3 *) Prevent the use of all but the "auth" directives from being used
4 anywhere but in global scope. Prevent more than one instance
5 of the FastCgiWrapper directive.
7 *) Return NOT_FOUND (404) or FORBIDDEN (403) instead of
8 INTERNAL_SERVER_ERROR (500) when there are configuration issues
9 or the script does't exist. Suggested by Jeff Lawson [bovine@ud.com].
13 *) [*nix Security] - When FastCgiWrapper (FastCgiSuexec) was in use
14 and a vhost configured to use the same uid/gid as the main
15 server, mod_fastcgi would not bother using the wrapper (suexec)
16 because its effective uid/gid was already appropriate. This is
17 consistent with Apache's v1.3 mod_cgi behaviour. There are two
18 problems with this approach: 1) when FastCgWrapper is in use
19 mod_fastcgi's process manager keeps its root privileges (as its
20 real uid/gid) so it can terminate the applications its starts -
21 this privilege was being passed to applications when the use of
22 the wrapper was bypassed 2) wrappers are often employed to
23 perform functionality beyond setting the uid/gid - by not calling
24 the wrapper under certain circumstances, application invocation
25 environments were inconsistent. With this change, the wrapper is
26 always used (when enabled) under both Apache 1.3 and 2.
27 Reported by Michael Richards [michael@fastmail.ca].
29 *) [*nix/AP2] Use the vhost uid/gid instead of the server uid/gid
30 for dynamic application invocation when the FastCgiWrapper is in use.
31 Reported by Michael Richards [michael@fastmail.ca]
33 *) [*nix] Fix handling of FastCgiWrapper when passed a real path,
34 i.e. other than "on" or "off". Michael Richards [michael@fastmail.ca]
36 *) Eliminate the logging of "incomplete headers (0 bytes) received from
37 server" when a client aborts.
39 *) [WIN32] Fix a delay in handling large POSTs to named pipe based
40 servers. Philip Gladstone [philip@okena.com]
42 *) [*nix/AP2] Prevent the module from being initalized twice at startup
43 (resulting in confusing error messages to the log).
45 *) Eliminate the need for SetHandler or AddHandler with static or
46 external applications.
48 *) Limit PM requests to start a dynamic application
49 to 5sec to prevent endless spinning (this is a drop-dead
50 limit that should only occur if the socket/named_pipe directory
51 is removed out from under a running server).
53 *) [*nix] Change the default socket directory from /tmp/fcgi to:
55 Apache2 - RUNTIMEDIR/fastcgi
57 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
58 for use with wrappers (in lieu of finding the user/group associated
59 with a virtual host - under Apache2 this isn't accomodated).
61 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
62 apr_proc_create() changes.
64 *) Log when invoking and restoring the restart backoff policy.
66 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
67 OVERLAPPED structure).
69 *) Eliminate need for dummy files for external servers under Apache2
71 *) Fix auth compatibility mode handling for access checker and authorizer
73 *) Fix HEAD request handling. Based on a patch by
74 Chris Lightfoot [chris@ex-parrot.com]
76 *) [*nix] When autoupdate is enabled touch the socket when restarting
77 the processes to prevent further requests.
78 Eckebrecht von Pappenheim [evp@heise.de]
80 *) Apache 2.0 support.
82 *) [WIN32] Don't read from a potentially closed named pipe.
83 Philip Gladstone [philip@okena.com]
85 *) Require the Apache version 1.3.6 or later to eliminate some signal
88 *) [WIN32] Use asyncronous io with named pipes instead of polled
89 nonblocking io. This should eliminate the last of the npipe issues.
91 *) Handle an application returning a complete and valid response without
92 having consumed all of the data sent to it.
94 *) Consume remaining client data (RESPONDERs only) if any.
96 *) Add support for backing off attempts to start applications that continuously
97 fail to start. Three new macros defined in mod_fastcgi.h control this
98 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
100 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
101 applications that do not (properly) support the shutdown event (this
102 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
103 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
104 interval between signaling a proper shutdown and wacking the process(s)
105 with a TerminateProcess().
107 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
108 setting it was just plain broken.
110 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
112 *) Fix 'FastCgiConfig -autoUpdate'.
114 *) Fix 'FastCgiConfig -flush'.
116 *) Prevent silly maxProcesses and processSlack combinations.
117 Dmitry Dorofeev [dima@yasp.com]
119 *) Properly handle the killing of idle processes when one takes a long time
120 to exit once signaled down (or the config is funky).
121 Dmitry Dorofeev [dima@yasp.com]
123 *) Always kill the youngest instance of an application. Suggested by
124 Dmitry Dorofeev [dima@yasp.com]
128 *) Delay the logging of write errors to the pm to account for shutdown/restart.
130 *) (Win32) An assortment of fixes.
132 *) Fix some broken casts that were likely the cause of an assert.
134 *) Win32. Eliminate forward slashes from the named pipe path name.
135 Gerald Richter [richter@ecos.de]
137 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
138 applications it spawns. [ryans@amazon.com]
140 *) Added support for the -flush argument to FastCgiConfig.
141 Eric Sit [esit@alum.mit.edu]
143 *) Change the "which call to module_init() is this" check to a more
144 reliable approach. Doru Petrescu [pdoru@kappa.ro]
146 *) Close the old pipe file descriptor in apache main on USR1/HUP
147 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
149 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
150 [tetsuya@secom-sis.co.jp]
152 *) Eliminate the use of locks to assist in the clean shutdown of
153 applications. Instead, it is assumed that applications handle
154 termination signals properly (this is now embedded in the C
157 *) Fix Win32 process termination. Proper operation requires the use of
158 an updated application lib (termination is now signalled with an
159 Event and handled by specialized thread).
161 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
163 *) Added code so if the last instance of a dynamic application died without
164 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
165 configuration allows the last instance to be killed, then allow it to die).
166 Andrew Benham [adsb@bigfoot.com]
168 *) Fix the loadFactor calculation used to determine when dyanmic
169 applications could be killed off due to low demand [adsb@bigfoot.com].
171 *) Fix a deadlock condition that could occur with Win32 named
174 *) Fix a potential deadlock condition when FastCGI application
175 sent responses while still reading the client request (POST data).
179 *) Allow absolute pathnames in the -socket argument. Suggested by
180 Christian Jaeger [christian.jaeger@sl.ethz.ch].
182 *) Don't invoke suexec when the user/group for the fastcgi application
183 is the same as the apache main server. This is consistent with
184 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
186 *) Reset the apache drop dead timer upon successful read or writes
187 to/from the client. This eliminates timeouts that were occuring
188 during the large file transfers to/from slow clients.
190 *) Support generic wrappers such as cgiwrap by eliminating dependencies
191 on Apache's SUEXEC, renaming the FastCgiSuexec directive
192 FastCgiWrapper and eliminating any checks regarding the target
193 application (this is the repsonibility of the wrapper).
195 *) Fix a nasty bug that occurred when a client aborted a POST request
196 before the connection to a dynamic FastCGI application was opened.
197 The application's lock file descriptor wasn't setup but was being
198 closed which resulted in FD0 being closed. Normally this is open to
199 /dev/null and should pose no problems except that because the FD
200 was available it was being returned by Apache's accept(). This
201 caused it to be registered for pool cleanup. mod_cgi though moves
202 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
203 cleanup. Problem identified by checksum@163.net.
205 Changes with mod_fastcgi 2.2.8
207 *) Eliminate the concept of disabled applications. If a failure
208 occurs trying to setup an application (e.g. bind() error) its
209 tried repeatedly every init-start-delay seconds.
211 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
213 Changes with mod_fastcgi 2.2.6
215 *) Shutdown the PM when Apache appears to have disappeared.
217 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
219 *) (Win32) More dynamic fixes.
221 *) (Win32) Eliminate the per application named pipe mutex. Its now
222 per process to keep the lib happy (the use _FCGI_MUTEX_ should
223 removed from the lib). This allows pipe-based applications to
224 now handle multiple simoultanous requests (one per process).
226 *) Increase the number of open FDs we look to close when spawning apps.
228 *) Prevent an assert from popping unnecessarily.
230 *) (Win32) Add support for interpretter scripts.
232 *) (Win32) Fix named pipe handling (problems with large responses).
234 *) (Win32) Remove the "can exec" check.
236 Changes with mod_fastcgi 2.2.4
238 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
239 Rob Saccoccio [robs@chelsea.net]
241 *) Remove request type restriction (GET and POST only) in order to support
242 Web DAV requests. Scott Robertson [sroberts@codeit.com]
244 *) Allow requests for URLs such as /server/some/other/qualifier to match a
245 FastCGI server defined as /server. This was done primarily for Java
246 Servlets, but is generically useful.
248 *) Change the comm between the PM and the request handlers from a regular
249 file to a pipe and an assortment of other dynamic fixes.
251 *) Log dynamic process termination scheduling and the resulting
252 process exit notification.
254 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
255 application process from being killed off due to low demand.
257 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
259 *) Fix bugs that prevented dynamic processes from being shutdown when
260 the load subsided. Lars Heete [heete@do.isst.fhg.de]
262 *) Prevent dynamic processes from being scheduled to be started before
263 the init-start-delay or restart-delay period expires. This prevents
264 the queuing of process start requests while an application which has
265 a long initialization period starts up.
267 *) When suexec is enabled, allow the process manager to become root again in
268 order to signal applications it spawned.
270 *) Change the "server started" log message level from INFO to WARN.
272 *) Fix dynamic server "Connection Refused" handling.
274 *) Fix demand-based dynamic process spawing (uses a more portable approach).
276 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
277 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
278 longer than this period. It applies to the initiation of connections as
279 well (and thus is similar to appConnTimeout). Default is 30 seconds.
281 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
282 is more platform portable/predicable.
284 *) Leave STDOUT and STDERR open to the main server error_log. This should
285 eliminate a pile of protocol errors and having to track down 3rd party
286 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
287 be fixed in the application to use the FastCGI I/O, it just allows these
288 applications to run when previously they'd crash and burn.
290 *) Add the pass-header arg to server directives.
292 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
293 is done for CGI) - at least until PR2580 is resolved.
295 *) Initialize the default (empty) environment in a more socially acceptable
298 *) Miscellaneous doc improvements based on notes I've collected up over the
301 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
302 that resulted in not or improperly terminated strings.
304 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
305 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
307 Changes with mod_fastcgi 2.2.2
309 *) Added support for blocking connect()s by setting appConnTimeout to 0.
310 Non-blocking connect()s (the default) can be troublesome on some platforms.
311 Its expected that blocking connect()s will become the default (and
312 non-blocking connect()s will become optional) in the next release.
314 *) Dump a compile time error if the version of Apache is too old.
316 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
317 to Apache releases prior to 1.3.6.
319 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
323 Always restart a failed dynamic application if it is the last instance.
324 This means there once started there will always be at least one process
325 instance of a dynamic application.
327 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
328 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
329 the PM to start (another) application instance may help empty it faster.
331 Change two sleep() calls to select() based snoozes because alarm() is in
332 effect and sleep() and alarm() don't always play nice together.
334 Fixed a couple of error messages.
336 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
337 was ignored and the default was always used.
339 *) Allow the -initial-env argument to be used to pass variables from the
340 Apache process environment to the FastCGI server (by specifying a
341 variable name without the "=" or a value). [suggested by
342 Martin Lichtin <lichtin@oanda.com>]
344 *) Cleanup some debug macros.
346 *) Improved script stderr handling. [based on suggestions by
347 David Birnbaum <davidb@chelsea.net>]
349 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
350 updates to the INSTALL doc.
352 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
353 5 to 100. Its still configurable with the -listen-queue-depth option.
354 This should help eliminate the FAQ - Why do I see "Connection Refused"
357 *) Fix a bug in FastCgiExternalServer that broke support for external servers
358 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
360 Changes with mod_fastcgi 2.2.1
362 *) Updates to the INSTALL doc to describe building as a DSO.
364 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
365 Apache was running ("apachectl restart" does this), the contents of the
366 dynamic directory (dynamic sockets and the mbox) would be blown away
369 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
370 Apache header file) to prevent a compile error on systems without
373 *) During auth requests, the subprocess_env table was left holding variables
374 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
375 were passed to other processing phases (such as CGI/FastCGI). The
376 subprocess_env is now restored to its pre-auth condition.
378 *) Added a FastCGI Authorizer Role compatibility mode which implements the
379 specification to the tee. Use -compat arg with any of the Auth server
380 directives. This is intended for new and existing FastCGI authorizer
381 applications that require compatibility with other server implementations.
383 *) Fix logging from the process manager. Previously, all calls to the
384 logging routines used NULL for the server_rec. This works fine in a dev
385 environment, but results in most of these messages being tossed by
386 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
388 *) Always setup fr->header so that in the event the FastCGI server
389 doesn't get a valid FastCGI protocol header over the wire, we
390 can still print our error message to the log from do_work().
392 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
393 defined (Don Locrasto [locrasto@iceminer.com]).
395 Changes with mod_fastcgi 2.2.0
397 *) Serious rewrite of mod_fastcgi.html. An example conf will have
398 to wait until another time - I'm burned out on docs updates.
400 *) Add to the mod_fastcgi to the server version string.
402 *) SCO doesn't like the const in the arg to inet_addr() and
405 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
407 *) Deleted redundant header files from fcgi.h.
409 *) Full path names are no longer required for directives which take
410 file names as arguments. Like other Apache directives you can
411 specify paths from server_root (don't start with a "/").
413 *) The fcgiKillMgr is gone. This intermediate process sat between
414 the Apache parent and the FastCGI Process Manager under Apache
415 1.2. The process manager tries to name itself fcgi_pm (it used
418 *) Authentication, Authorization, and Access phases are now
419 supported. I've bent the FastCgi spec a bit: as many of the standard
420 environment variables (that were easy) are sent (the spec says don't
421 send some by name), all headers (except Status) sent by the auth
422 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
423 as environment variables rather than just those prefixed by
424 "Variable-". Custom responses for auth failures aren't supported
425 (yet.. we'll see what the demand for this is like). In addition to
426 the FastCGI protocol defined environment variable "FCGI_ROLE" being
427 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
428 set indicating which of the three phases is being processed. See
429 the docs for info on the new per-directory directives:
431 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
432 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
433 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
435 *) The code has been broken into logical chunks making figuring it
436 out much easier. All of the #defines a user may want to change are
437 now in mod_fastcgi.h.
439 *) All the logging has been converted to the "supported" Apache
440 logging routines, ap_log_error() and ap_log_rerror(). Log entries
441 for request specific errors should now be directed to the correct
442 server errorlog (if your running more than one). Some attempt has
443 been made at setting appropriate log levels (comments of course are
444 welcome) and printing errno info when its of value. All of log
445 messages now have "FastCGI" in them (nice for grep).
447 *) All of the calls to Apache routines have been updated to the ap_
448 convention, eliminating the need for the compat header.
450 *) By default we no longer flush every piece of data we get from the
451 FastCGI server to the client. This allows the FastCGI server to
452 release without having to wait for the client flush to complete.
453 The old behaviour can revived with the -flush argument to AppClass.
455 *) I renamed some of the directives for consistency (the old names
458 AppClass -> FastCgiServer,
459 FCGIConfig -> FastCgiConfig,
460 ExternalAppClass -> FastCgiExternalServer
462 *) Directive arguments are no longer case sensitive.
464 *) The module now uses hard_timeout() rather than soft_timeout. This
465 means that the module will longjump out of the request if the
466 drop-dead timeout expires (set with Apache's Timeout directive, the
467 default is 5min) or if the client closes the connection (SIGPIPE).
468 This is more typical for a Apache modules and I'm not convinced we
469 properly handle all of the error cases well enough to use
470 soft_timeout (I can think of one place we don't anyway). This means
471 your FastCGI application can see SIGPIPE.
473 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
474 Apache. Without this the default behaviour is exit().
476 *) It should (I think) run now under DSO now.
478 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
480 *) The broken -initial-env bug ins 2.1b1 is fixed.
482 *) Maybe some other stuff.
486 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
487 commit notes and stick 'em in here.. when I get some free time. ;)
492 *** Originally from docs/README..
494 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
495 From: Stanley Gambarin <stanleyg@cs.bu.edu>
497 *) Yet more changes to the source distribution. Separated out
498 the Tcl dynamic string and buffer libraries into separate files.
499 Did the same for the OS library, however, it is not really an
500 abstraction... more like a bunch of wrappers. Added to DEVNOTES
501 a note about the problem of Apache not allowing including header
504 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
505 From: Stanley Gambarin <stanleyg@cs.bu.edu>
507 *) More changes to the installation scripts. Theoretically we
508 should be able to to use the same Makefile/installation script
509 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
510 layer to use the same source code for both 1.2.x and 1.3.x sources.
512 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
513 From: Stanley Gambarin <stanleyg@cs.bu.edu>
515 *) Created Makefile to compile the module into the archive library,
516 just like the proxy module. Also initial draft of the installation
517 script, which should work for both 1.2 and 1.3 sources of Apache.
519 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
520 From: Stanley Gambarin <stanleyg@cs.bu.edu>
522 *) More source reorg. Separated out some header files. Need much
523 more work on this. Updated the TODO file. Added fcgivers.h and
524 CHANGES file to track the history.
526 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
527 From: Stanley Gambarin <stanleyg@cs.bu.edu>
529 *) Source reorganization. This is done to provide a basis for later
530 functionality implementation. See TODO file for more information
531 of what possible future enhancements are possible/desired.
533 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
534 From: David MacKenzie <djm@va.pubnix.com>
536 *) Create the "dynamic" dir and "mbox" as the user specified in the
537 User and Group directives instead of as root. They are created with
538 restrictive permissions, and then the module checks to see whether the
539 user specified in the User and Group directives has read, write, and
540 execute permissions on them. Those permissions had been explicitly
541 denied by creating them as "root".
543 *) Some documentation was garbled or incomplete.
545 *) When a FastCGI application can't be execl'd, the code used a
546 value of errno that may have been stomped on by intervening system
547 calls. That happened on BSD/OS 2.1, where the error_log was
548 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
549 An intervening stdio call was setting errno as a side-effect, which
550 it has the right to do.
552 *) Erroneous fprintf arguments, a missing return value, an unused
553 function, missing declarations, and other problems detected by gcc
556 *) Duplicated code to create the lock file name merged into a single
557 function, closing several memory leaks.
559 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
560 FCGI app in a given dynamic class would live forever, even if it
561 hadn't been requested for weeks.
563 *) The test for whether to keep looking for dynamic app victims to
564 kill was backwards, so no dynamic apps were ever selected as
567 *) The killInterval and updateInterval were being ignored; the
568 sigsuspend() forced recalculations only when a child died,
569 instead of at the intervals specified.
571 *) When a dynamic app couldn't be started, the program SEGV'd when
572 trying to free the ipcAddrPtr twice.
574 *) Remove the dead lock file and socket when cleaning up after a
575 server whose last child has been killed, so the FastCGIHandler
576 isn't fooled into thinking there is still a process serving
579 *) Make file locking robust in the presence of signals.
581 *) Rename some badly named variables.
583 *) Fix typos in many comments.
585 *) Fix some memory and file descriptor leaks.
587 *) Make the blocking kill of a server closer to working.
589 *) More fixes to calculations in dynamic application management.
592 *** Originally from docs/README.OMI..
594 What's New: Version 2.0b1, 16 Apr 1997
596 *) Implemented a mechanism by which FastCGI applications are started by
597 the web server on the first request and continue running. Also, a
598 heuristics have been implemented to allow for dynamic killing of the
599 running FastCGI apps. The configuration options are supplied via
600 the new FCGIConfig directive, that is described in the
603 *) When performing internal redirect, since the original request's body
604 has already been read, do not allow the redirected request to think
605 that is has one [body].
607 *) More conditional compilation for OS/2.
609 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
611 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
613 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
614 force some output to be written in the period of inactivity. This
615 circumvents the Apache's buffering during the server-push.
617 *) Added a header-parser function placeholder for compatibility with an
620 *) Additional checks have been placed with regard to permissions of the
621 FastCGI processes. These include the existence of the file,
622 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
624 *) Implemented a restart cleanup, so that no parentlesses FastCGI
625 applications are left after Apache is restarted and/or terminated.
626 This was accomplished via two independent mechanisms, where the
627 first one prevents a start of the process manager on the first
628 reading of the configuration files. The second mechanism deals away
629 from Apache's implementation of fork() (via spawn_child()) and
630 implements its own forking, thus removing any dependencies on the
631 Apache to cleanup processes during termination phase.
633 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
634 itself in the file conf.h
636 What's New: Version 1.4.3, 15 Jan 1997
638 *) Fixed compilation warnings for various platforms, as well as
639 conditional compilation for OS/2.
641 What's New: Version 1.4.2, 12 Dec 1996
643 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
644 will proceed under this version of Apache web server.
646 *) As the result of porting, the "include virtual" construct of SSI
647 will now work correctly using either <Location> or <Directory>
650 What's New: Version 1.4.1, 4 Dec 1996
652 *) Checks have been removed from the ScanCGIHeaders that provided
653 for the presence of both Status and Location headers as being
654 an error. Contradictory to CGI/1.1 Internet Draft, both of
655 these headers are used by the current CGI applications.
657 What's New: Version 1.4, 22 Nov 1996
659 *) Added the -port option to AppClass, allowing TCP/IP communication.
660 Added the -socket option to AppClass, allowing Unix domain
661 communication via a configurable pathname.
663 *) Added the ExternalAppClass directive, allowing TCP/IP
664 communication with remote FastCGI applications.
666 *) The handler had its own code for generating HTTP response headers;
667 now it uses Apache's. This reduces the size of the module.
668 More importantly, it fixes the bug in which "include virtual"
669 sees the HTTP response headers.
671 *) The response header parser performed very little checking.
672 Now the parser enforces the guidelines in the CGI/1.1
673 Internet-Draft: Status and Location are mutually exclusive,
674 Location can only be a response to GET or HEAD,
675 CGI response headers can't be repeated, etc. (The CGI response
676 headers are the ones the handler interprets: Status, Location,
679 *) The response header parser used to miss CGI headers
680 with no whitespace after the colon, e.g. "Status:200 OK".
682 *) The response header parser sometimes interpreted the first line
683 of content as an RFC822 continuation line.
685 *) The handler implemented a nonstandard version of Location
686 which never used internal redirects. The handler
687 now attempts to implement Location as specified in the
688 CGI/1.1 Internet-Draft. The module documentation explains
691 *) Error log entries from the response header parser were
692 pretty uninformative; they are better now. If a header is
693 malformed, the log entry includes it. If the headers are
694 unterminated rather than malformed, the log entry says that,
695 and says how many bytes were received from the app.
697 *) When the application manager forked a new process, and that
698 process ran into trouble before executing the first instruction
699 of application code, the process used to exit with status =
700 errno. This made certain configuration problems (e.g. incorrect
701 file permissions when server parent is root) quite difficult
702 to pin down. Now the failing child process opens up the error log
703 and writes an informative entry before exiting.
705 *) The module now correctly handles a slash at the end of the
706 DocumentRoot directive. This was a one-line fix.
708 What's New: Version 1.3.3, 17 Oct 1996
710 *) The module now registers its request handler under the name
711 fastcgi-script in addition to the name application/x-httpd-fcgi.
712 This was a one-line addition to the module, but had quite
713 a large impact on the documentation and sample configuration.
715 What's New: Version 1.3.2, 27 Sept 1996
717 *) On some systems (SunOS, Linux?), fopen for append has a bug
718 that strikes when two processes append to the same file.
719 This bug causes the process manager to corrupt the error log.
720 Work around the bug by calling open, then fdopen. A patch
721 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
722 (Reported by Bob Ramstad.)
724 What's New: Version 1.3.1, 17 Sept 1996
726 *) On Linux and SunOS, the Apache default user_id (-1) is not a
727 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
728 Scott Langley, others.)
730 *) On some systems (Linux, some Solaris?, Irix?), connect requires
731 write access to a Unix Domain socket, so provide it. (Reported
732 by Scott Langley, freeform@wired.com, others.)
734 *) If you hit a .fcg file, but there's no AppClass defined, the
735 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
738 What's New: Version 1.3, 4 Sept 1996
740 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
742 *) Module violated the Apache buff abstraction.
743 Fixed. Might now work with SSL (not tested).
744 As part of the fix, eliminated support for
745 Apache 1.0x versions.
747 *) Module failed to chmod the Unix domain listening sockets
748 it created, so protections were set according to the current
751 *) Module forked too many process manager processes: One
752 per AppClass. When the Apache parent ran as root, these
753 process manager processes ran as root. New process
754 manager is a single process (only started if AppClass is
755 used) and runs with same privileges as other children.
756 The new process manager doesn't do any polling, so
757 there's less system overhead than before.
759 *) AppClass insisted on getting at least two arguments. Fixed.
761 *) Module used setjmp/longjmp, causing compiler warnings
762 on some platforms (e.g. Linux.) The module no longer
765 *) Module created listening sockets in /tmp, where they were
766 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
767 directive to give control over the location of listening
770 *) Module wrote error log entries without a timestamp. Fixed.
772 *) AppClass directive wrote error messages to stderr in addition
773 to returning a char * message to the Apache core. Fixed.
775 What's New: Version 1.2, 3 June 1996
777 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
780 Add version string: APACHEVERSION macro to mod_fastcgi.c
782 *) chown FastCGI socket to user_id and setuid to user_id for app
785 *) Modify GetFromStream() by having it call an OS dependent
786 function GetStreamSize(FILE *) which uses FILE internal data member.
787 Linux users might need to modify GetStreamSize().
789 What's New: Version 1.1, 10 May 1996
791 *) If you specify a non-existent executable in the AppClass directive,
792 or if the file exists and it does not have execute permission you
793 get a constant stream of error messages telling you that "program
794 terminated due to a signal".
796 *) The mod_fastcgi module should use the standard Apache error logging
797 facility instead of writing to stderr.
799 Version 1.0, 30 April 1996