3 *) [WIN32] Fix a delay in handling large POSTs to named pipe based
4 servers. Philip Gladstone [philip@okena.com]
6 *) [*nix/AP2] Prevent the module from being initalized twice at startup
7 (resulting in confusing error messages to the log).
9 *) Eliminate the need for SetHandler or AddHandler with static or
10 external applications.
12 *) Limit PM requests to start a dynamic application
13 to 5sec to prevent endless spinning (this is a drop-dead
14 limit that should only occur if the socket/named_pipe directory
15 is removed out from under a running server).
17 *) [*nix] Change the default socket directory from /tmp/fcgi to:
19 Apache2 - RUNTIMEDIR/fastcgi
21 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
22 for use with wrappers (in lieu of finding the user/group associated
23 with a virtual host - under Apache2 this isn't accomodated).
25 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
26 apr_proc_create() changes.
28 *) Log when invoking and restoring the restart backoff policy.
30 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
31 OVERLAPPED structure).
33 *) Eliminate need for dummy files for external servers under Apache2
35 *) Fix auth compatibility mode handling for access checker and authorizer
37 *) Fix HEAD request handling. Based on a patch by
38 Chris Lightfoot [chris@ex-parrot.com]
40 *) [*nix] When autoupdate is enabled touch the socket when restarting
41 the processes to prevent further requests.
42 Eckebrecht von Pappenheim [evp@heise.de]
44 *) Apache 2.0 support.
46 *) [WIN32] Don't read from a potentially closed named pipe.
47 Philip Gladstone [philip@okena.com]
49 *) Require the Apache version 1.3.6 or later to eliminate some signal
52 *) [WIN32] Use asyncronous io with named pipes instead of polled
53 nonblocking io. This should eliminate the last of the npipe issues.
55 *) Handle an application returning a complete and valid response without
56 having consumed all of the data sent to it.
58 *) Consume remaining client data (RESPONDERs only) if any.
60 *) Add support for backing off attempts to start applications that continuously
61 fail to start. Three new macros defined in mod_fastcgi.h control this
62 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
64 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
65 applications that do not (properly) support the shutdown event (this
66 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
67 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
68 interval between signaling a proper shutdown and wacking the process(s)
69 with a TerminateProcess().
71 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
72 setting it was just plain broken.
74 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
76 *) Fix 'FastCgiConfig -autoUpdate'.
78 *) Fix 'FastCgiConfig -flush'.
80 *) Prevent silly maxProcesses and processSlack combinations.
81 Dmitry Dorofeev [dima@yasp.com]
83 *) Properly handle the killing of idle processes when one takes a long time
84 to exit once signaled down (or the config is funky).
85 Dmitry Dorofeev [dima@yasp.com]
87 *) Always kill the youngest instance of an application. Suggested by
88 Dmitry Dorofeev [dima@yasp.com]
92 *) Delay the logging of write errors to the pm to account for shutdown/restart.
94 *) (Win32) An assortment of fixes.
96 *) Fix some broken casts that were likely the cause of an assert.
98 *) Win32. Eliminate forward slashes from the named pipe path name.
99 Gerald Richter [richter@ecos.de]
101 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
102 applications it spawns. [ryans@amazon.com]
104 *) Added support for the -flush argument to FastCgiConfig.
105 Eric Sit [esit@alum.mit.edu]
107 *) Change the "which call to module_init() is this" check to a more
108 reliable approach. Doru Petrescu [pdoru@kappa.ro]
110 *) Close the old pipe file descriptor in apache main on USR1/HUP
111 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
113 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
114 [tetsuya@secom-sis.co.jp]
116 *) Eliminate the use of locks to assist in the clean shutdown of
117 applications. Instead, it is assumed that applications handle
118 termination signals properly (this is now embedded in the C
121 *) Fix Win32 process termination. Proper operation requires the use of
122 an updated application lib (termination is now signalled with an
123 Event and handled by specialized thread).
125 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
127 *) Added code so if the last instance of a dynamic application died without
128 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
129 configuration allows the last instance to be killed, then allow it to die).
130 Andrew Benham [adsb@bigfoot.com]
132 *) Fix the loadFactor calculation used to determine when dyanmic
133 applications could be killed off due to low demand [adsb@bigfoot.com].
135 *) Fix a deadlock condition that could occur with Win32 named
138 *) Fix a potential deadlock condition when FastCGI application
139 sent responses while still reading the client request (POST data).
143 *) Allow absolute pathnames in the -socket argument. Suggested by
144 Christian Jaeger [christian.jaeger@sl.ethz.ch].
146 *) Don't invoke suexec when the user/group for the fastcgi application
147 is the same as the apache main server. This is consistent with
148 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
150 *) Reset the apache drop dead timer upon successful read or writes
151 to/from the client. This eliminates timeouts that were occuring
152 during the large file transfers to/from slow clients.
154 *) Support generic wrappers such as cgiwrap by eliminating dependencies
155 on Apache's SUEXEC, renaming the FastCgiSuexec directive
156 FastCgiWrapper and eliminating any checks regarding the target
157 application (this is the repsonibility of the wrapper).
159 *) Fix a nasty bug that occurred when a client aborted a POST request
160 before the connection to a dynamic FastCGI application was opened.
161 The application's lock file descriptor wasn't setup but was being
162 closed which resulted in FD0 being closed. Normally this is open to
163 /dev/null and should pose no problems except that because the FD
164 was available it was being returned by Apache's accept(). This
165 caused it to be registered for pool cleanup. mod_cgi though moves
166 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
167 cleanup. Problem identified by checksum@163.net.
169 Changes with mod_fastcgi 2.2.8
171 *) Eliminate the concept of disabled applications. If a failure
172 occurs trying to setup an application (e.g. bind() error) its
173 tried repeatedly every init-start-delay seconds.
175 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
177 Changes with mod_fastcgi 2.2.6
179 *) Shutdown the PM when Apache appears to have disappeared.
181 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
183 *) (Win32) More dynamic fixes.
185 *) (Win32) Eliminate the per application named pipe mutex. Its now
186 per process to keep the lib happy (the use _FCGI_MUTEX_ should
187 removed from the lib). This allows pipe-based applications to
188 now handle multiple simoultanous requests (one per process).
190 *) Increase the number of open FDs we look to close when spawning apps.
192 *) Prevent an assert from popping unnecessarily.
194 *) (Win32) Add support for interpretter scripts.
196 *) (Win32) Fix named pipe handling (problems with large responses).
198 *) (Win32) Remove the "can exec" check.
200 Changes with mod_fastcgi 2.2.4
202 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
203 Rob Saccoccio [robs@chelsea.net]
205 *) Remove request type restriction (GET and POST only) in order to support
206 Web DAV requests. Scott Robertson [sroberts@codeit.com]
208 *) Allow requests for URLs such as /server/some/other/qualifier to match a
209 FastCGI server defined as /server. This was done primarily for Java
210 Servlets, but is generically useful.
212 *) Change the comm between the PM and the request handlers from a regular
213 file to a pipe and an assortment of other dynamic fixes.
215 *) Log dynamic process termination scheduling and the resulting
216 process exit notification.
218 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
219 application process from being killed off due to low demand.
221 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
223 *) Fix bugs that prevented dynamic processes from being shutdown when
224 the load subsided. Lars Heete [heete@do.isst.fhg.de]
226 *) Prevent dynamic processes from being scheduled to be started before
227 the init-start-delay or restart-delay period expires. This prevents
228 the queuing of process start requests while an application which has
229 a long initialization period starts up.
231 *) When suexec is enabled, allow the process manager to become root again in
232 order to signal applications it spawned.
234 *) Change the "server started" log message level from INFO to WARN.
236 *) Fix dynamic server "Connection Refused" handling.
238 *) Fix demand-based dynamic process spawing (uses a more portable approach).
240 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
241 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
242 longer than this period. It applies to the initiation of connections as
243 well (and thus is similar to appConnTimeout). Default is 30 seconds.
245 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
246 is more platform portable/predicable.
248 *) Leave STDOUT and STDERR open to the main server error_log. This should
249 eliminate a pile of protocol errors and having to track down 3rd party
250 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
251 be fixed in the application to use the FastCGI I/O, it just allows these
252 applications to run when previously they'd crash and burn.
254 *) Add the pass-header arg to server directives.
256 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
257 is done for CGI) - at least until PR2580 is resolved.
259 *) Initialize the default (empty) environment in a more socially acceptable
262 *) Miscellaneous doc improvements based on notes I've collected up over the
265 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
266 that resulted in not or improperly terminated strings.
268 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
269 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
271 Changes with mod_fastcgi 2.2.2
273 *) Added support for blocking connect()s by setting appConnTimeout to 0.
274 Non-blocking connect()s (the default) can be troublesome on some platforms.
275 Its expected that blocking connect()s will become the default (and
276 non-blocking connect()s will become optional) in the next release.
278 *) Dump a compile time error if the version of Apache is too old.
280 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
281 to Apache releases prior to 1.3.6.
283 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
287 Always restart a failed dynamic application if it is the last instance.
288 This means there once started there will always be at least one process
289 instance of a dynamic application.
291 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
292 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
293 the PM to start (another) application instance may help empty it faster.
295 Change two sleep() calls to select() based snoozes because alarm() is in
296 effect and sleep() and alarm() don't always play nice together.
298 Fixed a couple of error messages.
300 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
301 was ignored and the default was always used.
303 *) Allow the -initial-env argument to be used to pass variables from the
304 Apache process environment to the FastCGI server (by specifying a
305 variable name without the "=" or a value). [suggested by
306 Martin Lichtin <lichtin@oanda.com>]
308 *) Cleanup some debug macros.
310 *) Improved script stderr handling. [based on suggestions by
311 David Birnbaum <davidb@chelsea.net>]
313 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
314 updates to the INSTALL doc.
316 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
317 5 to 100. Its still configurable with the -listen-queue-depth option.
318 This should help eliminate the FAQ - Why do I see "Connection Refused"
321 *) Fix a bug in FastCgiExternalServer that broke support for external servers
322 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
324 Changes with mod_fastcgi 2.2.1
326 *) Updates to the INSTALL doc to describe building as a DSO.
328 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
329 Apache was running ("apachectl restart" does this), the contents of the
330 dynamic directory (dynamic sockets and the mbox) would be blown away
333 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
334 Apache header file) to prevent a compile error on systems without
337 *) During auth requests, the subprocess_env table was left holding variables
338 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
339 were passed to other processing phases (such as CGI/FastCGI). The
340 subprocess_env is now restored to its pre-auth condition.
342 *) Added a FastCGI Authorizer Role compatibility mode which implements the
343 specification to the tee. Use -compat arg with any of the Auth server
344 directives. This is intended for new and existing FastCGI authorizer
345 applications that require compatibility with other server implementations.
347 *) Fix logging from the process manager. Previously, all calls to the
348 logging routines used NULL for the server_rec. This works fine in a dev
349 environment, but results in most of these messages being tossed by
350 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
352 *) Always setup fr->header so that in the event the FastCGI server
353 doesn't get a valid FastCGI protocol header over the wire, we
354 can still print our error message to the log from do_work().
356 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
357 defined (Don Locrasto [locrasto@iceminer.com]).
359 Changes with mod_fastcgi 2.2.0
361 *) Serious rewrite of mod_fastcgi.html. An example conf will have
362 to wait until another time - I'm burned out on docs updates.
364 *) Add to the mod_fastcgi to the server version string.
366 *) SCO doesn't like the const in the arg to inet_addr() and
369 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
371 *) Deleted redundant header files from fcgi.h.
373 *) Full path names are no longer required for directives which take
374 file names as arguments. Like other Apache directives you can
375 specify paths from server_root (don't start with a "/").
377 *) The fcgiKillMgr is gone. This intermediate process sat between
378 the Apache parent and the FastCGI Process Manager under Apache
379 1.2. The process manager tries to name itself fcgi_pm (it used
382 *) Authentication, Authorization, and Access phases are now
383 supported. I've bent the FastCgi spec a bit: as many of the standard
384 environment variables (that were easy) are sent (the spec says don't
385 send some by name), all headers (except Status) sent by the auth
386 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
387 as environment variables rather than just those prefixed by
388 "Variable-". Custom responses for auth failures aren't supported
389 (yet.. we'll see what the demand for this is like). In addition to
390 the FastCGI protocol defined environment variable "FCGI_ROLE" being
391 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
392 set indicating which of the three phases is being processed. See
393 the docs for info on the new per-directory directives:
395 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
396 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
397 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
399 *) The code has been broken into logical chunks making figuring it
400 out much easier. All of the #defines a user may want to change are
401 now in mod_fastcgi.h.
403 *) All the logging has been converted to the "supported" Apache
404 logging routines, ap_log_error() and ap_log_rerror(). Log entries
405 for request specific errors should now be directed to the correct
406 server errorlog (if your running more than one). Some attempt has
407 been made at setting appropriate log levels (comments of course are
408 welcome) and printing errno info when its of value. All of log
409 messages now have "FastCGI" in them (nice for grep).
411 *) All of the calls to Apache routines have been updated to the ap_
412 convention, eliminating the need for the compat header.
414 *) By default we no longer flush every piece of data we get from the
415 FastCGI server to the client. This allows the FastCGI server to
416 release without having to wait for the client flush to complete.
417 The old behaviour can revived with the -flush argument to AppClass.
419 *) I renamed some of the directives for consistency (the old names
422 AppClass -> FastCgiServer,
423 FCGIConfig -> FastCgiConfig,
424 ExternalAppClass -> FastCgiExternalServer
426 *) Directive arguments are no longer case sensitive.
428 *) The module now uses hard_timeout() rather than soft_timeout. This
429 means that the module will longjump out of the request if the
430 drop-dead timeout expires (set with Apache's Timeout directive, the
431 default is 5min) or if the client closes the connection (SIGPIPE).
432 This is more typical for a Apache modules and I'm not convinced we
433 properly handle all of the error cases well enough to use
434 soft_timeout (I can think of one place we don't anyway). This means
435 your FastCGI application can see SIGPIPE.
437 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
438 Apache. Without this the default behaviour is exit().
440 *) It should (I think) run now under DSO now.
442 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
444 *) The broken -initial-env bug ins 2.1b1 is fixed.
446 *) Maybe some other stuff.
450 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
451 commit notes and stick 'em in here.. when I get some free time. ;)
456 *** Originally from docs/README..
458 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
459 From: Stanley Gambarin <stanleyg@cs.bu.edu>
461 *) Yet more changes to the source distribution. Separated out
462 the Tcl dynamic string and buffer libraries into separate files.
463 Did the same for the OS library, however, it is not really an
464 abstraction... more like a bunch of wrappers. Added to DEVNOTES
465 a note about the problem of Apache not allowing including header
468 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
469 From: Stanley Gambarin <stanleyg@cs.bu.edu>
471 *) More changes to the installation scripts. Theoretically we
472 should be able to to use the same Makefile/installation script
473 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
474 layer to use the same source code for both 1.2.x and 1.3.x sources.
476 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
477 From: Stanley Gambarin <stanleyg@cs.bu.edu>
479 *) Created Makefile to compile the module into the archive library,
480 just like the proxy module. Also initial draft of the installation
481 script, which should work for both 1.2 and 1.3 sources of Apache.
483 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
484 From: Stanley Gambarin <stanleyg@cs.bu.edu>
486 *) More source reorg. Separated out some header files. Need much
487 more work on this. Updated the TODO file. Added fcgivers.h and
488 CHANGES file to track the history.
490 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
491 From: Stanley Gambarin <stanleyg@cs.bu.edu>
493 *) Source reorganization. This is done to provide a basis for later
494 functionality implementation. See TODO file for more information
495 of what possible future enhancements are possible/desired.
497 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
498 From: David MacKenzie <djm@va.pubnix.com>
500 *) Create the "dynamic" dir and "mbox" as the user specified in the
501 User and Group directives instead of as root. They are created with
502 restrictive permissions, and then the module checks to see whether the
503 user specified in the User and Group directives has read, write, and
504 execute permissions on them. Those permissions had been explicitly
505 denied by creating them as "root".
507 *) Some documentation was garbled or incomplete.
509 *) When a FastCGI application can't be execl'd, the code used a
510 value of errno that may have been stomped on by intervening system
511 calls. That happened on BSD/OS 2.1, where the error_log was
512 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
513 An intervening stdio call was setting errno as a side-effect, which
514 it has the right to do.
516 *) Erroneous fprintf arguments, a missing return value, an unused
517 function, missing declarations, and other problems detected by gcc
520 *) Duplicated code to create the lock file name merged into a single
521 function, closing several memory leaks.
523 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
524 FCGI app in a given dynamic class would live forever, even if it
525 hadn't been requested for weeks.
527 *) The test for whether to keep looking for dynamic app victims to
528 kill was backwards, so no dynamic apps were ever selected as
531 *) The killInterval and updateInterval were being ignored; the
532 sigsuspend() forced recalculations only when a child died,
533 instead of at the intervals specified.
535 *) When a dynamic app couldn't be started, the program SEGV'd when
536 trying to free the ipcAddrPtr twice.
538 *) Remove the dead lock file and socket when cleaning up after a
539 server whose last child has been killed, so the FastCGIHandler
540 isn't fooled into thinking there is still a process serving
543 *) Make file locking robust in the presence of signals.
545 *) Rename some badly named variables.
547 *) Fix typos in many comments.
549 *) Fix some memory and file descriptor leaks.
551 *) Make the blocking kill of a server closer to working.
553 *) More fixes to calculations in dynamic application management.
556 *** Originally from docs/README.OMI..
558 What's New: Version 2.0b1, 16 Apr 1997
560 *) Implemented a mechanism by which FastCGI applications are started by
561 the web server on the first request and continue running. Also, a
562 heuristics have been implemented to allow for dynamic killing of the
563 running FastCGI apps. The configuration options are supplied via
564 the new FCGIConfig directive, that is described in the
567 *) When performing internal redirect, since the original request's body
568 has already been read, do not allow the redirected request to think
569 that is has one [body].
571 *) More conditional compilation for OS/2.
573 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
575 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
577 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
578 force some output to be written in the period of inactivity. This
579 circumvents the Apache's buffering during the server-push.
581 *) Added a header-parser function placeholder for compatibility with an
584 *) Additional checks have been placed with regard to permissions of the
585 FastCGI processes. These include the existence of the file,
586 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
588 *) Implemented a restart cleanup, so that no parentlesses FastCGI
589 applications are left after Apache is restarted and/or terminated.
590 This was accomplished via two independent mechanisms, where the
591 first one prevents a start of the process manager on the first
592 reading of the configuration files. The second mechanism deals away
593 from Apache's implementation of fork() (via spawn_child()) and
594 implements its own forking, thus removing any dependencies on the
595 Apache to cleanup processes during termination phase.
597 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
598 itself in the file conf.h
600 What's New: Version 1.4.3, 15 Jan 1997
602 *) Fixed compilation warnings for various platforms, as well as
603 conditional compilation for OS/2.
605 What's New: Version 1.4.2, 12 Dec 1996
607 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
608 will proceed under this version of Apache web server.
610 *) As the result of porting, the "include virtual" construct of SSI
611 will now work correctly using either <Location> or <Directory>
614 What's New: Version 1.4.1, 4 Dec 1996
616 *) Checks have been removed from the ScanCGIHeaders that provided
617 for the presence of both Status and Location headers as being
618 an error. Contradictory to CGI/1.1 Internet Draft, both of
619 these headers are used by the current CGI applications.
621 What's New: Version 1.4, 22 Nov 1996
623 *) Added the -port option to AppClass, allowing TCP/IP communication.
624 Added the -socket option to AppClass, allowing Unix domain
625 communication via a configurable pathname.
627 *) Added the ExternalAppClass directive, allowing TCP/IP
628 communication with remote FastCGI applications.
630 *) The handler had its own code for generating HTTP response headers;
631 now it uses Apache's. This reduces the size of the module.
632 More importantly, it fixes the bug in which "include virtual"
633 sees the HTTP response headers.
635 *) The response header parser performed very little checking.
636 Now the parser enforces the guidelines in the CGI/1.1
637 Internet-Draft: Status and Location are mutually exclusive,
638 Location can only be a response to GET or HEAD,
639 CGI response headers can't be repeated, etc. (The CGI response
640 headers are the ones the handler interprets: Status, Location,
643 *) The response header parser used to miss CGI headers
644 with no whitespace after the colon, e.g. "Status:200 OK".
646 *) The response header parser sometimes interpreted the first line
647 of content as an RFC822 continuation line.
649 *) The handler implemented a nonstandard version of Location
650 which never used internal redirects. The handler
651 now attempts to implement Location as specified in the
652 CGI/1.1 Internet-Draft. The module documentation explains
655 *) Error log entries from the response header parser were
656 pretty uninformative; they are better now. If a header is
657 malformed, the log entry includes it. If the headers are
658 unterminated rather than malformed, the log entry says that,
659 and says how many bytes were received from the app.
661 *) When the application manager forked a new process, and that
662 process ran into trouble before executing the first instruction
663 of application code, the process used to exit with status =
664 errno. This made certain configuration problems (e.g. incorrect
665 file permissions when server parent is root) quite difficult
666 to pin down. Now the failing child process opens up the error log
667 and writes an informative entry before exiting.
669 *) The module now correctly handles a slash at the end of the
670 DocumentRoot directive. This was a one-line fix.
672 What's New: Version 1.3.3, 17 Oct 1996
674 *) The module now registers its request handler under the name
675 fastcgi-script in addition to the name application/x-httpd-fcgi.
676 This was a one-line addition to the module, but had quite
677 a large impact on the documentation and sample configuration.
679 What's New: Version 1.3.2, 27 Sept 1996
681 *) On some systems (SunOS, Linux?), fopen for append has a bug
682 that strikes when two processes append to the same file.
683 This bug causes the process manager to corrupt the error log.
684 Work around the bug by calling open, then fdopen. A patch
685 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
686 (Reported by Bob Ramstad.)
688 What's New: Version 1.3.1, 17 Sept 1996
690 *) On Linux and SunOS, the Apache default user_id (-1) is not a
691 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
692 Scott Langley, others.)
694 *) On some systems (Linux, some Solaris?, Irix?), connect requires
695 write access to a Unix Domain socket, so provide it. (Reported
696 by Scott Langley, freeform@wired.com, others.)
698 *) If you hit a .fcg file, but there's no AppClass defined, the
699 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
702 What's New: Version 1.3, 4 Sept 1996
704 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
706 *) Module violated the Apache buff abstraction.
707 Fixed. Might now work with SSL (not tested).
708 As part of the fix, eliminated support for
709 Apache 1.0x versions.
711 *) Module failed to chmod the Unix domain listening sockets
712 it created, so protections were set according to the current
715 *) Module forked too many process manager processes: One
716 per AppClass. When the Apache parent ran as root, these
717 process manager processes ran as root. New process
718 manager is a single process (only started if AppClass is
719 used) and runs with same privileges as other children.
720 The new process manager doesn't do any polling, so
721 there's less system overhead than before.
723 *) AppClass insisted on getting at least two arguments. Fixed.
725 *) Module used setjmp/longjmp, causing compiler warnings
726 on some platforms (e.g. Linux.) The module no longer
729 *) Module created listening sockets in /tmp, where they were
730 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
731 directive to give control over the location of listening
734 *) Module wrote error log entries without a timestamp. Fixed.
736 *) AppClass directive wrote error messages to stderr in addition
737 to returning a char * message to the Apache core. Fixed.
739 What's New: Version 1.2, 3 June 1996
741 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
744 Add version string: APACHEVERSION macro to mod_fastcgi.c
746 *) chown FastCGI socket to user_id and setuid to user_id for app
749 *) Modify GetFromStream() by having it call an OS dependent
750 function GetStreamSize(FILE *) which uses FILE internal data member.
751 Linux users might need to modify GetStreamSize().
753 What's New: Version 1.1, 10 May 1996
755 *) If you specify a non-existent executable in the AppClass directive,
756 or if the file exists and it does not have execute permission you
757 get a constant stream of error messages telling you that "program
758 terminated due to a signal".
760 *) The mod_fastcgi module should use the standard Apache error logging
761 facility instead of writing to stderr.
763 Version 1.0, 30 April 1996