions it can also be counter-productive if the stream being received is already using C carriage-control. To prevent the C-RTL reinterpreting data passed to it it often, perhaps invariably, necessary to reopen the input stream as binary using a construct similar to following.
This, and its <stdin> equivalent (below), are used consistently in WASD scripts.
The input stream should be read before generating any output. If an error occurs during the body processing it should be reported via a CGI response header indicating an error (i.e. non-200). With HTTP/1.1 request processing there is also a requirement (that CGILIB fulfills) to return a "100 Continue" interim response after receiving the client request header and before the client sends the request body. Output of anything before this "100 Continue" is delivered will cause it to be interleaved with the script response body.
A source code collection of C language functions useful for processing the more vexing aspects of CGI/CGIplus programming (1.13 Scripting Function Library).
This assists with the generation of HTTP responses, including the transfer of binary content from files (copying a file back to the client as part of the request), and the processing of the contents of POSTed requests from DCL (2.5 CGIUTL Utility).
3.1CGIplus Programming |
3.2Code Examples |
3.3Other Considerations |
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
Common Gateway Interface | … | plus lower latency, |
---|---|---|
plus greater throughput, | ||
plus far less system impact! |
I know, I know! The term CGIplus is a bit too cute but I had to call it something!
CGIplus attempts to eliminate the overhead associated with creating the script process and then executing the image of a CGI script. It does this by allowing the script process and any associated image/application to continue executing between uses, eliminating startup overheads. This reduces both the load on the system and the request latency. In this sense these advantages parallel those offered by commercial HTTP server-integration APIs, such as Netscape NSAPI and Microsoft ISAPI, without the disadvantages of such proprietory interfaces, the API complexity, language dependency and server process integration.
Existing CGI scripts can rapidly and elegantly be modified to additionally support CGIplus. The capability of scripts to easily differentiate between and operate in both standard CGI and CGIplus environments with a minimum of code revision offers great versatility. Many WASD scripts operate in both environments.
A simple performance evaluation indicates the advantage of CGIplus. See "Techncial Overview, Performance" for some test results comparing the CGI and CGIplus environments.
Without a doubt, the subjective difference in activating the same script within the standard CGI and CGIplus environments is quite startling!
The script interface is still CGI, with all the usual environment variables and input/output streams available, which means a new API does not need to be learned and existing CGI scripts are simple to modify.
See examples in WASD_ROOT:[SRC.CGIPLUS]
Instead of having the CGI variables available from the environment (generally accessed via the C Language getenv() standard library call, or via DCL symbols) a CGIplus script must read the CGI variables from an I/O stream named CGIPLUSIN. The variables can be supplied in one of two modes.
This default and simple record-oriented mode allows any environment that can read records from an input source to process CGIplus variables. This of course includes DCL (examples referenced below).
This mode offers significantly lower system overheads and improves latency and performance at the cost of the additional complexity of recovering the variables from a binary structure. Code examples and CGILIB functions make this relatively trivial at the application level.
After processing, the CGIplus script can loop, waiting to read the details of the next request from CGIPLUSIN.
Request output (to the client) is written to SYS$OUTPUT (<stdout>) as per normal CGI behaviour. End of output MUST be indicated by writing a special EOF record to the output stream. A unique EOF sequence is generated for each use of DCL via a zombie or CGIplus script process. A non-repeating series of bits most unlikely to occur in normal output is employed … but there is still a very, very, very small chance of premature termination of output (one in 2^224 I think!) See WASD_ROOT:[SRC.HTTPD]CGI.C for how the value is generated.
The CGIplus EOF string is obtained by the script from the logical name CGIPLUSEOF, defined in the script process' process table, using the scripting language's equivalent of F$TRNLNM(), SYS$TRNLNM(), or a getenv() call (in the C standard library). This string will always contain less than 64 characters and comprise only printable characters. It must be written at the conclusion of a request's output to the output stream as a single record (line) but may also contain a <CR><LF> or just <LF> trailing carriage-control (to allow for programming language requirements). It only has to be evaluated once, as the processing begins, remaining the same for all requests over the life-time of that instance of the script.
HTTP input (raw request body stream) is still available to a CGIplus script.
A source code collection of C language functions useful for processing the more vexing aspects of CGI/CGIplus/RTE programming (1.13 Scripting Function Library).
Of course a CGIplus script should only have a single exit point and should explicitly close files, free allocated memory, etc., after processing a request (i.e. not rely on image run-down to clean-up after itself). It is particularly important when modifying existing scripts to work in the CGIplus environment to ensure this requirement is met (who of us hasn't thought "well, this file will close when the image exits anyway"?)
It is a simple task to design a script to modify its behaviour according to the environment it is executing in. Detecting the presence or absence of the CGIPLUSEOF logical is sufficient indication. The following C code fragment shows simultaneously determining whether it is a standard or CGIplus environment (and setting an appropriate boolean), and getting the CGIplus EOF sequence (if it exists).
The following C code fragment shows a basic CGIplus record-mode request loop, reading lines from CGIPLUSIN, and some basic processing to select required CGI variables for request processing. Generally this level of coding is not required as it is recommended to employ the functionality of something like the CGILIB functiona library.
This mode requires significantly more code than record-mode. A self-contained C language function, allowing CGI variable processing in standard CGI, CGIplus record-mode and CGIplus struct-mode, is available for inclusion in user applications. It automatically detects the environment and changes behaviour to suit. This or CGILIB is strongly recommended.
See source in WASD_ROOT:[SRC.CGIPLUS]CGIPLUS_CGIVAR.C
CGI scripts can write output in record (line-by-line) or binary mode (more efficient because of buffering by the C RTL). When in binary mode the output stream must be flushed immediately before and after writing the CGIplus EOF sequence (note that in binary a full HTTP stream must also be used). This code fragment shows placing a script output stream into binary mode and the flushing steps.
See examples in WASD_ROOT:[SRC.CGIPLUS]
Scripting processes may be purged or deleted using (see HTTPd Command Line in WASD Features).
Multiple CGIplus scripts may be executing in multiple processes at any one time. This includes multiple instances of any particular script. It is the server's task to track these, distributing appropriate requests to idle processes, monitoring those currently processing requests, creating new instances if and when necessary, and deleting the least-used, idle CGIplus processes when configurable thresholds are reached. Of course it is the script's job to maintain coherency if multiple instances may result in resource conflicts or race conditions, etc., between the scripts.
The CGIplus script process can be given a finite life-time set by configuration parameter (see "Features and Facilities, Server Configuration"). If this life-time is not set then the CGIplus will persist indefinitely (i.e. until purged due to soft-limits being reached, or explicitly purged/deleted). When a life-time has been set the CGIplus process is automatically deleted after being idle for the specified period (i.e. not having processed a request). This can be useful in preventing sporadically used scripts from cluttering up the system indefinitely.
In addition, an idle CGIplus script can be run-down by the server at any time the script process soft-limit is reached, so resources should be largely quiescent when not actually processing (1.2.6 Script Process Run-Down). Of course, in extreme situations, a CGIplus process may also be manually terminated from the command line (e.g. STOP/ID=).
Some CGIplus scripting information and management is available via the server administration menu, see "Features and Facilities, Server Reports".
CGIplus scripts are differentiated from standard CGI scripts in the mapping rule configuration file using the "script+" and "exec+" directives (see Request Processing Configuration in WASD Configuration).
Scripts capable of operating in both standard CGI and CGIplus environments may simply be accessed in either via rules such as
4.1RTE Programming |
4.2Server Configuration |
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
A Run-Time Environment (RTE) is a persistant scripting environment with similar objectives to CGIplus … reducing script response time, increasing server throughput and reducing system impact. In fact the RTE environment is implemented using CGIplus! There is very little difference in the behaviour of CGIplus scripts and RTEs. Both are activated by the server, process multiple requests (reading the request CGI environment from a data stream supplied by the server), persist between requests in a quiescent state, and may be removed by the server if idle for a specified period or when it wishes to use the process for some other purpose. Like CGIplus an RTE must be purpose-written for the environment! What is the difference then?
With CGIplus the script itself persists between uses, retaining all of its state. With an RTE the script does not persist or retain state, only the RTE itself.
A RTE is intended as an environment in which a script source is interpreted or otherwise processed, that is for scripting engines, although it is not limited to that. The essential difference between an RTE and a CGIplus script is this script source. In CGIplus the SCRIPT_NAME and SCRIPT_FILENAME CGI variables reflect the script itself, and remain constant for each activation of the script, with PATH_INFO and PATH_TRANSLATED providing the additional "location" information for the script processing. With an RTE the SCRIPT_NAME and SCRIPT_FILENAME can vary with each activation. This allows the RTE to process multiple, successive different (or the same) scripts, each with its own PATH_INFO and PATH_TRANSLATED. Hence, it is not unreasonable to consider the two environments to be the same, with a slight difference in the mapping of resources passed to them.
This might be best illustrated with examples.
Consider the mapping rule
If the script was an executable it would be activated as
By contrast with a request to activate an RTE the following mapping rule
As can be seen the script information is constant for each request to a CGIplus script, while with RTE the script information could vary with each request (although of course it would be the same if the same script is requested). In the case of CGIplus the process what? request information is provided only by path information, however with RTE both script and path information are used.
The RTE interface is still CGI, with all the usual environment variables and input/output streams available, just in a CGIplus environment! Hence when coding a Run-Time Environment the same considerations involved in CGIplus programming apply (3. CGIplus).
In particular it is important a RTE should explicitly close files, free allocated memory, etc., after processing a request (of course it cannot rely on image run-down to clean-up after itself). It is particularly important that all traces of each script's processing are removed after it concludes. This does not mean for example that databases need to be completely closed, etc., which might defeat the purpose of using a persistant environment, just that great care must be exercised by the programmer to prevent one script interfering with another!
An example RTE, WASD_ROOT:[SRC.CGIPLUS]RTE_EXAMPLE.C provides the basics of the environment.
A source code collection of C language functions useful for processing the more vexing aspects of CGI/CGIplus/RTE programming (). The example RTE implementation uses this library.
The following configuration information uses the supplied Perl RTE as the example. Note that RTE scripting engines must always be mapped using the EXEC+ rules. The SCRIPT+ rule does not apply.
The following rule in WASD_CONFIG_MAP maps the /pl-bin/ location to where the site wishes to locate its CGI Perl scripts (not necessarily the same as in the example).
With this rule Perl scripts may be accessed using
This WASD_CONFIG_GLOBAL rule ensures Perl scripts could be activated via the Perl RTE even if the WASD_CONFIG_MAP rule did not exist (1.7 Script Run-Time).
It does by default however, check that the file used as the script source exists as with other scripting environments. If it does not this is reported as a "script not found". For RTEs that wish to report on this themselves, or that possibly construct their own script specification via some internal processing, this server behaviour may be suppressed for the script activation path using the WASD_CONFIG_MAP path SETting "script=nofind" as in the following example.
CGI environment variables SCRIPT_FILENAME and PATH_TRANSLATED can be provided to any RTE script in Unix file-system syntax should that script require or prefer it using this format. See 1.8 Unix Syntax.
If the RTE executable requires wrapping in a DCL procedure (perhaps to provide some command-line specific parameter or define a C-RTL logical name) this can be specified in place of an executable. Merely prefix the specification with a "@". The default is to run an executable (this can explicitly be specified using a leading "$") while the leading "@" provides a DCL procedure activation.
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
WebSocket is a capability introduced with HTML5, providing an asynchronous, bidirectional, full-duplex connection over which messages can be sent between agents, commonly a browser client and a server application. Compatible browsers provide a JavaScript interface that allows connections to be set up, maintained, messages exchanged, and connections closed down, using a callable and event-based interface.
WASD provides a WebSocket compatible scripting environment, one that is activated in the same fashion as an equivalent CGI/CGIplus/RTE and has an identical CGI environment (variables, streams, etc.) but which uses a unique HTTP response and communicates with its client using the WebSocket protocol.
Client supplied data is available to the script via the WEBSOCKET_INPUT mailbox and data from the script supplied via the WEBSOCKET_OUTPUT mailbox (indicated via CGI variables). Communication using a WebSocket requires the use of a framing protocol while WEBSOCKET_INPUT and WEBSOCKET_OUTPUT are opaque octet-streams providing communication to and from the WebSocket application. CGI variables WEBSOCKET_INPUT_MRS and WEBSOCKET_OUTPUT_MRS indicate the respective mailbox capacity.
The WASD server largely acts as a conduit for the WebSocket octet-stream. It provides the upgrade from HTTP to WebSocket protocol handshake and then connects the bidirectional data stream to the WebSocket application activated in WASD's scripting environment which then is required to perform all of the protocol requirements, etc. The baseline WASD implementation is via the wsLIB library (see below). The complexity and potential extensibility of the WebSocket protocol means this decoupling of server infrastructure and protocol implementation offers a number of advantages, including more straight-forward updates and bug fixing (just the library), and alternate, concurrent implementations.
Long-lived WebSocket scripts by default have timeouts and other limits set to infinite. If control is required it must be exercised using the appropriate mapping SETings or DCL callouts.
A WASD RawSocket is an analogue to the WebSocket, providing a bidirectional, asynchronous, opaque data stream input and output on a per-service basis. See 5.8 WASD "Raw"Socket.
A single WASD WebSocket server application (script) can support multiple clients by using some form of multi-threading such as AST-based I/O, POSIX Threads, multi-thread interpreter environment, etc. The WASD wsLIB library (5.3 WebSocket Library) supports native AST concurrency.
A WebSocket connection to a script is maintained by the WEBSOCKET_INPUT and WEBSOCKET_OUTPUT channels remaining connected to the script. If the script closes them (or the image or process exits, etc.) the WebSocket connection is closed. WebSocket requests are maintained as long as the script maintains them, for a CGIplus script, until it exits. If a CGIplus script requires to disconnect from a WebSocket client without exiting it must do so explicitly (by using the wsLIB close function (and associated WebSocket protocol close handshake), closing C streams, deassigning channels, etc.)
Of course this is the underlying mechanism allowing a single CGIplus script to maintain connections with multiple WebSocket clients. Provided the script remains connected to the WebSocket IPC mailboxes and processes that I/O asynchronously a single script can concurrently handle multiple clients. The script just processes each request it is given, adding the new client to the existing group (and removing them as the IPC indicates they disconnect).
Obviously the script must remain resident via CGIplus or RTE.
The server will continue to provide requests to the script for as long as it appears idle (i.e. the CGIplus sentinal EOF is returned even though concurrent processing may continue). Obviously a single scripting process cannot accept an unlimited number of concurrent WebSockets. When a script decides it can process no more it should not return the sentinal EOF from the most recent request until it is in a position to process more, when it then provides the EOF and the server again will supply another request.
The original request is access logged at request run-down (when the WebSocket is finally closed either because the client disconnected or the script closed its connection to the WEBSOCKET_.. mailboxes). The access log status is 101 (Switching Protocols) and the bytes rx and tx reflect the total for the duration.
WebSocket server applications are essentially CGIplus scripts and so have similar programming considerations (see 3. CGIplus).
A WebSocket application however is typically long-lived and involves significant interaction between the participants. Either party can initiate independent communication with the other according to the required business logic.
A WASD WebSocket application relies on asynchronous I/O and other events to provide the communication granularity required for application interaction. The following pseudo-code shows the structure of one such hypothetical application. It accepts multiple, concurrent requests in it's main loop, creates the required WebSocket protocol supporting data structure, and then services application requirements in two event loops.
The first reads from the remote client and processes according to the business logic of client-initiated processing, asynchronously and/or synchronously writing data to the client. The second loop pushes data asynchronously to the client based on the application business logic providing those events. The close event occurs when the client or application close the WebSocket, or are otherwise disconnected, and finalises the request.
A second model for request synchronisation allows the initialisation to specify a routine to be called when another request is available, making all processing event-driven. In all other respects the processing is the same as above.
These basic structures are seen in all the WebSocket example applications.
wsLIB is a C code module that provides the basic infrastructure for building WebSocket applications to run as WASD scripting under both the models decribed above.
It abstracts much of the required functionality into a few callable functions using optional string descriptors so as to minimise dependency on the C language and on knowing the internals of the library data structure. The list of functions and associated parameters would unnecessarily clutter this document and so WebSocket application designers and programmers are referred to the descriptive prologue in the library code module itself (see below). While wsLIB usage is relatively straight-forward, the detail of any multi-threaded, asynchronous application can be daunting and so the example WebSocket applications (scripts) should be used as a wsLIB reference and tutorial.
wsLIB also contains routines for synchronising request acceptance and accessing CGI variables associated with that request. These variables are available for the period from request acceptance to the issuing of the CGIplus sentinal EOF indicating to the server the script is ready to accept another request. Any CGI variable values required during ongoing processing must be copied to request-specific storage. Again, the example WebSocket applications (scripts) should be used as a CGI variable processing reference and tutorial.
The library contains WATCH points. Network [x]Data and [x]Script provide a useful combination of traffic data. The library function WsLibWatchScript() allows WebSocket applications (scripts) to provide additional WATCHable information via the [x]Script item.
WASD_ROOT:[SRC.WEBSOCKET]WSLIB.C
The WASD WebSocket implementation provides a number of scripting examples illustrating WebSocket programming basics and the use of the WASD wsLIB library. All of these illustrate multi-client support using asynchonous I/O. Each has a server component (the C code) and a client component (the HTML file containing the JavaScript code).
The following examples concentrate on the server C code as this is WASD-specific. Any WebSocket reference can adequitely cover the essentials of the client JavaScript implementation.
This almost has to be the classic example of asynchronous, bidirectional communications without HTTP kludges. Each connected client can enter a message and it is distributed to all connected clients.
WASD_ROOT:[SRC.WEBSOCKET]WS_CHAT.C
Each connected client can enter a message which is then returned to them.
WASD_ROOT:[SRC.WEBSOCKET]WS_ECHO.C
The HTML/JavaScript/WebSocket client end connects to the script. Each mouse movement is then reported to the script. These data are distributed to all connected clients. This provides an asynchronous update facility from all clients to all clients.
The script is implemented using VMS I/O-driven ASTs. The code is also interesting because it implements all required functionality explicitly; no WebSocket library functions are employed.
WASD_ROOT:[SRC.WEBSOCKET]WS_MOUSE.C
WebSocket server applications are essentially CGIplus scripts and so are mapped and activated in the same fashion as any other CGIplus script (3.3 Other Considerations).
Throttle mapping rules may be applied to WebSocket requests. There is however, a fundamental difference between request throttling and WebSocket throttling though. HTTP request throttling applies control to the entire life of the response. WebSocket throttling applies only to establishing connection to the underlying server application. Once the script responds to accept the connection or reject it throttling is concluded.
Long-lived WebSocket connections are considered less suitable to full life-cycle throttling and should use internal mechanisms to control resource utilisation (i.e. using the delayed sentinal EOF mechanism described in 5.1 Multi-Client WebSocket Applications). Essentially it is used to limit the impact concurrent requests have on the number of supporting script processes allowed to be instantiated to support the application.
For example, the rule
To support concurrent requests distributed across multiple application scripts specify the throttle value as the number of separate scripts
To return a "too busy" 503 to clients (almost) immediately upon all processes become full and blocking (maximum application concurrency has been reached) then set the "t/o-busy" value to 1 second.
Unconditionally disconnects all WebSocket applications.
For VMS V8.2 and later, more selective disconnects are possible. Disconnects WebSocket applications with connect number, with matching script names, and with matching scripting account usernames, respectively.
CGI variable WEBSOCKET_VERSION provides the WebSocket protocol version number negotiated by the server at connection establishment.
At the time of writing the WebSocket protocol has just gone to IETF Draft RFC and has during development been very volatile and may continue to be so as it evolves. WASD supports the current base protocol number and any higher. At some time in the future it may be necessary to constrain that to a supported version number or set of numbers. Defining the logical name WASD_WEBSOCKET_VERSION to be one or more comma-separated numbers will limit the supported protocol versions. For example
This string is also used as the list of versions reported in a 426 (upgrade required) response when a client makes a request using an unsupported version.
The raw WebSocket throughput of a platform (hardware plus VMS plus TCP/IP stack plus WASD and optionally network infrastructure) can be measured using the WSB utility. Measures of raw message and byte throughput for a series of messages of various sizes can provide useful information on the underlying maximum messaging characteristics of that platform.
The following example shows usage on an Alpha XP1000:
For more information see the description in the prologue of the program. (A zero-size message is legitimate with the WebSocket protocol.)
The WASD RawSocket is a variant of, and is heavily based on, the WASD WebSocket infrastructure. It allows a service (listening host and port) to accept a connection and immediately activate a configured WASD CGIplus script to service that connection. Full-duplex, asynchronous, bidirectional input/output is supported. While being referred to as a "socket", of course it is not network socket communication (BSD or any other abstraction), but a read/write abstraction via mailbox I/O. Input is a stream of octets; output a stream of octets. The streams are opaque in this context and depend entirely on the abstraction (protocol) being implemented by the script. The "RawSocket" is just WASD nomenclature for WebSocket-style scripting without the WebSocket protocol.
The script when activated enters a CGIplus wait-service-wait loop using the standard CGIplus mechanism. When a request activates the script it issues a "100 Continue" CGI/HTTP header to signal the server that the request is accepted and now being processed. This header is NOT relayed to the client. The activation parameters (normally referred to as request parameters) are available via the usual CGI variables. The script can then read from and write to the RAWSOCKET_INPUT and RAWSOCKET_OUTPUT devices respectively. As with WebSocket applications, a single RawSocket application (script) can concurrently support multiple clients. The end-of-script is indicated by issuing an EOF to the mailbox. The script activation can be as long-lived as required and the server will not run a request down for any timeout reason.
RawSocket applications share the same lifecycle as described for WebSockets in 5.2 WebSocket Application.
rawLIB is a C code module that provides a similar API and functionality to the wsLIB library described in 5.3 WebSocket Library.
WASD_ROOT:[SRC.WEBSOCKET]RAWLIB.C
The WASD RawSocket implementation provides a number of scripting examples illustrating programming basics and the use of the WASD rawLIB library. All of these illustrate multi-client support using asynchonouse I/O. All use telnet as a client interface although this is only for convenience. RawSocket is completely protocol agnostic.
This almost has to be the classic example of asynchronous, bidirectional communications. Each connected client can enter a message and it is distributed to all connected clients.
WASD_ROOT:[SRC.WEBSOCKET]RAW_CHAT.C
Each connected client can enter a message which is then returned to them.
WASD_ROOT:[SRC.WEBSOCKET]RAW_ECHO.C
This example implements a simple-minded telnet server. Definitely not intended for production.
WASD_ROOT:[SRC.WEBSOCKET]RAW_PTD.C
RawSocket server applications are essentially CGIplus scripts and so are mapped and activated in the same fashion as any other CGIplus script (3.3 Other Considerations). They are a little unique in that there is generally a one-to-one relationship between a script and a service. The service is flagged as implementing a RawSocket and that service is mapped directly to a script.
It is possible to have conditional mapping based on the (rather limited) "request" parameters (WATCH is useful for understanding what request data is available). For example, the script activated may be varied on the client address.
6.1Requests and Responses |
6.2Code Examples |
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
During CGI or CGIplus processing (though not DECnet-based CGI) it is possible to suspend normal script output to the client and for the script to interact directly with the server, then resume output to the client. This may done more than once during processing. During the callout the script makes a request of the server and receives a response. These requests are to perform some server function, such as the mapping of a path to a file specification, on behalf of the script. Naturally, this makes the script no longer portable, but may be extrememly useful in some circumstances.
It is this general callout mechanism that allows specific authentication agents (Authentication and Authorization in WASD Features) to be implemented as standard CGIplus scripts.
The mechanism is quite simple.
This is a basic callout. Between the callout escape and end-of-text sequences multiple request/responses transactions may be undertaken.
The request record is plain text, comprising a request key-word (case-insensitive), full-colon, and following optional white-space and parameter(s). It is designed not to be implementation language specific.
The response record is also plain-text. It begins with a three-digit response code, with similar meanings and used for the same purpose as HTTP response codes. That is 200 is a success status, 500 a server error, 400 a request error, etc. Following the response code is white-space and the plain text result or error message. A response to any given callout request may be suppressed by providing a request record with a leading ! (exclamation point) or # (hash symbol).
WASD v12.0 and later agents require the agent to declare its role and typically provide some information about itself. An example callout is shown below.
WASD v12.0 and later agents are required to declare process concluded, and optionally to return any results to the calling code.
If the specialized /PROFILE capability is enabled (SYSUAF Security Profile in WASD Features) this can determine whether the specified file name is allowed access by the request's username.
Create a temporary global section to act as a memory buffer shared between a script process and the server. The default is Megabytes. See 2.2.3 Bulk Content Output for a description of the purpose and use of all the BUFFER-.. callouts.
Dispose of the shared memory buffer created by callout BUFFER-BEGIN.
Instruct the server to write <integer> bytes from the shared memory buffer to the client.
Specialised callout allowing an access agent, lookup agent or meta agent to insert, modify or delete a CGI variable for a request. These agents are currently only proof-of-concept. See sources in
This callout is used to indicate to the server that a CGIplus script can process the CGI variables in "struct" mode. By default each CGI variable is transfered to a CGIplus script one "record" at a time. In "struct" mode all variables are transfered in a single, binary I/O which must the be parsed by the the script. It is of course a much more efficient method for CGIplus (‘Struct-Mode CGIplus’ in 3.1 CGIplus Programming).
Provide Content Security Policy (CSP) configuration(s) from a script.
1.9 Content Security Policy (CSP)
Content Security Policy (CSP) in WASD Configuration
Specialised callout allowing an access agent, lookup agent or meta agent to insert, modify or delete a dictionary entry. These agents are currently only proof-of-concept. See sources in
When using the raw TCP/IP socket for output (11. Raw TCP/IP Socket) this callout is used to notify the server that the script will be using the gateway device and the HTTP status code (e.g. 200, 302, 404, etc.)
When using the raw TCP/IP socket for output (11. Raw TCP/IP Socket) this can be used to change the BG: device carriage-control. A value of 1 enables a <CR><LF> with each record (the default), while 0 disables it. This is analagous to the APACHE$SET_CCL utility.
When using the raw TCP/IP socket for output (11. Raw TCP/IP Socket) this callout is used to notify the server of the quantity of data transfered directly to the client by the script.
Sets/resets a scripting process' lifetime which may be expressed as an integer number of minutes or in the format hh:mm:ss. For instance, use to give frequently used CGIplus scripts an extended lifetime before being rundown by the server (override the [DclCgiPlusLifeTime] configuration parameter). Specifying "none" (or -1) gives it an infinite lifetime, zero resets to default.
Map the supplied file specification to its URL-style path equivalent, and against the server's mapping rule. This does not check the file name is legal RMS syntax.
Map the supplied URL-style path against the server's rule database into a VMS file specification. Note that this does not verify the file name legaility or that the file actually exists.
No operation. Just return a success response.
Place the supplied string into the server process log. Used to report incidental processing or other errors.
Send the supplied string to OPCOM.
Equivalent to the script issuing a "Script-Control:" response header field (although of course some script control directives will not apply after header generation).
Specifies the period for which a script continues to execute if the client disconnects. Overrides the WASD_CONFIG_GLOBAL [DclBitBucketTimeout] confiuration directive.
Sets/resets a script request lifetime (in minutes, overrides the [TimeoutOutput] configuration parameter). Specifying "none" (or -1) gives it an infinite lifetime, zero resets to default.
Sets/resets a script request no-progress (in minutes, overrides the [TimeoutNoProgress] configuration parameter). The no-progress period is the maximum number of seconds that there may be no output from the script before it is aborted. Specifying "none" (or -1) gives it an infinite lifetime, zero resets to default.
The record-oriented callout sequences and request/response makes implementation quite straight-forward. The following C language and DCL procedure code fragments illustrate the basics.
Also see working examples in WASD_ROOT:[SRC.CGIPLUS]
7.1CGIsapi |
7.2Writing ISAPI Scripts |
7.3Server Configuration |
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
ISAPI (procounced eye-sap-ee) was developed by Process Software Corporation (the developer of Purveyor Encrypt Web Server available under VMS), Microsoft Corporation and a small number of other vendors. It has an software infrastructure similar to CGI but a different architecture. It is designed to eliminate the expensive process creation overheads of CGI (under Unix, let alone VMS), reduce latency for expensive-to-activate resources, and generally improve server throughput, particularly on busy sites.
Unlike standard CGI, which creates a child process to service each request, ISAPI is designed to load additional sharable code (DLLs, or Dynamic Link Libraries in MSWindows, shareable images under VMS) into the Web server's process space. These are known as server extensions. This radically reduces the overheads of subsequent request processing and makes possible server functionality that can maintain resources between requests (for instance keep open a large database), again contributing to reduced latency and increased throughput.
Of course there is a down-side! Loading foreign executable code into the server compromises its integrity. Poorly written extensions can seriously impact server performance and in the worst-case even crash a server process. The other significant concern is the multi-threaded environment of most servers. Extensions must be carefully constructed so as not to impact the granularity of processing in a server and to be thread-safe, not creating circumstances where processing corruption or deadlock occurs.
WASD provides an environment for executing ISAPI based extensions. Unlike classic ISAPI the DLLs are not loaded into server space but into autonomous processes, in much the same way as CGIplus scripts are handled (3. CGIplus). This still offers significantly improved performance through the persistance of the ISAPI extension between requests. Measurements show a potential five-fold, to in excess of ten-fold increase in throughput compared to an equivalent CGI script! This is comparable to reported performance differences between the two environments in the Microsoft IIS environment.
While the script process context does add more overhead than if the DLL was loaded directly into the server process space, it does have two significant advantages.
WASD implements the ISAPI environment as an instance of its CGIplus environment. CGIplus shares two significant characteristics with ISAPI, persistance and a CGI-like environment. This allows a simple CGIplus wrapper script to be used to load and interface with the ISAPI DLL. After being loaded the ISAPI-compliant code cannot tell the difference between the WASD environment and any other vanilla ISAPI one!
This wrapper is known as CGIsapi (pronounced see-gee-eye-sap-ee).
Wrapping another layer does introduce overhead not present in the native CGIplus itself, however measurements indicate in the real world (tm) performance of the two is quite comparable. See Server Performance in WASD Features for further information. The advantage of ISAPI over CGIplus is not performance but the fact it's a well documented interface. Writing a script to that specification may be an easier option, particularly for sites with a mixture or history of different Web servers, than learning the CGIplus interface (simple as CGIplus is).
This section is by-no-means a tutorial on how to write for ISAPI.
First, get a book on ISAPI. Second, ignore most of it! Generally these tomes concentrate on the Microsoft environment. Still, information on the basic behaviour of ISAPI extensions and the Internet Server API is valuable. Other resources are available at no cost from the Microsoft and Process Software Corporation sites.
Have a look at the WASD example DLL and its build procedure in WASD_ROOT:[SRC.CGIPLUS]
The CGIsapi wrapper, WASD_ROOT:[SRC.CGIPLUS]CGISAPI.C, is relatively straight-forward, relying on CGIplus for IPC with the parent server process. A brief description of the detail of the implementation is provided in the source code.
CGIsapi has a simple facility to assist with debugging DLLs. When enabled, information on passed parameters is output whenever a call is made to an ISAPI function. This debugging can be toggled on and off whenever desired. Once enabled DLL debugging remains active through multiple uses of a CGISAPI instance, or until disabled, or until the particular CGISAPI process' lifetime expires. Check detail in the CGIsapi source code description.
The wrapper is designed to be ISAPI 1.0 compliant. It should also be vanilla ISAPI 2.0 compliant (not the Microsoft WIN32 variety, so don't think you'll necessarily be able to grab all those IIS extensions and just recompile and use ;^)
With CGIsapi multiple instances of any one extension may be active on the one server (each in an autonomous process, unlike a server-process-space loaded extension where only one would ever be active at any one time). Be aware this could present different concurrency issues than one multiple or single threaded instance.
When CGIplus processes are idle they can be run-down at any time by the server at expiry of lifetime or to free up required server resources. For this reason ISAPI extensions (scripts) should finalize the processing of transactions when finished, not leave anything in a state where its unexpected demise might corrupt resources or otherwise cause problems (which is fairly good general advice anyway ;^) That is, when finished tidy up as much as is necessary.
CGIsapi loaded extensions can exit at any time they wish. The process context allows this. Of course, normally a server-process-space loaded instance would not be able to do so!
For other technical detail refer to the description with the source code.
Scripting processes may be purged or deleted using ( "Techncial Overview, Server Command Line Control"):
Ensure the following are in the appropriate sections of WASD_CONFIG_GLOBAL.
Ensure this rule exists in the scripting section of WASD_CONFIG_MAP.
With this rule DLLs may be accessed using something like
8.1Script System Environment |
8.1.1Proxy Access |
8.1.2DECnet Objects |
8.1.3Reducing Script Latency |
8.1.4DECnet/OSU Startup |
8.2CGI |
8.3OSU (DECthreads) Emulation |
8.4User Scripts |
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
"Imitation is the sincerest form of flattery" – proverb
By default WASD executes scripts within detached processes, but can also provide scripting using DECnet for the process management. DECnet scripting is not provided to generally supplant the detached process-based scripting but augment it for certain circumstances:
Any DECnet based processing incurs some overheads:
As of version 5.2 WASD provides reuse of DECnet connections for both CGI and OSU scripting, in-line with OSU v3.3 which provided reuse for OSU scripts. This means multiple script requests can be made for the cost of a single DECnet connection establishment and task object activation. Note that the OSU task procedure requires the definition of the logical name WWW_SCRIPT_MAX_REUSE representing the number of times a script may be reused. The WASD startup procedures can provide this.
In practice both the WASD CGI and OSU scripts seem to provide acceptable responsiveness.
DECnet-based scripts are mapped using the same rules as process-based scripts, using the SCRIPT and EXEC rules (Mapping User Directories in WASD Configuration) for general information on mapping rules). DECnet scripts have a DECnet node and task specification string as part of the mapping rule. There are minor variations within these to further identify it as a WASD or an OSU script (8.4 User Scripts).
The specification string follows basic VMS file system syntax (RMS), preceding the file components of the specification. The following example illustrates declaring that paths beginning with FRODO will allow the execution of scripts from the CGI-BIN:[000000] directory on DECnet node FRODO.
In similar fashion the following example illustrates a script "frodo_show" that might do a "SHOW SYSTEM" on node FRODO. Note that these rules are case-insensitive.
Both of the above examples would use the WASD CGI DECnet environment (the default if no task specification string is provided). By including task information other environments, in particular the OSU scripting enviroment, can be specified for the script to be executed within. The default task is named CGIWASD and can also be explicitly specified (although this behaviour would be the same as that in the first example)
All task specification strings may also use zero as the task abbreviation.
To execute a script within the OSU environment specify the standard OSU task executive WWWEXEC, as in the following example:
This would allow any URL beginning with "/osu/" to execute a script in the OSU environment.
By default the script process is created using the HTTPd scripting account (usually HTTP$NOBODY, although versions prior to 8.1 have used HTTP$SERVER). It is possible to specify alternate accounts for the scripts to be executed within.
The first examples are explicitly specifying an account in the script rule.
It is also possible to have scripts that have been subject to SYSUAF authorization executed within the authenticated account. The dollar symbol in the following examples directs the server to substitute the authenticated username into the access string.
The set script=as= rule used for PERSONA controlled process scripting can also be applied to DECnet scripts. This includes explicitly specified usernames as well as SYSUAF authenticated usernames. The server creates an appropriate access string when the script is activated.
User scripts can also be activated using these rules, either explicitly specifying the "~" in an access string or using the set script=as= mapping rule.
See 8.4 User Scripts for more detail.
To specify any script to execute on the same system as the HTTP server specify the node name as zero or SYS$NODE.
Mapping rules are included in the examples WASD_ROOT:[EXAMPLE] providing this. After the DECnet environment has been started any CGI script may be executed on the local system via DECnet by substituting "/decnet/" for "/cgi-bin/" as the script path, and any OSU script available by using "/osu/". Behaviour is indeterminate, though it shouldn't be catastrophic, if one is invoked using the incorrect path (i.e. an OSU script using /decnet/ or a CGI script using /osu/).
The target system must have sufficient of the WASD server environment to support the required CGI script activation and activity. If the target system is actually the same system as the HTTP server then it already exists, or if part of the local system's cluster, then providing this should be relatively straight-forward. If the target system has none of the server environment then at a minimum it must have the logical name CGI-BIN defined representing the directory containing the required DECnet object procedure and scripts. The following fragment illustrates this:
The local system must have proxy access to each target scripting system (even if that "target" system is the same system as the HTTP server). This involves creating a proxy entry in each target hosts's authorization database. The following example assumes the existance of a local HTTP$NOBODY account. If it does not exist on the target node then one must be created with the same security profile as the HTTP server's.
The zero represents the system the server is currently executing on.
It is necessary to ensure the account has permission to write into its home directory. A network process creates a NETSERVER.LOG (Phase-IV) or NET$SERVER.LOG (DECnet-Plus) file in the home directory, and will fail to start if it cannot!
To provide DECnet scripting DECnet object(s) must be specified for any system on which the scripts will be executed. The DECnet object is the program or procedure that is activated at the target system inside of a network-mode process to interact with the HTTP server.
DECnet-Plus uses the NCL utility to administer the network environment. The following NCL scripting shows the creation of a network application for the WASD CGI object:
To create a DECnet-Plus OSU WWWEXEC object:
These must be executed at each system (or server) startup, and may be executed standalone, as illustrated, or incorporated in the NCL script SYS$STARTUP:NET$APPLICATION_STARTUP.NCL for automatic creation at each system startup. Examples may be found in WASD_ROOT:[EXAMPLE]
DECnet Phase-IV uses the NCP utility to administer the network environment. The following NCP commands may be used each time during server startup to create the required DECnet objects. With Phase-IV the SET verb may be replaced with a DEFINE verb and the commands issued just once to permanently create the objects (a SET must also be done that first time to create working instances of the DEFINEd objects).
To create a DECnet CGI object:
To create a DECnet OSU WWWEXEC object:
Examples may be found in WASD_ROOT:[EXAMPLE]
Script system network process persistance may be configured using NETSERVER logical names. These can control the number and quiescent period of the server processes. These logical names must be defined in the LOGIN.COM of the HTTP server account on the target script system.
The example STARTUP.COM and STARTUP_DECNET.COM procedures found in the WASD_ROOT:[EXAMPLE] directory provide the essentials for DECnet/OSU scripting. If the INSTALL.COM startup environment is used setting the PROVIDE_DECNET symbol to 1 in STARTUP.COM will create the DECnet scripting environment during server startup.
CGI scripts that use HTTP GET (the default) may be transparently executed within the DECnet scripting environment. This means that the script is executed within a network process, on the target system (which could be the local system), instead of within a process on the local system. Other than that the WASD DECnet CGI environment behaves identically to the standard (sub)process CGI environment. CGIplus scripting is not supported and if CGIplus-only scripts are executed the behaviour is indeterminate.
Scripts that wish to use HTTP POST will need to read the request body from the NET$LINK stream, rather than from HTTP$INPUT as with (sub)process based scripts. End of body is indicated by an empty record rather than EOF. Scripts may quite simply be made to function appropriately in both environments. The following C code fragment illustrates this.
An example of making the HELP database on a system other than that hosting the HTTP server (using the CONAN script) would be done using the mapping rules
The WASD HTTP server provides an emulation of the OSU scripting environment. This is provided so that OSU-based scripts (both CGI-style and OSU-specific) may be employed by WASD with no modification. As this emulation has been designed through examining OSU code and lots of trial and error its behaviour may be incomplete or present errors. A list of OSU scripts known to work with WASD is provided at the end of this section (‘Known Working Scripts’ in 8.3 OSU (DECthreads) Emulation).
Supported scripts include only those that depend on the OSU WWWEXEC object and dialog for all functionality. Any script that uses other OSU-specific functionality is not supported. Interactions between WASD's and OSU's authentication/authorization schemes may be expected.
OSU scripts expect to get the path information unmapped, whereas WASD always additionally maps any path after the script component has been derived from the request URI. It may be necessary to ensure OSU scripts are activated with the associated path SET to provide what they expect. For example:
The author would like to know of any OSU scripts the WASD emulation barfs on, and will attempt to address the associated limitation(s) and/or problem(s).
Software necessary for supporting the OSU scripting environment (e.g. WWWEXEC.COM) and selected OSU scripts (mainly for testing purposes) have been extracted from the OSU v3.4 package and included in the WASD_ROOT:[SRC.OSU] directory. This has been done within the express OSU licensing conditions.
An example DECnet and OSU scripting startup may be found in WASD_ROOT:[EXAMPLE]. This should be called from or used within the HTTP server startup. It includes two logical definitions required for common OSU scripts. Other tailoring may be required for specific OSU scripts.
David Jones, the author of the DECthreads (OSU) HTTP server, outlines his reasons for basing OSUs scripting on DECnet (reproduced from a USENET NEWS reply to a comment this author made about DECnet-based scripting).
The OSU server's DECnet scripting is not based on arbitrary considerations. This author does not disagree with any of the concerns, and as may be seen from WASD documentation the design of WASD also directly addresses points 1, 3 and 5 with the use of persistant processes and CGIplus. Certainly DECnet-based scripting addresses the very legitimate point 4 (and also allows nodes with specific resources to participate without installing full HTTP server environments). Point 2 is not an issue with the use of detached scripting processes, or for all practical purposes addressed by adjusting account quotas to support the required number of subprocesses. Point 6 is only too true (possibly at least until Java servers and servlets become ubiquitous :^)
The following is a list of OSU-specific scripts that the WASD implementation has either been developed or tested against, and any installation notes or other WASD specifics. The author would like to know of any OSU scripts the WASD emulation has problems or works successfully with.
Comment out the Conan The Librarian mappings for the "/help" path and provide the following in WASD_CONFIG_MAP:
Yes, backward compatibility can be provided for those old OSU .HTMLX files in your new WASD environment ;^) All that is needed is a file type mapping to the script in the WASD_CONFIG_GLOBAL configuration file.
The WASD DECnet environment provides a simple mechanism for executing scripts within accounts other than the server's. This allows configured users to write and maintain scripts within their own areas and have them execute as themselves. Both standard CGI and OSU scripting may be provided for with this facility.
Of course there is always a down-side. Be careful to whom this capability is granted. User scripts are executed within a user network-mode process created by DECnet. Script actions cannot generally affect server behaviour, but they can access any WORLD-readable and modify any WORLD-writable resource in the system/cluster, opening a window for information leakage or mischievous/malicious actions. Script authors should be aware of any potential side-effects of their scripts and Web administrators vigilant against possible destructive behaviours of scripts they do not author.
User scripting is not enabled by default. To provide this facility mapping rules into the user area must be provided in much the same way as for user directories, see (Mapping User Directories in WASD Configuration).
The "EXEC" rule provides a wildcard representation of users' script paths. As part of this mapping a subdirectory specifically for the web scripts should always be included. Never map users' top-level directories. For instance if a user's account home directory was located in the area WWW_USER:[DANIEL] the following rule would potentially allow the user DANIEL to provide scripts from the home subdirectory [.WWW.CGI-BIN] using the accompanying rules (first for CGI, second for OSU scripts):
Scripts located in these directories are accessible via paths such as the following:
For each user account permitted to execute local scripts proxy access to that account must be granted to the HTTP server account.
For example, the following would allow the HTTP server to execute scripts on behalf of the username DANIEL.
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
WASD supports a number of scripting engines.
Earlier releases of the WASD package included some of these in the basic package. Due to the growing number, greater complexity of the environments, and increasing version dependencies, these environments will be distributed independently of the main WASD package. Current releases may be found at the main WASD download site
Pages generated by scripting environments can optionally be cached by the server. For a certain class of script output this can offer reduced response latency and system impact. See 1.4 Caching Script Output.
Java classes may be used to perform CGI/CGIplus scripting with WASD. This is not Java Server Pages, Tomcat, or anything of the like. The Java refered to here is a small, self-contained Java environment that may used with WASD "out-of-the-box". All you need is java installed on your VMS system. These may be designed as standard CGI scripts (with the inevitable latency of the class loading) or as CGIplus scripts (with the attendant benefit of lower latency).
WASD provides a class to allow a relatively simple interface to the CGI environment for both GET and POST method scripts. This and a collection of demonstration scripts may be found in the WASD_ROOT:[SRC.JAVA] directory.
As the Java environment is constantly under development, both as a platform-independent environment and on the VMS platform in particular, it is possible that the latest VMS Java kit may not integrate well with the WASD Java environment. Of course every effort will be made to keep the WASD Java environment current.
Java CGI/CGIplus scripts must always be mapped and executed using the CGIplus path, however some can behave as standard CGI scripts, exiting after responding to the request, while others can persist, responding to multiple requests (3. CGIplus). The CGIplus path is always necessary as Java does not have direct access to a process' general environment, the traditional way of passing CGI variables, so the WASD implementation uses the CGIplus data stream to provide CGI information.
Ensure the Java class file type is mapped to the Java run-time in the WASD_CONFIG_GLOBAL configuration file.
The following content types are configured, also in WASD_CONFIG_GLOBAL.
Class files should be copied to the [CGI-BIN] directory (where all architecture neutral script files should be located).
Getting carriage-control to make sense is often a challenge. System.out.print() only expresses carriage-control embedded in the string. System.out.println() the same but then issues an independent linefeed (the newline of the ln) which appears to WASD as an empty record. Choose your poison (and antidote). Using the "Script-Control: X-stream-mode", "Script-Control: X-record-mode" or "Script-Control: X-record0-mode" can assist WASD interpreting the output. See 2. CGI.
WASD supports Perl scripting in the CGI, CGIplus and RTE environments. Generally no source changes are required to use standard CGI Perl scripts! Information in this section pertains specifically to VMS Perl 5.6 and following. Earlier versions may have some limitations. VMS Perl 5.6 is a relatively complete Perl implementation and standard distributions contain some VMS-specific functionality. In particular the VMS::DCLsym and VMS::Stdio can make life simpler for the VMS perl developer.
Users of VMS Perl are directed to "Perl on VMS" at http://www.sidhe.org/vmsperl/ providing access to the latest release of Perl for VMS.
There are a number of ways to activate a Perl script under VMS. Any of these may be used with the WASD server. If the script file is accessible via the exec or script rules of the WASD_CONFIG_MAP configuration file it can be activated by the server. The simplest example is to place the scripts somewhere in the CGI-BIN:[000000] path and execute via /cgi-bin/, although in common with other scripts it may be located anywhere a rule provides a path to access it (1.6 Script Mapping).
When Perl is available from the command-line, either as a DCLTABLES defined verb, a DCL$PATH available verb, or as a foreign verb. The script (the file containg the Perl source) is provided to the Perl interpreter as a parameter to the Perl verb.
If DCL pre-processing, or some other specific environment needs to be set up, the activation of the Perl script can be placed inside a DCL wrapper procedure. This is often used to allow the transparent activation of Perl scripts via the DCL$PATH mechanism.
The Perl source is embedded as in-line data within a DCL procedure.
Due to changes in environment handling sometime between versions 5.0 and 5.6 it was impossible to access DCL symbols via the %ENV hash, making CGI-based scripts impossible to use under VMS Web servers without modification. Version 5.6 addresses this issue by providing a versatile mechanism for controlling where the environment variables are manipulated. The logical name PERL_ENV_TABLES specifies this location, or if defined as a search list, the locations.
Name | Location |
---|---|
CRTL_ENV | C run-time environment array (i.e. getenv()) |
CLISYM_LOCAL | get DCL symbols, set local |
CLISYM_GLOBAL | get DCL symbols, set global |
logical name table | any logical name table, including LNM$FILE_DEV |
For WASD Perl scripting it is recommended that this be defined as CLISYM_GLOBAL,LNM$PROCESS. The CLISYM_GLOBAL allows access to the CGI variable environment, and LNM$PROCESS to significant logical name definitions for the subprocess (e.g. HTTP$INPUT and callout sequences). This can be done on a system-wide basis (i.e. for all Perl scripting) using
Requests using the POST method contain all the content in the body of the request. In particular, requests generated via HTML <form> contructs do not deliver the form data via the request query string, it is provided in a URL-form-encoded body. This requires some explicit processing to recover the form elements. A number of Perl CGI modules exist to ease this chore, including the most popular CGI.pm. All of these should work in the VMS environment, and of course then with WASD.
For POSTed requests it is necessary for the script to have access to the request body. In Unix environments this is available via the <stdin> stream, and under Perl via STDIN, <>, etc. This equates to SYS$INPUT under VMS.
With WASD, when activating the .PL script file directly via a [DclScriptRunTime] entry (i.e. without a DCL procedure wrapper) STDIN is directly available without further issues.
If the script has a DCL wrapper procedure the DCL CLI has control of the SYS$INPUT stream and it becomes necessary to temporarily redirect this for the duration of the script. WASD provides the HTTP$INPUT process-level logical name to identify the script body stream (along with WWW_IN and APACHE$INPUT names for easing script portability). The redirection is simply done, as shown in the following example.
If the script is embedded in a DCL procedure the DCL CLI is using SYS$INPUT to provide the script source to the Perl interpreter and so is completely unavailable for use. The request body is still available to the script however but must be explicitly read from HTTP$INPUT. This example provides the basics.
Perl is an interpreter, meaning scripts are provided and activated as source form, the interpreter processing the program "on-the-fly". Perl actually translates the entire script into an intermediate form before beginning execution. This has the advantage of discovering and reporting syntax errors before beginning any actual processing, and also improves the final run-time performance.
While having Perl an interpreter eases development and portability it does incur a performance penalty, particularly in activation latency, due to both interpreter image activation, and script and associated Perl module preprocessing. With standard CGI, where each request processed is handled as an autonomous activation, this becomes quite noticable and can have significant system impact.
WASD provides two solutions for this and other persistent scripting issues. Both of these require the supplementary Perl package available from the WASD download page. Both are briefly described below.
CGIplus substantially eliminates the overhead associated with CGI processing by allowing the subprocess and any associated image/application to continue executing between uses (3. CGIplus). The good news is, CGIplus is relatively simple to support, even using Perl. The great news is, CGIplus can reduce latency and improve performance by some twenty-fold!!
With CGIplus the Perl script remains active for the life of the subprocess. That is it persists! Read the general philosphy and implementation details in the above reference. Note that it is still substantially CGI! The notable differences are two. CGI variables are obtained by reading a stream, not using the %ENV hash. The end-of-script is indicated by writing a special byte sequence (detected and used by the server). Of course the request body is still available via the usual stream.
Using the basic principles described in the CGIplus Chapter a Perl CGIplus script would be relatively simple to build from scratch. To assist in deploying CGIplus Perl scripting a CGIplus.pm Perl module has been provided as part of the supplementary package.
A Run-Time Environment (RTE) is almost identical to CGIplus. It allows an environment to persist between requests, substantially improving response latency and reducing system impact (4. Run-Time Environments). There is a significant difference between RTE and CGIplus scripts. With CGIplus the script itself persists between uses, retaining all of its state. With an RTE the script does not persist or retain state, only the RTE itself.
The WASD RTE Perl interpreter contains an embedded Perl engine and an associated Perl module that allows multiple scripts to be activated, preprocessed once and remain loaded read-to-run. This eliminates the overhead associated with activating the interpreter and Perl script with each request. This mechanism parallels the Apache perl_mod module and works on substantially unmodified CGI scripts. The test-bench indicates an improvement of some twenty-five fold!
These are the configuration requirements for using the basic CGI Perl.
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
Callout processing may redact (completely rewrite and restart) a request.
To do this a script must use the REDACT:<opaque> callout to send back to the server a completely new request header and body (if applicable) which the server then treats as if received from the client over the network. This allows a request to be partially or completely rewritten (as required) and restarted. The data supplied to this callout is treated as completely opaque and care must be taken to include all and no extra carriage-control, etc.
Request redaction may only be initiated (using the REDACT: callout) if the CGI response header has not been sent. Once request redaction has been initiated no CGI output subsequently can be generated. The server will generate an error if such a protocol error occurs.
The REDACT-SIZE:<integer> callout may be used prior to any REDACT: callout. By default the server allocates memory on demand to accomodate the redacted request. If the redacted request is large (more than [BufferSizeDclOutput]) and the total size of the redacted request known in advance there is some efficiency in requesting the server to preallocate this amount of space using the REDACT-SIZE: callout.
An elementary (and somewhat contrived) example:
Once the request has been redacted the script just finishes processing without other output and the server transparently restarts processing.
An actual usage example may be found in the WASD PAPI authentication agent (not a component of the standard WASD package).
This facility was originally incorporated to allow a PAPI
http://papi.rediris.es/ |
authentication agent to store a request on-disk and then some time and several processing steps later restart the original request processing.
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
For detached and subprocess scripting the raw TCP/IP socket can be made available for scripts to transfer data directly to the client. The socket BGnnnn: device name is made available via the CGI variable WWW_GATEWAY_BG. This is enabled using the [DclGatewayBg] configuration directive. As it is a completely raw stream it cannot be used, and is not made available for SSL ("https:") encrypted requests.
Although one might imagine this direct transfer to be significantly more efficient than the standard script mailbox the test-bench indicates that to all purposes it provides a negligable improvement in throughput, even under high load. It probably only translates into measurable benefits for scripts producing large quantities of output (for instance hundreds of thousands or millions of bytes). For the average script the overhead of opening a stream to the raw TCP/IP device (which is not insignificant) and complications of the callout requirements isn't worth the effort. Still, it's there if someone wants or requires it.
The socket is created shareable between processes, and so a channel may be assigned by the script subprocess and have data written to it. The data is raw, in the sense the script must provide all carriage control, etc. All data transfered this way is outside of the server and so may not be WATCHed, etc.
The script must supply a full HTTP response. This means a NPH-style header (2.2.2 Non-Parsed-Header Output) and body, with full carriage-control as required, etc. The server must be notified that the script is using the gateway device, by providing a CGI callout (6. CGI Callouts) before any script output and after output is concluded. The first callout provides the response HTTP status code, the second the number of bytes transfered. These are also required for correct logging of the request. If a channel to the BG: device is opened it should always be closed when it is finished with. Failure to do so could lead to resource starvation for the server.
The /wasd_root/src/other/gateway_bg_example.com example script demonstrates the use of the raw socket from DCL. The priciples can be applied to any scripting laguage.
The following code fragment shows the essential requirements using the C language.
By default the TCP/IP BG device driver supplies a <CR><LF> sequence as carriage control for each record. This supports record-oriented output such as DCL and various VMS utilities but is an issue when needing to output a binary object such as a large graphic. The CGI callout (6. CGI Callouts) GATEWAY-CCL: directive allows the device carriage control to be set and reset programmatically. A value of 1 enables a <CR><LF> with each record, while 0 disables it. This is analagous to the APACHE$SET_CCL utility.
Not all vendor's TCP/IP package BG drivers, or not all older versions, may support the C_SHARE option when creating sockets. Symptoms may range from it being ignored (and the script being unable to open a channel to the BGnnnn: device) to an error being reported as the socket is being created (and the server being unable to start at all). If this occurs merely disable the [DclGatewayBg] configuration option. Script output is of course still available via the standard script output mailbox.
For portability scripts that use the raw socket for output should always use a construct similar to the above example code so only to redirect output when the GATEWAY_BG device is indicated as available.
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |
This package uses the Expat XML parsing toolkit.
This package uses essential algorithm and code from Flexible and Economical UTF-8 Decoder.
This package contains software made available by the Free Software Foundation under the GNU General Public License.
This package contains software provided with the OSU (DECthreads) HTTP server package, authored by David Jones:
This product can include software developed by the OpenSSL Project for use in the OpenSSL Toolkit (https://www.openssl.org/).
This package uses SHA-1 hash code.
This software contains code derived in part from RSA Data Security, Inc:
SortTable version 2
Stuart Langridge, http://www.kryogenix.org/code/browser/sorttable/
nghttp2 - HTTP/2 C Library
Tatsuhiro Tsujikawa, https://github.com/tatsuhiro-t
VSI OpenVMS,
VSI TCP/IP Services for OpenVMS,
VSI C
are registered trademarks of VMS Software Inc.
OpenVMS,
HP TCP/IP Services for OpenVMS,
HP C,
Alpha,
Itanium and
VAX
are registered trademarks of Hewlett Packard Enterprise
MultiNet and TCPware are registered trademarks of Process Software Corporation
↩︎ | ↖︎ | ↑︎ | ↘︎ | ↪︎ |