3 *) Do the connect() to the application *after* collecting a chunk
4 of client data. This reinstates the pre-2.4.0 behaviour.
5 James Jurach [muaddib@erf.net].
7 *) Remove an assert that was triggering on WIN when spawn() failed.
9 *) Provide the NO_SUEXEC_FOR_AP_USER_N_GROUP macro for building
10 mod_fastcgi with the AP13 suexec behaviour (don't use suexec
11 if httpd's user and group match that needed for the application).
13 *) Prevent the use of all but the "auth" directives from being used
14 anywhere but in global scope. Prevent more than one instance
15 of the FastCgiWrapper directive.
17 *) Return NOT_FOUND (404) or FORBIDDEN (403) instead of
18 INTERNAL_SERVER_ERROR (500) when there are configuration issues
19 or the script does't exist. Suggested by Jeff Lawson [bovine@ud.com].
23 *) [*nix Security] - When FastCgiWrapper (FastCgiSuexec) was in use
24 and a vhost configured to use the same uid/gid as the main
25 server, mod_fastcgi would not bother using the wrapper (suexec)
26 because its effective uid/gid was already appropriate. This is
27 consistent with Apache's v1.3 mod_cgi behaviour. There are two
28 problems with this approach: 1) when FastCgWrapper is in use
29 mod_fastcgi's process manager keeps its root privileges (as its
30 real uid/gid) so it can terminate the applications its starts -
31 this privilege was being passed to applications when the use of
32 the wrapper was bypassed 2) wrappers are often employed to
33 perform functionality beyond setting the uid/gid - by not calling
34 the wrapper under certain circumstances, application invocation
35 environments were inconsistent. With this change, the wrapper is
36 always used (when enabled) under both Apache 1.3 and 2.
37 Reported by Michael Richards [michael@fastmail.ca].
39 *) [*nix/AP2] Use the vhost uid/gid instead of the server uid/gid
40 for dynamic application invocation when the FastCgiWrapper is in use.
41 Reported by Michael Richards [michael@fastmail.ca]
43 *) [*nix] Fix handling of FastCgiWrapper when passed a real path,
44 i.e. other than "on" or "off". Michael Richards [michael@fastmail.ca]
46 *) Eliminate the logging of "incomplete headers (0 bytes) received from
47 server" when a client aborts.
49 *) [WIN32] Fix a delay in handling large POSTs to named pipe based
50 servers. Philip Gladstone [philip@okena.com]
52 *) [*nix/AP2] Prevent the module from being initalized twice at startup
53 (resulting in confusing error messages to the log).
55 *) Eliminate the need for SetHandler or AddHandler with static or
56 external applications.
58 *) Limit PM requests to start a dynamic application
59 to 5sec to prevent endless spinning (this is a drop-dead
60 limit that should only occur if the socket/named_pipe directory
61 is removed out from under a running server).
63 *) [*nix] Change the default socket directory from /tmp/fcgi to:
65 Apache2 - RUNTIMEDIR/fastcgi
67 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
68 for use with wrappers (in lieu of finding the user/group associated
69 with a virtual host - under Apache2 this isn't accomodated).
71 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
72 apr_proc_create() changes.
74 *) Log when invoking and restoring the restart backoff policy.
76 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
77 OVERLAPPED structure).
79 *) Eliminate need for dummy files for external servers under Apache2
81 *) Fix auth compatibility mode handling for access checker and authorizer
83 *) Fix HEAD request handling. Based on a patch by
84 Chris Lightfoot [chris@ex-parrot.com]
86 *) [*nix] When autoupdate is enabled touch the socket when restarting
87 the processes to prevent further requests.
88 Eckebrecht von Pappenheim [evp@heise.de]
90 *) Apache 2.0 support.
92 *) [WIN32] Don't read from a potentially closed named pipe.
93 Philip Gladstone [philip@okena.com]
95 *) Require the Apache version 1.3.6 or later to eliminate some signal
98 *) [WIN32] Use asyncronous io with named pipes instead of polled
99 nonblocking io. This should eliminate the last of the npipe issues.
101 *) Handle an application returning a complete and valid response without
102 having consumed all of the data sent to it.
104 *) Consume remaining client data (RESPONDERs only) if any.
106 *) Add support for backing off attempts to start applications that continuously
107 fail to start. Three new macros defined in mod_fastcgi.h control this
108 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
110 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
111 applications that do not (properly) support the shutdown event (this
112 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
113 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
114 interval between signaling a proper shutdown and wacking the process(s)
115 with a TerminateProcess().
117 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
118 setting it was just plain broken.
120 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
122 *) Fix 'FastCgiConfig -autoUpdate'.
124 *) Fix 'FastCgiConfig -flush'.
126 *) Prevent silly maxProcesses and processSlack combinations.
127 Dmitry Dorofeev [dima@yasp.com]
129 *) Properly handle the killing of idle processes when one takes a long time
130 to exit once signaled down (or the config is funky).
131 Dmitry Dorofeev [dima@yasp.com]
133 *) Always kill the youngest instance of an application. Suggested by
134 Dmitry Dorofeev [dima@yasp.com]
138 *) Delay the logging of write errors to the pm to account for shutdown/restart.
140 *) (Win32) An assortment of fixes.
142 *) Fix some broken casts that were likely the cause of an assert.
144 *) Win32. Eliminate forward slashes from the named pipe path name.
145 Gerald Richter [richter@ecos.de]
147 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
148 applications it spawns. [ryans@amazon.com]
150 *) Added support for the -flush argument to FastCgiConfig.
151 Eric Sit [esit@alum.mit.edu]
153 *) Change the "which call to module_init() is this" check to a more
154 reliable approach. Doru Petrescu [pdoru@kappa.ro]
156 *) Close the old pipe file descriptor in apache main on USR1/HUP
157 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
159 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
160 [tetsuya@secom-sis.co.jp]
162 *) Eliminate the use of locks to assist in the clean shutdown of
163 applications. Instead, it is assumed that applications handle
164 termination signals properly (this is now embedded in the C
167 *) Fix Win32 process termination. Proper operation requires the use of
168 an updated application lib (termination is now signalled with an
169 Event and handled by specialized thread).
171 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
173 *) Added code so if the last instance of a dynamic application died without
174 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
175 configuration allows the last instance to be killed, then allow it to die).
176 Andrew Benham [adsb@bigfoot.com]
178 *) Fix the loadFactor calculation used to determine when dyanmic
179 applications could be killed off due to low demand [adsb@bigfoot.com].
181 *) Fix a deadlock condition that could occur with Win32 named
184 *) Fix a potential deadlock condition when FastCGI application
185 sent responses while still reading the client request (POST data).
189 *) Allow absolute pathnames in the -socket argument. Suggested by
190 Christian Jaeger [christian.jaeger@sl.ethz.ch].
192 *) Don't invoke suexec when the user/group for the fastcgi application
193 is the same as the apache main server. This is consistent with
194 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
196 *) Reset the apache drop dead timer upon successful read or writes
197 to/from the client. This eliminates timeouts that were occuring
198 during the large file transfers to/from slow clients.
200 *) Support generic wrappers such as cgiwrap by eliminating dependencies
201 on Apache's SUEXEC, renaming the FastCgiSuexec directive
202 FastCgiWrapper and eliminating any checks regarding the target
203 application (this is the repsonibility of the wrapper).
205 *) Fix a nasty bug that occurred when a client aborted a POST request
206 before the connection to a dynamic FastCGI application was opened.
207 The application's lock file descriptor wasn't setup but was being
208 closed which resulted in FD0 being closed. Normally this is open to
209 /dev/null and should pose no problems except that because the FD
210 was available it was being returned by Apache's accept(). This
211 caused it to be registered for pool cleanup. mod_cgi though moves
212 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
213 cleanup. Problem identified by checksum@163.net.
215 Changes with mod_fastcgi 2.2.8
217 *) Eliminate the concept of disabled applications. If a failure
218 occurs trying to setup an application (e.g. bind() error) its
219 tried repeatedly every init-start-delay seconds.
221 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
223 Changes with mod_fastcgi 2.2.6
225 *) Shutdown the PM when Apache appears to have disappeared.
227 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
229 *) (Win32) More dynamic fixes.
231 *) (Win32) Eliminate the per application named pipe mutex. Its now
232 per process to keep the lib happy (the use _FCGI_MUTEX_ should
233 removed from the lib). This allows pipe-based applications to
234 now handle multiple simoultanous requests (one per process).
236 *) Increase the number of open FDs we look to close when spawning apps.
238 *) Prevent an assert from popping unnecessarily.
240 *) (Win32) Add support for interpretter scripts.
242 *) (Win32) Fix named pipe handling (problems with large responses).
244 *) (Win32) Remove the "can exec" check.
246 Changes with mod_fastcgi 2.2.4
248 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
249 Rob Saccoccio [robs@chelsea.net]
251 *) Remove request type restriction (GET and POST only) in order to support
252 Web DAV requests. Scott Robertson [sroberts@codeit.com]
254 *) Allow requests for URLs such as /server/some/other/qualifier to match a
255 FastCGI server defined as /server. This was done primarily for Java
256 Servlets, but is generically useful.
258 *) Change the comm between the PM and the request handlers from a regular
259 file to a pipe and an assortment of other dynamic fixes.
261 *) Log dynamic process termination scheduling and the resulting
262 process exit notification.
264 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
265 application process from being killed off due to low demand.
267 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
269 *) Fix bugs that prevented dynamic processes from being shutdown when
270 the load subsided. Lars Heete [heete@do.isst.fhg.de]
272 *) Prevent dynamic processes from being scheduled to be started before
273 the init-start-delay or restart-delay period expires. This prevents
274 the queuing of process start requests while an application which has
275 a long initialization period starts up.
277 *) When suexec is enabled, allow the process manager to become root again in
278 order to signal applications it spawned.
280 *) Change the "server started" log message level from INFO to WARN.
282 *) Fix dynamic server "Connection Refused" handling.
284 *) Fix demand-based dynamic process spawing (uses a more portable approach).
286 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
287 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
288 longer than this period. It applies to the initiation of connections as
289 well (and thus is similar to appConnTimeout). Default is 30 seconds.
291 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
292 is more platform portable/predicable.
294 *) Leave STDOUT and STDERR open to the main server error_log. This should
295 eliminate a pile of protocol errors and having to track down 3rd party
296 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
297 be fixed in the application to use the FastCGI I/O, it just allows these
298 applications to run when previously they'd crash and burn.
300 *) Add the pass-header arg to server directives.
302 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
303 is done for CGI) - at least until PR2580 is resolved.
305 *) Initialize the default (empty) environment in a more socially acceptable
308 *) Miscellaneous doc improvements based on notes I've collected up over the
311 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
312 that resulted in not or improperly terminated strings.
314 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
315 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
317 Changes with mod_fastcgi 2.2.2
319 *) Added support for blocking connect()s by setting appConnTimeout to 0.
320 Non-blocking connect()s (the default) can be troublesome on some platforms.
321 Its expected that blocking connect()s will become the default (and
322 non-blocking connect()s will become optional) in the next release.
324 *) Dump a compile time error if the version of Apache is too old.
326 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
327 to Apache releases prior to 1.3.6.
329 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
333 Always restart a failed dynamic application if it is the last instance.
334 This means there once started there will always be at least one process
335 instance of a dynamic application.
337 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
338 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
339 the PM to start (another) application instance may help empty it faster.
341 Change two sleep() calls to select() based snoozes because alarm() is in
342 effect and sleep() and alarm() don't always play nice together.
344 Fixed a couple of error messages.
346 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
347 was ignored and the default was always used.
349 *) Allow the -initial-env argument to be used to pass variables from the
350 Apache process environment to the FastCGI server (by specifying a
351 variable name without the "=" or a value). [suggested by
352 Martin Lichtin <lichtin@oanda.com>]
354 *) Cleanup some debug macros.
356 *) Improved script stderr handling. [based on suggestions by
357 David Birnbaum <davidb@chelsea.net>]
359 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
360 updates to the INSTALL doc.
362 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
363 5 to 100. Its still configurable with the -listen-queue-depth option.
364 This should help eliminate the FAQ - Why do I see "Connection Refused"
367 *) Fix a bug in FastCgiExternalServer that broke support for external servers
368 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
370 Changes with mod_fastcgi 2.2.1
372 *) Updates to the INSTALL doc to describe building as a DSO.
374 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
375 Apache was running ("apachectl restart" does this), the contents of the
376 dynamic directory (dynamic sockets and the mbox) would be blown away
379 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
380 Apache header file) to prevent a compile error on systems without
383 *) During auth requests, the subprocess_env table was left holding variables
384 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
385 were passed to other processing phases (such as CGI/FastCGI). The
386 subprocess_env is now restored to its pre-auth condition.
388 *) Added a FastCGI Authorizer Role compatibility mode which implements the
389 specification to the tee. Use -compat arg with any of the Auth server
390 directives. This is intended for new and existing FastCGI authorizer
391 applications that require compatibility with other server implementations.
393 *) Fix logging from the process manager. Previously, all calls to the
394 logging routines used NULL for the server_rec. This works fine in a dev
395 environment, but results in most of these messages being tossed by
396 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
398 *) Always setup fr->header so that in the event the FastCGI server
399 doesn't get a valid FastCGI protocol header over the wire, we
400 can still print our error message to the log from do_work().
402 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
403 defined (Don Locrasto [locrasto@iceminer.com]).
405 Changes with mod_fastcgi 2.2.0
407 *) Serious rewrite of mod_fastcgi.html. An example conf will have
408 to wait until another time - I'm burned out on docs updates.
410 *) Add to the mod_fastcgi to the server version string.
412 *) SCO doesn't like the const in the arg to inet_addr() and
415 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
417 *) Deleted redundant header files from fcgi.h.
419 *) Full path names are no longer required for directives which take
420 file names as arguments. Like other Apache directives you can
421 specify paths from server_root (don't start with a "/").
423 *) The fcgiKillMgr is gone. This intermediate process sat between
424 the Apache parent and the FastCGI Process Manager under Apache
425 1.2. The process manager tries to name itself fcgi_pm (it used
428 *) Authentication, Authorization, and Access phases are now
429 supported. I've bent the FastCgi spec a bit: as many of the standard
430 environment variables (that were easy) are sent (the spec says don't
431 send some by name), all headers (except Status) sent by the auth
432 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
433 as environment variables rather than just those prefixed by
434 "Variable-". Custom responses for auth failures aren't supported
435 (yet.. we'll see what the demand for this is like). In addition to
436 the FastCGI protocol defined environment variable "FCGI_ROLE" being
437 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
438 set indicating which of the three phases is being processed. See
439 the docs for info on the new per-directory directives:
441 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
442 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
443 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
445 *) The code has been broken into logical chunks making figuring it
446 out much easier. All of the #defines a user may want to change are
447 now in mod_fastcgi.h.
449 *) All the logging has been converted to the "supported" Apache
450 logging routines, ap_log_error() and ap_log_rerror(). Log entries
451 for request specific errors should now be directed to the correct
452 server errorlog (if your running more than one). Some attempt has
453 been made at setting appropriate log levels (comments of course are
454 welcome) and printing errno info when its of value. All of log
455 messages now have "FastCGI" in them (nice for grep).
457 *) All of the calls to Apache routines have been updated to the ap_
458 convention, eliminating the need for the compat header.
460 *) By default we no longer flush every piece of data we get from the
461 FastCGI server to the client. This allows the FastCGI server to
462 release without having to wait for the client flush to complete.
463 The old behaviour can revived with the -flush argument to AppClass.
465 *) I renamed some of the directives for consistency (the old names
468 AppClass -> FastCgiServer,
469 FCGIConfig -> FastCgiConfig,
470 ExternalAppClass -> FastCgiExternalServer
472 *) Directive arguments are no longer case sensitive.
474 *) The module now uses hard_timeout() rather than soft_timeout. This
475 means that the module will longjump out of the request if the
476 drop-dead timeout expires (set with Apache's Timeout directive, the
477 default is 5min) or if the client closes the connection (SIGPIPE).
478 This is more typical for a Apache modules and I'm not convinced we
479 properly handle all of the error cases well enough to use
480 soft_timeout (I can think of one place we don't anyway). This means
481 your FastCGI application can see SIGPIPE.
483 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
484 Apache. Without this the default behaviour is exit().
486 *) It should (I think) run now under DSO now.
488 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
490 *) The broken -initial-env bug ins 2.1b1 is fixed.
492 *) Maybe some other stuff.
496 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
497 commit notes and stick 'em in here.. when I get some free time. ;)
502 *** Originally from docs/README..
504 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
505 From: Stanley Gambarin <stanleyg@cs.bu.edu>
507 *) Yet more changes to the source distribution. Separated out
508 the Tcl dynamic string and buffer libraries into separate files.
509 Did the same for the OS library, however, it is not really an
510 abstraction... more like a bunch of wrappers. Added to DEVNOTES
511 a note about the problem of Apache not allowing including header
514 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
515 From: Stanley Gambarin <stanleyg@cs.bu.edu>
517 *) More changes to the installation scripts. Theoretically we
518 should be able to to use the same Makefile/installation script
519 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
520 layer to use the same source code for both 1.2.x and 1.3.x sources.
522 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
523 From: Stanley Gambarin <stanleyg@cs.bu.edu>
525 *) Created Makefile to compile the module into the archive library,
526 just like the proxy module. Also initial draft of the installation
527 script, which should work for both 1.2 and 1.3 sources of Apache.
529 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
530 From: Stanley Gambarin <stanleyg@cs.bu.edu>
532 *) More source reorg. Separated out some header files. Need much
533 more work on this. Updated the TODO file. Added fcgivers.h and
534 CHANGES file to track the history.
536 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
537 From: Stanley Gambarin <stanleyg@cs.bu.edu>
539 *) Source reorganization. This is done to provide a basis for later
540 functionality implementation. See TODO file for more information
541 of what possible future enhancements are possible/desired.
543 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
544 From: David MacKenzie <djm@va.pubnix.com>
546 *) Create the "dynamic" dir and "mbox" as the user specified in the
547 User and Group directives instead of as root. They are created with
548 restrictive permissions, and then the module checks to see whether the
549 user specified in the User and Group directives has read, write, and
550 execute permissions on them. Those permissions had been explicitly
551 denied by creating them as "root".
553 *) Some documentation was garbled or incomplete.
555 *) When a FastCGI application can't be execl'd, the code used a
556 value of errno that may have been stomped on by intervening system
557 calls. That happened on BSD/OS 2.1, where the error_log was
558 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
559 An intervening stdio call was setting errno as a side-effect, which
560 it has the right to do.
562 *) Erroneous fprintf arguments, a missing return value, an unused
563 function, missing declarations, and other problems detected by gcc
566 *) Duplicated code to create the lock file name merged into a single
567 function, closing several memory leaks.
569 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
570 FCGI app in a given dynamic class would live forever, even if it
571 hadn't been requested for weeks.
573 *) The test for whether to keep looking for dynamic app victims to
574 kill was backwards, so no dynamic apps were ever selected as
577 *) The killInterval and updateInterval were being ignored; the
578 sigsuspend() forced recalculations only when a child died,
579 instead of at the intervals specified.
581 *) When a dynamic app couldn't be started, the program SEGV'd when
582 trying to free the ipcAddrPtr twice.
584 *) Remove the dead lock file and socket when cleaning up after a
585 server whose last child has been killed, so the FastCGIHandler
586 isn't fooled into thinking there is still a process serving
589 *) Make file locking robust in the presence of signals.
591 *) Rename some badly named variables.
593 *) Fix typos in many comments.
595 *) Fix some memory and file descriptor leaks.
597 *) Make the blocking kill of a server closer to working.
599 *) More fixes to calculations in dynamic application management.
602 *** Originally from docs/README.OMI..
604 What's New: Version 2.0b1, 16 Apr 1997
606 *) Implemented a mechanism by which FastCGI applications are started by
607 the web server on the first request and continue running. Also, a
608 heuristics have been implemented to allow for dynamic killing of the
609 running FastCGI apps. The configuration options are supplied via
610 the new FCGIConfig directive, that is described in the
613 *) When performing internal redirect, since the original request's body
614 has already been read, do not allow the redirected request to think
615 that is has one [body].
617 *) More conditional compilation for OS/2.
619 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
621 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
623 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
624 force some output to be written in the period of inactivity. This
625 circumvents the Apache's buffering during the server-push.
627 *) Added a header-parser function placeholder for compatibility with an
630 *) Additional checks have been placed with regard to permissions of the
631 FastCGI processes. These include the existence of the file,
632 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
634 *) Implemented a restart cleanup, so that no parentlesses FastCGI
635 applications are left after Apache is restarted and/or terminated.
636 This was accomplished via two independent mechanisms, where the
637 first one prevents a start of the process manager on the first
638 reading of the configuration files. The second mechanism deals away
639 from Apache's implementation of fork() (via spawn_child()) and
640 implements its own forking, thus removing any dependencies on the
641 Apache to cleanup processes during termination phase.
643 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
644 itself in the file conf.h
646 What's New: Version 1.4.3, 15 Jan 1997
648 *) Fixed compilation warnings for various platforms, as well as
649 conditional compilation for OS/2.
651 What's New: Version 1.4.2, 12 Dec 1996
653 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
654 will proceed under this version of Apache web server.
656 *) As the result of porting, the "include virtual" construct of SSI
657 will now work correctly using either <Location> or <Directory>
660 What's New: Version 1.4.1, 4 Dec 1996
662 *) Checks have been removed from the ScanCGIHeaders that provided
663 for the presence of both Status and Location headers as being
664 an error. Contradictory to CGI/1.1 Internet Draft, both of
665 these headers are used by the current CGI applications.
667 What's New: Version 1.4, 22 Nov 1996
669 *) Added the -port option to AppClass, allowing TCP/IP communication.
670 Added the -socket option to AppClass, allowing Unix domain
671 communication via a configurable pathname.
673 *) Added the ExternalAppClass directive, allowing TCP/IP
674 communication with remote FastCGI applications.
676 *) The handler had its own code for generating HTTP response headers;
677 now it uses Apache's. This reduces the size of the module.
678 More importantly, it fixes the bug in which "include virtual"
679 sees the HTTP response headers.
681 *) The response header parser performed very little checking.
682 Now the parser enforces the guidelines in the CGI/1.1
683 Internet-Draft: Status and Location are mutually exclusive,
684 Location can only be a response to GET or HEAD,
685 CGI response headers can't be repeated, etc. (The CGI response
686 headers are the ones the handler interprets: Status, Location,
689 *) The response header parser used to miss CGI headers
690 with no whitespace after the colon, e.g. "Status:200 OK".
692 *) The response header parser sometimes interpreted the first line
693 of content as an RFC822 continuation line.
695 *) The handler implemented a nonstandard version of Location
696 which never used internal redirects. The handler
697 now attempts to implement Location as specified in the
698 CGI/1.1 Internet-Draft. The module documentation explains
701 *) Error log entries from the response header parser were
702 pretty uninformative; they are better now. If a header is
703 malformed, the log entry includes it. If the headers are
704 unterminated rather than malformed, the log entry says that,
705 and says how many bytes were received from the app.
707 *) When the application manager forked a new process, and that
708 process ran into trouble before executing the first instruction
709 of application code, the process used to exit with status =
710 errno. This made certain configuration problems (e.g. incorrect
711 file permissions when server parent is root) quite difficult
712 to pin down. Now the failing child process opens up the error log
713 and writes an informative entry before exiting.
715 *) The module now correctly handles a slash at the end of the
716 DocumentRoot directive. This was a one-line fix.
718 What's New: Version 1.3.3, 17 Oct 1996
720 *) The module now registers its request handler under the name
721 fastcgi-script in addition to the name application/x-httpd-fcgi.
722 This was a one-line addition to the module, but had quite
723 a large impact on the documentation and sample configuration.
725 What's New: Version 1.3.2, 27 Sept 1996
727 *) On some systems (SunOS, Linux?), fopen for append has a bug
728 that strikes when two processes append to the same file.
729 This bug causes the process manager to corrupt the error log.
730 Work around the bug by calling open, then fdopen. A patch
731 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
732 (Reported by Bob Ramstad.)
734 What's New: Version 1.3.1, 17 Sept 1996
736 *) On Linux and SunOS, the Apache default user_id (-1) is not a
737 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
738 Scott Langley, others.)
740 *) On some systems (Linux, some Solaris?, Irix?), connect requires
741 write access to a Unix Domain socket, so provide it. (Reported
742 by Scott Langley, freeform@wired.com, others.)
744 *) If you hit a .fcg file, but there's no AppClass defined, the
745 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
748 What's New: Version 1.3, 4 Sept 1996
750 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
752 *) Module violated the Apache buff abstraction.
753 Fixed. Might now work with SSL (not tested).
754 As part of the fix, eliminated support for
755 Apache 1.0x versions.
757 *) Module failed to chmod the Unix domain listening sockets
758 it created, so protections were set according to the current
761 *) Module forked too many process manager processes: One
762 per AppClass. When the Apache parent ran as root, these
763 process manager processes ran as root. New process
764 manager is a single process (only started if AppClass is
765 used) and runs with same privileges as other children.
766 The new process manager doesn't do any polling, so
767 there's less system overhead than before.
769 *) AppClass insisted on getting at least two arguments. Fixed.
771 *) Module used setjmp/longjmp, causing compiler warnings
772 on some platforms (e.g. Linux.) The module no longer
775 *) Module created listening sockets in /tmp, where they were
776 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
777 directive to give control over the location of listening
780 *) Module wrote error log entries without a timestamp. Fixed.
782 *) AppClass directive wrote error messages to stderr in addition
783 to returning a char * message to the Apache core. Fixed.
785 What's New: Version 1.2, 3 June 1996
787 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
790 Add version string: APACHEVERSION macro to mod_fastcgi.c
792 *) chown FastCGI socket to user_id and setuid to user_id for app
795 *) Modify GetFromStream() by having it call an OS dependent
796 function GetStreamSize(FILE *) which uses FILE internal data member.
797 Linux users might need to modify GetStreamSize().
799 What's New: Version 1.1, 10 May 1996
801 *) If you specify a non-existent executable in the AppClass directive,
802 or if the file exists and it does not have execute permission you
803 get a constant stream of error messages telling you that "program
804 terminated due to a signal".
806 *) The mod_fastcgi module should use the standard Apache error logging
807 facility instead of writing to stderr.
809 Version 1.0, 30 April 1996