3 *) [WIN] Fix handle leaks in the process manager.
4 Sascha Schumann [sascha@schumann.cx]
6 *) [WIN] Use a permanent pool for allocating the SystemRoot
7 environment variable. Sakamoto [ringring@zb4.so-net.ne.jp]
9 *) [WIN] Fix starting of scripts under AP2.
11 *) Do the connect() to the application *after* collecting a chunk
12 of client data. This reinstates the pre-2.4.0 behaviour.
13 James Jurach [muaddib@erf.net].
15 *) Remove an assert that was triggering on WIN when spawn() failed.
17 *) Provide the NO_SUEXEC_FOR_AP_USER_N_GROUP macro for building
18 mod_fastcgi with the AP13 suexec behaviour (don't use suexec
19 if httpd's user and group match that needed for the application).
21 *) Prevent the use of all but the "auth" directives from being used
22 anywhere but in global scope. Prevent more than one instance
23 of the FastCgiWrapper directive.
25 *) Return NOT_FOUND (404) or FORBIDDEN (403) instead of
26 INTERNAL_SERVER_ERROR (500) when there are configuration issues
27 or the script does't exist. Suggested by Jeff Lawson [bovine@ud.com].
31 *) [*nix Security] - When FastCgiWrapper (FastCgiSuexec) was in use
32 and a vhost configured to use the same uid/gid as the main
33 server, mod_fastcgi would not bother using the wrapper (suexec)
34 because its effective uid/gid was already appropriate. This is
35 consistent with Apache's v1.3 mod_cgi behaviour. There are two
36 problems with this approach: 1) when FastCgWrapper is in use
37 mod_fastcgi's process manager keeps its root privileges (as its
38 real uid/gid) so it can terminate the applications its starts -
39 this privilege was being passed to applications when the use of
40 the wrapper was bypassed 2) wrappers are often employed to
41 perform functionality beyond setting the uid/gid - by not calling
42 the wrapper under certain circumstances, application invocation
43 environments were inconsistent. With this change, the wrapper is
44 always used (when enabled) under both Apache 1.3 and 2.
45 Reported by Michael Richards [michael@fastmail.ca].
47 *) [*nix/AP2] Use the vhost uid/gid instead of the server uid/gid
48 for dynamic application invocation when the FastCgiWrapper is in use.
49 Reported by Michael Richards [michael@fastmail.ca]
51 *) [*nix] Fix handling of FastCgiWrapper when passed a real path,
52 i.e. other than "on" or "off". Michael Richards [michael@fastmail.ca]
54 *) Eliminate the logging of "incomplete headers (0 bytes) received from
55 server" when a client aborts.
57 *) [WIN32] Fix a delay in handling large POSTs to named pipe based
58 servers. Philip Gladstone [philip@okena.com]
60 *) [*nix/AP2] Prevent the module from being initalized twice at startup
61 (resulting in confusing error messages to the log).
63 *) Eliminate the need for SetHandler or AddHandler with static or
64 external applications.
66 *) Limit PM requests to start a dynamic application
67 to 5sec to prevent endless spinning (this is a drop-dead
68 limit that should only occur if the socket/named_pipe directory
69 is removed out from under a running server).
71 *) [*nix] Change the default socket directory from /tmp/fcgi to:
73 Apache2 - RUNTIMEDIR/fastcgi
75 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
76 for use with wrappers (in lieu of finding the user/group associated
77 with a virtual host - under Apache2 this isn't accomodated).
79 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
80 apr_proc_create() changes.
82 *) Log when invoking and restoring the restart backoff policy.
84 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
85 OVERLAPPED structure).
87 *) Eliminate need for dummy files for external servers under Apache2
89 *) Fix auth compatibility mode handling for access checker and authorizer
91 *) Fix HEAD request handling. Based on a patch by
92 Chris Lightfoot [chris@ex-parrot.com]
94 *) [*nix] When autoupdate is enabled touch the socket when restarting
95 the processes to prevent further requests.
96 Eckebrecht von Pappenheim [evp@heise.de]
98 *) Apache 2.0 support.
100 *) [WIN32] Don't read from a potentially closed named pipe.
101 Philip Gladstone [philip@okena.com]
103 *) Require the Apache version 1.3.6 or later to eliminate some signal
106 *) [WIN32] Use asyncronous io with named pipes instead of polled
107 nonblocking io. This should eliminate the last of the npipe issues.
109 *) Handle an application returning a complete and valid response without
110 having consumed all of the data sent to it.
112 *) Consume remaining client data (RESPONDERs only) if any.
114 *) Add support for backing off attempts to start applications that continuously
115 fail to start. Three new macros defined in mod_fastcgi.h control this
116 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
118 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
119 applications that do not (properly) support the shutdown event (this
120 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
121 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
122 interval between signaling a proper shutdown and wacking the process(s)
123 with a TerminateProcess().
125 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
126 setting it was just plain broken.
128 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
130 *) Fix 'FastCgiConfig -autoUpdate'.
132 *) Fix 'FastCgiConfig -flush'.
134 *) Prevent silly maxProcesses and processSlack combinations.
135 Dmitry Dorofeev [dima@yasp.com]
137 *) Properly handle the killing of idle processes when one takes a long time
138 to exit once signaled down (or the config is funky).
139 Dmitry Dorofeev [dima@yasp.com]
141 *) Always kill the youngest instance of an application. Suggested by
142 Dmitry Dorofeev [dima@yasp.com]
146 *) Delay the logging of write errors to the pm to account for shutdown/restart.
148 *) (Win32) An assortment of fixes.
150 *) Fix some broken casts that were likely the cause of an assert.
152 *) Win32. Eliminate forward slashes from the named pipe path name.
153 Gerald Richter [richter@ecos.de]
155 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
156 applications it spawns. [ryans@amazon.com]
158 *) Added support for the -flush argument to FastCgiConfig.
159 Eric Sit [esit@alum.mit.edu]
161 *) Change the "which call to module_init() is this" check to a more
162 reliable approach. Doru Petrescu [pdoru@kappa.ro]
164 *) Close the old pipe file descriptor in apache main on USR1/HUP
165 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
167 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
168 [tetsuya@secom-sis.co.jp]
170 *) Eliminate the use of locks to assist in the clean shutdown of
171 applications. Instead, it is assumed that applications handle
172 termination signals properly (this is now embedded in the C
175 *) Fix Win32 process termination. Proper operation requires the use of
176 an updated application lib (termination is now signalled with an
177 Event and handled by specialized thread).
179 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
181 *) Added code so if the last instance of a dynamic application died without
182 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
183 configuration allows the last instance to be killed, then allow it to die).
184 Andrew Benham [adsb@bigfoot.com]
186 *) Fix the loadFactor calculation used to determine when dyanmic
187 applications could be killed off due to low demand [adsb@bigfoot.com].
189 *) Fix a deadlock condition that could occur with Win32 named
192 *) Fix a potential deadlock condition when FastCGI application
193 sent responses while still reading the client request (POST data).
197 *) Allow absolute pathnames in the -socket argument. Suggested by
198 Christian Jaeger [christian.jaeger@sl.ethz.ch].
200 *) Don't invoke suexec when the user/group for the fastcgi application
201 is the same as the apache main server. This is consistent with
202 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
204 *) Reset the apache drop dead timer upon successful read or writes
205 to/from the client. This eliminates timeouts that were occuring
206 during the large file transfers to/from slow clients.
208 *) Support generic wrappers such as cgiwrap by eliminating dependencies
209 on Apache's SUEXEC, renaming the FastCgiSuexec directive
210 FastCgiWrapper and eliminating any checks regarding the target
211 application (this is the repsonibility of the wrapper).
213 *) Fix a nasty bug that occurred when a client aborted a POST request
214 before the connection to a dynamic FastCGI application was opened.
215 The application's lock file descriptor wasn't setup but was being
216 closed which resulted in FD0 being closed. Normally this is open to
217 /dev/null and should pose no problems except that because the FD
218 was available it was being returned by Apache's accept(). This
219 caused it to be registered for pool cleanup. mod_cgi though moves
220 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
221 cleanup. Problem identified by checksum@163.net.
223 Changes with mod_fastcgi 2.2.8
225 *) Eliminate the concept of disabled applications. If a failure
226 occurs trying to setup an application (e.g. bind() error) its
227 tried repeatedly every init-start-delay seconds.
229 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
231 Changes with mod_fastcgi 2.2.6
233 *) Shutdown the PM when Apache appears to have disappeared.
235 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
237 *) (Win32) More dynamic fixes.
239 *) (Win32) Eliminate the per application named pipe mutex. Its now
240 per process to keep the lib happy (the use _FCGI_MUTEX_ should
241 removed from the lib). This allows pipe-based applications to
242 now handle multiple simoultanous requests (one per process).
244 *) Increase the number of open FDs we look to close when spawning apps.
246 *) Prevent an assert from popping unnecessarily.
248 *) (Win32) Add support for interpretter scripts.
250 *) (Win32) Fix named pipe handling (problems with large responses).
252 *) (Win32) Remove the "can exec" check.
254 Changes with mod_fastcgi 2.2.4
256 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
257 Rob Saccoccio [robs@chelsea.net]
259 *) Remove request type restriction (GET and POST only) in order to support
260 Web DAV requests. Scott Robertson [sroberts@codeit.com]
262 *) Allow requests for URLs such as /server/some/other/qualifier to match a
263 FastCGI server defined as /server. This was done primarily for Java
264 Servlets, but is generically useful.
266 *) Change the comm between the PM and the request handlers from a regular
267 file to a pipe and an assortment of other dynamic fixes.
269 *) Log dynamic process termination scheduling and the resulting
270 process exit notification.
272 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
273 application process from being killed off due to low demand.
275 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
277 *) Fix bugs that prevented dynamic processes from being shutdown when
278 the load subsided. Lars Heete [heete@do.isst.fhg.de]
280 *) Prevent dynamic processes from being scheduled to be started before
281 the init-start-delay or restart-delay period expires. This prevents
282 the queuing of process start requests while an application which has
283 a long initialization period starts up.
285 *) When suexec is enabled, allow the process manager to become root again in
286 order to signal applications it spawned.
288 *) Change the "server started" log message level from INFO to WARN.
290 *) Fix dynamic server "Connection Refused" handling.
292 *) Fix demand-based dynamic process spawing (uses a more portable approach).
294 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
295 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
296 longer than this period. It applies to the initiation of connections as
297 well (and thus is similar to appConnTimeout). Default is 30 seconds.
299 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
300 is more platform portable/predicable.
302 *) Leave STDOUT and STDERR open to the main server error_log. This should
303 eliminate a pile of protocol errors and having to track down 3rd party
304 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
305 be fixed in the application to use the FastCGI I/O, it just allows these
306 applications to run when previously they'd crash and burn.
308 *) Add the pass-header arg to server directives.
310 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
311 is done for CGI) - at least until PR2580 is resolved.
313 *) Initialize the default (empty) environment in a more socially acceptable
316 *) Miscellaneous doc improvements based on notes I've collected up over the
319 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
320 that resulted in not or improperly terminated strings.
322 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
323 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
325 Changes with mod_fastcgi 2.2.2
327 *) Added support for blocking connect()s by setting appConnTimeout to 0.
328 Non-blocking connect()s (the default) can be troublesome on some platforms.
329 Its expected that blocking connect()s will become the default (and
330 non-blocking connect()s will become optional) in the next release.
332 *) Dump a compile time error if the version of Apache is too old.
334 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
335 to Apache releases prior to 1.3.6.
337 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
341 Always restart a failed dynamic application if it is the last instance.
342 This means there once started there will always be at least one process
343 instance of a dynamic application.
345 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
346 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
347 the PM to start (another) application instance may help empty it faster.
349 Change two sleep() calls to select() based snoozes because alarm() is in
350 effect and sleep() and alarm() don't always play nice together.
352 Fixed a couple of error messages.
354 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
355 was ignored and the default was always used.
357 *) Allow the -initial-env argument to be used to pass variables from the
358 Apache process environment to the FastCGI server (by specifying a
359 variable name without the "=" or a value). [suggested by
360 Martin Lichtin <lichtin@oanda.com>]
362 *) Cleanup some debug macros.
364 *) Improved script stderr handling. [based on suggestions by
365 David Birnbaum <davidb@chelsea.net>]
367 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
368 updates to the INSTALL doc.
370 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
371 5 to 100. Its still configurable with the -listen-queue-depth option.
372 This should help eliminate the FAQ - Why do I see "Connection Refused"
375 *) Fix a bug in FastCgiExternalServer that broke support for external servers
376 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
378 Changes with mod_fastcgi 2.2.1
380 *) Updates to the INSTALL doc to describe building as a DSO.
382 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
383 Apache was running ("apachectl restart" does this), the contents of the
384 dynamic directory (dynamic sockets and the mbox) would be blown away
387 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
388 Apache header file) to prevent a compile error on systems without
391 *) During auth requests, the subprocess_env table was left holding variables
392 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
393 were passed to other processing phases (such as CGI/FastCGI). The
394 subprocess_env is now restored to its pre-auth condition.
396 *) Added a FastCGI Authorizer Role compatibility mode which implements the
397 specification to the tee. Use -compat arg with any of the Auth server
398 directives. This is intended for new and existing FastCGI authorizer
399 applications that require compatibility with other server implementations.
401 *) Fix logging from the process manager. Previously, all calls to the
402 logging routines used NULL for the server_rec. This works fine in a dev
403 environment, but results in most of these messages being tossed by
404 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
406 *) Always setup fr->header so that in the event the FastCGI server
407 doesn't get a valid FastCGI protocol header over the wire, we
408 can still print our error message to the log from do_work().
410 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
411 defined (Don Locrasto [locrasto@iceminer.com]).
413 Changes with mod_fastcgi 2.2.0
415 *) Serious rewrite of mod_fastcgi.html. An example conf will have
416 to wait until another time - I'm burned out on docs updates.
418 *) Add to the mod_fastcgi to the server version string.
420 *) SCO doesn't like the const in the arg to inet_addr() and
423 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
425 *) Deleted redundant header files from fcgi.h.
427 *) Full path names are no longer required for directives which take
428 file names as arguments. Like other Apache directives you can
429 specify paths from server_root (don't start with a "/").
431 *) The fcgiKillMgr is gone. This intermediate process sat between
432 the Apache parent and the FastCGI Process Manager under Apache
433 1.2. The process manager tries to name itself fcgi_pm (it used
436 *) Authentication, Authorization, and Access phases are now
437 supported. I've bent the FastCgi spec a bit: as many of the standard
438 environment variables (that were easy) are sent (the spec says don't
439 send some by name), all headers (except Status) sent by the auth
440 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
441 as environment variables rather than just those prefixed by
442 "Variable-". Custom responses for auth failures aren't supported
443 (yet.. we'll see what the demand for this is like). In addition to
444 the FastCGI protocol defined environment variable "FCGI_ROLE" being
445 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
446 set indicating which of the three phases is being processed. See
447 the docs for info on the new per-directory directives:
449 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
450 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
451 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
453 *) The code has been broken into logical chunks making figuring it
454 out much easier. All of the #defines a user may want to change are
455 now in mod_fastcgi.h.
457 *) All the logging has been converted to the "supported" Apache
458 logging routines, ap_log_error() and ap_log_rerror(). Log entries
459 for request specific errors should now be directed to the correct
460 server errorlog (if your running more than one). Some attempt has
461 been made at setting appropriate log levels (comments of course are
462 welcome) and printing errno info when its of value. All of log
463 messages now have "FastCGI" in them (nice for grep).
465 *) All of the calls to Apache routines have been updated to the ap_
466 convention, eliminating the need for the compat header.
468 *) By default we no longer flush every piece of data we get from the
469 FastCGI server to the client. This allows the FastCGI server to
470 release without having to wait for the client flush to complete.
471 The old behaviour can revived with the -flush argument to AppClass.
473 *) I renamed some of the directives for consistency (the old names
476 AppClass -> FastCgiServer,
477 FCGIConfig -> FastCgiConfig,
478 ExternalAppClass -> FastCgiExternalServer
480 *) Directive arguments are no longer case sensitive.
482 *) The module now uses hard_timeout() rather than soft_timeout. This
483 means that the module will longjump out of the request if the
484 drop-dead timeout expires (set with Apache's Timeout directive, the
485 default is 5min) or if the client closes the connection (SIGPIPE).
486 This is more typical for a Apache modules and I'm not convinced we
487 properly handle all of the error cases well enough to use
488 soft_timeout (I can think of one place we don't anyway). This means
489 your FastCGI application can see SIGPIPE.
491 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
492 Apache. Without this the default behaviour is exit().
494 *) It should (I think) run now under DSO now.
496 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
498 *) The broken -initial-env bug ins 2.1b1 is fixed.
500 *) Maybe some other stuff.
504 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
505 commit notes and stick 'em in here.. when I get some free time. ;)
510 *** Originally from docs/README..
512 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
513 From: Stanley Gambarin <stanleyg@cs.bu.edu>
515 *) Yet more changes to the source distribution. Separated out
516 the Tcl dynamic string and buffer libraries into separate files.
517 Did the same for the OS library, however, it is not really an
518 abstraction... more like a bunch of wrappers. Added to DEVNOTES
519 a note about the problem of Apache not allowing including header
522 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
523 From: Stanley Gambarin <stanleyg@cs.bu.edu>
525 *) More changes to the installation scripts. Theoretically we
526 should be able to to use the same Makefile/installation script
527 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
528 layer to use the same source code for both 1.2.x and 1.3.x sources.
530 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
531 From: Stanley Gambarin <stanleyg@cs.bu.edu>
533 *) Created Makefile to compile the module into the archive library,
534 just like the proxy module. Also initial draft of the installation
535 script, which should work for both 1.2 and 1.3 sources of Apache.
537 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
538 From: Stanley Gambarin <stanleyg@cs.bu.edu>
540 *) More source reorg. Separated out some header files. Need much
541 more work on this. Updated the TODO file. Added fcgivers.h and
542 CHANGES file to track the history.
544 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
545 From: Stanley Gambarin <stanleyg@cs.bu.edu>
547 *) Source reorganization. This is done to provide a basis for later
548 functionality implementation. See TODO file for more information
549 of what possible future enhancements are possible/desired.
551 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
552 From: David MacKenzie <djm@va.pubnix.com>
554 *) Create the "dynamic" dir and "mbox" as the user specified in the
555 User and Group directives instead of as root. They are created with
556 restrictive permissions, and then the module checks to see whether the
557 user specified in the User and Group directives has read, write, and
558 execute permissions on them. Those permissions had been explicitly
559 denied by creating them as "root".
561 *) Some documentation was garbled or incomplete.
563 *) When a FastCGI application can't be execl'd, the code used a
564 value of errno that may have been stomped on by intervening system
565 calls. That happened on BSD/OS 2.1, where the error_log was
566 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
567 An intervening stdio call was setting errno as a side-effect, which
568 it has the right to do.
570 *) Erroneous fprintf arguments, a missing return value, an unused
571 function, missing declarations, and other problems detected by gcc
574 *) Duplicated code to create the lock file name merged into a single
575 function, closing several memory leaks.
577 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
578 FCGI app in a given dynamic class would live forever, even if it
579 hadn't been requested for weeks.
581 *) The test for whether to keep looking for dynamic app victims to
582 kill was backwards, so no dynamic apps were ever selected as
585 *) The killInterval and updateInterval were being ignored; the
586 sigsuspend() forced recalculations only when a child died,
587 instead of at the intervals specified.
589 *) When a dynamic app couldn't be started, the program SEGV'd when
590 trying to free the ipcAddrPtr twice.
592 *) Remove the dead lock file and socket when cleaning up after a
593 server whose last child has been killed, so the FastCGIHandler
594 isn't fooled into thinking there is still a process serving
597 *) Make file locking robust in the presence of signals.
599 *) Rename some badly named variables.
601 *) Fix typos in many comments.
603 *) Fix some memory and file descriptor leaks.
605 *) Make the blocking kill of a server closer to working.
607 *) More fixes to calculations in dynamic application management.
610 *** Originally from docs/README.OMI..
612 What's New: Version 2.0b1, 16 Apr 1997
614 *) Implemented a mechanism by which FastCGI applications are started by
615 the web server on the first request and continue running. Also, a
616 heuristics have been implemented to allow for dynamic killing of the
617 running FastCGI apps. The configuration options are supplied via
618 the new FCGIConfig directive, that is described in the
621 *) When performing internal redirect, since the original request's body
622 has already been read, do not allow the redirected request to think
623 that is has one [body].
625 *) More conditional compilation for OS/2.
627 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
629 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
631 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
632 force some output to be written in the period of inactivity. This
633 circumvents the Apache's buffering during the server-push.
635 *) Added a header-parser function placeholder for compatibility with an
638 *) Additional checks have been placed with regard to permissions of the
639 FastCGI processes. These include the existence of the file,
640 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
642 *) Implemented a restart cleanup, so that no parentlesses FastCGI
643 applications are left after Apache is restarted and/or terminated.
644 This was accomplished via two independent mechanisms, where the
645 first one prevents a start of the process manager on the first
646 reading of the configuration files. The second mechanism deals away
647 from Apache's implementation of fork() (via spawn_child()) and
648 implements its own forking, thus removing any dependencies on the
649 Apache to cleanup processes during termination phase.
651 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
652 itself in the file conf.h
654 What's New: Version 1.4.3, 15 Jan 1997
656 *) Fixed compilation warnings for various platforms, as well as
657 conditional compilation for OS/2.
659 What's New: Version 1.4.2, 12 Dec 1996
661 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
662 will proceed under this version of Apache web server.
664 *) As the result of porting, the "include virtual" construct of SSI
665 will now work correctly using either <Location> or <Directory>
668 What's New: Version 1.4.1, 4 Dec 1996
670 *) Checks have been removed from the ScanCGIHeaders that provided
671 for the presence of both Status and Location headers as being
672 an error. Contradictory to CGI/1.1 Internet Draft, both of
673 these headers are used by the current CGI applications.
675 What's New: Version 1.4, 22 Nov 1996
677 *) Added the -port option to AppClass, allowing TCP/IP communication.
678 Added the -socket option to AppClass, allowing Unix domain
679 communication via a configurable pathname.
681 *) Added the ExternalAppClass directive, allowing TCP/IP
682 communication with remote FastCGI applications.
684 *) The handler had its own code for generating HTTP response headers;
685 now it uses Apache's. This reduces the size of the module.
686 More importantly, it fixes the bug in which "include virtual"
687 sees the HTTP response headers.
689 *) The response header parser performed very little checking.
690 Now the parser enforces the guidelines in the CGI/1.1
691 Internet-Draft: Status and Location are mutually exclusive,
692 Location can only be a response to GET or HEAD,
693 CGI response headers can't be repeated, etc. (The CGI response
694 headers are the ones the handler interprets: Status, Location,
697 *) The response header parser used to miss CGI headers
698 with no whitespace after the colon, e.g. "Status:200 OK".
700 *) The response header parser sometimes interpreted the first line
701 of content as an RFC822 continuation line.
703 *) The handler implemented a nonstandard version of Location
704 which never used internal redirects. The handler
705 now attempts to implement Location as specified in the
706 CGI/1.1 Internet-Draft. The module documentation explains
709 *) Error log entries from the response header parser were
710 pretty uninformative; they are better now. If a header is
711 malformed, the log entry includes it. If the headers are
712 unterminated rather than malformed, the log entry says that,
713 and says how many bytes were received from the app.
715 *) When the application manager forked a new process, and that
716 process ran into trouble before executing the first instruction
717 of application code, the process used to exit with status =
718 errno. This made certain configuration problems (e.g. incorrect
719 file permissions when server parent is root) quite difficult
720 to pin down. Now the failing child process opens up the error log
721 and writes an informative entry before exiting.
723 *) The module now correctly handles a slash at the end of the
724 DocumentRoot directive. This was a one-line fix.
726 What's New: Version 1.3.3, 17 Oct 1996
728 *) The module now registers its request handler under the name
729 fastcgi-script in addition to the name application/x-httpd-fcgi.
730 This was a one-line addition to the module, but had quite
731 a large impact on the documentation and sample configuration.
733 What's New: Version 1.3.2, 27 Sept 1996
735 *) On some systems (SunOS, Linux?), fopen for append has a bug
736 that strikes when two processes append to the same file.
737 This bug causes the process manager to corrupt the error log.
738 Work around the bug by calling open, then fdopen. A patch
739 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
740 (Reported by Bob Ramstad.)
742 What's New: Version 1.3.1, 17 Sept 1996
744 *) On Linux and SunOS, the Apache default user_id (-1) is not a
745 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
746 Scott Langley, others.)
748 *) On some systems (Linux, some Solaris?, Irix?), connect requires
749 write access to a Unix Domain socket, so provide it. (Reported
750 by Scott Langley, freeform@wired.com, others.)
752 *) If you hit a .fcg file, but there's no AppClass defined, the
753 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
756 What's New: Version 1.3, 4 Sept 1996
758 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
760 *) Module violated the Apache buff abstraction.
761 Fixed. Might now work with SSL (not tested).
762 As part of the fix, eliminated support for
763 Apache 1.0x versions.
765 *) Module failed to chmod the Unix domain listening sockets
766 it created, so protections were set according to the current
769 *) Module forked too many process manager processes: One
770 per AppClass. When the Apache parent ran as root, these
771 process manager processes ran as root. New process
772 manager is a single process (only started if AppClass is
773 used) and runs with same privileges as other children.
774 The new process manager doesn't do any polling, so
775 there's less system overhead than before.
777 *) AppClass insisted on getting at least two arguments. Fixed.
779 *) Module used setjmp/longjmp, causing compiler warnings
780 on some platforms (e.g. Linux.) The module no longer
783 *) Module created listening sockets in /tmp, where they were
784 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
785 directive to give control over the location of listening
788 *) Module wrote error log entries without a timestamp. Fixed.
790 *) AppClass directive wrote error messages to stderr in addition
791 to returning a char * message to the Apache core. Fixed.
793 What's New: Version 1.2, 3 June 1996
795 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
798 Add version string: APACHEVERSION macro to mod_fastcgi.c
800 *) chown FastCGI socket to user_id and setuid to user_id for app
803 *) Modify GetFromStream() by having it call an OS dependent
804 function GetStreamSize(FILE *) which uses FILE internal data member.
805 Linux users might need to modify GetStreamSize().
807 What's New: Version 1.1, 10 May 1996
809 *) If you specify a non-existent executable in the AppClass directive,
810 or if the file exists and it does not have execute permission you
811 get a constant stream of error messages telling you that "program
812 terminated due to a signal".
814 *) The mod_fastcgi module should use the standard Apache error logging
815 facility instead of writing to stderr.
817 Version 1.0, 30 April 1996