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