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