3 *) Eliminate need for dummy files for external servers under Apache2
\r
5 *) Fix auth compatibility mode handling for access checker and authorizer
7 *) Fix HEAD request handling. Based on a patch by
8 Chris Lightfoot [chris@ex-parrot.com]
10 *) [*nix] When autoupdate is enabled touch the socket when restarting
11 the processes to prevent further requests.
12 Eckebrecht von Pappenheim [evp@heise.de]
14 *) Apache 2.0 support.
16 *) [WIN32] Don't read from a potentially closed named pipe.
17 Philip Gladstone [philip@okena.com]
19 *) Require the Apache version 1.3.6 or later to eliminate some signal
22 *) [WIN32] Use asyncronous io with named pipes instead of polled
23 nonblocking io. This should eliminate the last of the npipe issues.
25 *) Handle an application returning a complete and valid response without
26 having consumed all of the data sent to it.
28 *) Consume remaining client data (RESPONDERs only) if any.
30 *) Add support for backing off attempts to start applications that continuously
31 fail to start. Three new macros defined in mod_fastcgi.h control this
32 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
34 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
35 applications that do not (properly) support the shutdown event (this
36 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
37 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
38 interval between signaling a proper shutdown and wacking the process(s)
39 with a TerminateProcess().
41 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
42 setting it was just plain broken.
44 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
46 *) Fix 'FastCgiConfig -autoUpdate'.
48 *) Fix 'FastCgiConfig -flush'.
50 *) Prevent silly maxProcesses and processSlack combinations.
51 Dmitry Dorofeev [dima@yasp.com]
53 *) Properly handle the killing of idle processes when one takes a long time
54 to exit once signaled down (or the config is funky).
55 Dmitry Dorofeev [dima@yasp.com]
57 *) Always kill the youngest instance of an application. Suggested by
58 Dmitry Dorofeev [dima@yasp.com]
62 *) Delay the logging of write errors to the pm to account for shutdown/restart.
64 *) (Win32) An assortment of fixes.
66 *) Fix some broken casts that were likely the cause of an assert.
68 *) Win32. Eliminate forward slashes from the named pipe path name.
69 Gerald Richter [richter@ecos.de]
71 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
72 applications it spawns. [ryans@amazon.com]
74 *) Added support for the -flush argument to FastCgiConfig.
75 Eric Sit [esit@alum.mit.edu]
77 *) Change the "which call to module_init() is this" check to a more
78 reliable approach. Doru Petrescu [pdoru@kappa.ro]
80 *) Close the old pipe file descriptor in apache main on USR1/HUP
81 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
83 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
84 [tetsuya@secom-sis.co.jp]
86 *) Eliminate the use of locks to assist in the clean shutdown of
87 applications. Instead, it is assumed that applications handle
88 termination signals properly (this is now embedded in the C
91 *) Fix Win32 process termination. Proper operation requires the use of
92 an updated application lib (termination is now signalled with an
93 Event and handled by specialized thread).
95 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
97 *) Added code so if the last instance of a dynamic application died without
98 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
99 configuration allows the last instance to be killed, then allow it to die).
100 Andrew Benham [adsb@bigfoot.com]
102 *) Fix the loadFactor calculation used to determine when dyanmic
103 applications could be killed off due to low demand [adsb@bigfoot.com].
105 *) Fix a deadlock condition that could occur with Win32 named
108 *) Fix a potential deadlock condition when FastCGI application
109 sent responses while still reading the client request (POST data).
113 *) Allow absolute pathnames in the -socket argument. Suggested by
114 Christian Jaeger [christian.jaeger@sl.ethz.ch].
116 *) Don't invoke suexec when the user/group for the fastcgi application
117 is the same as the apache main server. This is consistent with
118 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
120 *) Reset the apache drop dead timer upon successful read or writes
121 to/from the client. This eliminates timeouts that were occuring
122 during the large file transfers to/from slow clients.
124 *) Support generic wrappers such as cgiwrap by eliminating dependencies
125 on Apache's SUEXEC, renaming the FastCgiSuexec directive
126 FastCgiWrapper and eliminating any checks regarding the target
127 application (this is the repsonibility of the wrapper).
129 *) Fix a nasty bug that occurred when a client aborted a POST request
130 before the connection to a dynamic FastCGI application was opened.
131 The application's lock file descriptor wasn't setup but was being
132 closed which resulted in FD0 being closed. Normally this is open to
133 /dev/null and should pose no problems except that because the FD
134 was available it was being returned by Apache's accept(). This
135 caused it to be registered for pool cleanup. mod_cgi though moves
136 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
137 cleanup. Problem identified by checksum@163.net.
139 Changes with mod_fastcgi 2.2.8
141 *) Eliminate the concept of disabled applications. If a failure
142 occurs trying to setup an application (e.g. bind() error) its
143 tried repeatedly every init-start-delay seconds.
145 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
147 Changes with mod_fastcgi 2.2.6
149 *) Shutdown the PM when Apache appears to have disappeared.
151 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
153 *) (Win32) More dynamic fixes.
155 *) (Win32) Eliminate the per application named pipe mutex. Its now
156 per process to keep the lib happy (the use _FCGI_MUTEX_ should
157 removed from the lib). This allows pipe-based applications to
158 now handle multiple simoultanous requests (one per process).
160 *) Increase the number of open FDs we look to close when spawning apps.
162 *) Prevent an assert from popping unnecessarily.
164 *) (Win32) Add support for interpretter scripts.
166 *) (Win32) Fix named pipe handling (problems with large responses).
168 *) (Win32) Remove the "can exec" check.
170 Changes with mod_fastcgi 2.2.4
172 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
173 Rob Saccoccio [robs@chelsea.net]
175 *) Remove request type restriction (GET and POST only) in order to support
176 Web DAV requests. Scott Robertson [sroberts@codeit.com]
178 *) Allow requests for URLs such as /server/some/other/qualifier to match a
179 FastCGI server defined as /server. This was done primarily for Java
180 Servlets, but is generically useful.
182 *) Change the comm between the PM and the request handlers from a regular
183 file to a pipe and an assortment of other dynamic fixes.
185 *) Log dynamic process termination scheduling and the resulting
186 process exit notification.
188 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
189 application process from being killed off due to low demand.
191 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
193 *) Fix bugs that prevented dynamic processes from being shutdown when
194 the load subsided. Lars Heete [heete@do.isst.fhg.de]
196 *) Prevent dynamic processes from being scheduled to be started before
197 the init-start-delay or restart-delay period expires. This prevents
198 the queuing of process start requests while an application which has
199 a long initialization period starts up.
201 *) When suexec is enabled, allow the process manager to become root again in
202 order to signal applications it spawned.
204 *) Change the "server started" log message level from INFO to WARN.
206 *) Fix dynamic server "Connection Refused" handling.
208 *) Fix demand-based dynamic process spawing (uses a more portable approach).
210 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
211 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
212 longer than this period. It applies to the initiation of connections as
213 well (and thus is similar to appConnTimeout). Default is 30 seconds.
215 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
216 is more platform portable/predicable.
218 *) Leave STDOUT and STDERR open to the main server error_log. This should
219 eliminate a pile of protocol errors and having to track down 3rd party
220 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
221 be fixed in the application to use the FastCGI I/O, it just allows these
222 applications to run when previously they'd crash and burn.
224 *) Add the pass-header arg to server directives.
226 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
227 is done for CGI) - at least until PR2580 is resolved.
229 *) Initialize the default (empty) environment in a more socially acceptable
232 *) Miscellaneous doc improvements based on notes I've collected up over the
235 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
236 that resulted in not or improperly terminated strings.
238 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
239 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
241 Changes with mod_fastcgi 2.2.2
243 *) Added support for blocking connect()s by setting appConnTimeout to 0.
244 Non-blocking connect()s (the default) can be troublesome on some platforms.
245 Its expected that blocking connect()s will become the default (and
246 non-blocking connect()s will become optional) in the next release.
248 *) Dump a compile time error if the version of Apache is too old.
250 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
251 to Apache releases prior to 1.3.6.
253 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
257 Always restart a failed dynamic application if it is the last instance.
258 This means there once started there will always be at least one process
259 instance of a dynamic application.
261 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
262 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
263 the PM to start (another) application instance may help empty it faster.
265 Change two sleep() calls to select() based snoozes because alarm() is in
266 effect and sleep() and alarm() don't always play nice together.
268 Fixed a couple of error messages.
270 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
271 was ignored and the default was always used.
273 *) Allow the -initial-env argument to be used to pass variables from the
274 Apache process environment to the FastCGI server (by specifying a
275 variable name without the "=" or a value). [suggested by
276 Martin Lichtin <lichtin@oanda.com>]
278 *) Cleanup some debug macros.
280 *) Improved script stderr handling. [based on suggestions by
281 David Birnbaum <davidb@chelsea.net>]
283 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
284 updates to the INSTALL doc.
286 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
287 5 to 100. Its still configurable with the -listen-queue-depth option.
288 This should help eliminate the FAQ - Why do I see "Connection Refused"
291 *) Fix a bug in FastCgiExternalServer that broke support for external servers
292 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
294 Changes with mod_fastcgi 2.2.1
296 *) Updates to the INSTALL doc to describe building as a DSO.
298 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
299 Apache was running ("apachectl restart" does this), the contents of the
300 dynamic directory (dynamic sockets and the mbox) would be blown away
303 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
304 Apache header file) to prevent a compile error on systems without
307 *) During auth requests, the subprocess_env table was left holding variables
308 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
309 were passed to other processing phases (such as CGI/FastCGI). The
310 subprocess_env is now restored to its pre-auth condition.
312 *) Added a FastCGI Authorizer Role compatibility mode which implements the
313 specification to the tee. Use -compat arg with any of the Auth server
314 directives. This is intended for new and existing FastCGI authorizer
315 applications that require compatibility with other server implementations.
317 *) Fix logging from the process manager. Previously, all calls to the
318 logging routines used NULL for the server_rec. This works fine in a dev
319 environment, but results in most of these messages being tossed by
320 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
322 *) Always setup fr->header so that in the event the FastCGI server
323 doesn't get a valid FastCGI protocol header over the wire, we
324 can still print our error message to the log from do_work().
326 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
327 defined (Don Locrasto [locrasto@iceminer.com]).
329 Changes with mod_fastcgi 2.2.0
331 *) Serious rewrite of mod_fastcgi.html. An example conf will have
332 to wait until another time - I'm burned out on docs updates.
334 *) Add to the mod_fastcgi to the server version string.
336 *) SCO doesn't like the const in the arg to inet_addr() and
339 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
341 *) Deleted redundant header files from fcgi.h.
343 *) Full path names are no longer required for directives which take
344 file names as arguments. Like other Apache directives you can
345 specify paths from server_root (don't start with a "/").
347 *) The fcgiKillMgr is gone. This intermediate process sat between
348 the Apache parent and the FastCGI Process Manager under Apache
349 1.2. The process manager tries to name itself fcgi_pm (it used
352 *) Authentication, Authorization, and Access phases are now
353 supported. I've bent the FastCgi spec a bit: as many of the standard
354 environment variables (that were easy) are sent (the spec says don't
355 send some by name), all headers (except Status) sent by the auth
356 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
357 as environment variables rather than just those prefixed by
358 "Variable-". Custom responses for auth failures aren't supported
359 (yet.. we'll see what the demand for this is like). In addition to
360 the FastCGI protocol defined environment variable "FCGI_ROLE" being
361 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
362 set indicating which of the three phases is being processed. See
363 the docs for info on the new per-directory directives:
365 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
366 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
367 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
369 *) The code has been broken into logical chunks making figuring it
370 out much easier. All of the #defines a user may want to change are
371 now in mod_fastcgi.h.
373 *) All the logging has been converted to the "supported" Apache
374 logging routines, ap_log_error() and ap_log_rerror(). Log entries
375 for request specific errors should now be directed to the correct
376 server errorlog (if your running more than one). Some attempt has
377 been made at setting appropriate log levels (comments of course are
378 welcome) and printing errno info when its of value. All of log
379 messages now have "FastCGI" in them (nice for grep).
381 *) All of the calls to Apache routines have been updated to the ap_
382 convention, eliminating the need for the compat header.
384 *) By default we no longer flush every piece of data we get from the
385 FastCGI server to the client. This allows the FastCGI server to
386 release without having to wait for the client flush to complete.
387 The old behaviour can revived with the -flush argument to AppClass.
389 *) I renamed some of the directives for consistency (the old names
392 AppClass -> FastCgiServer,
393 FCGIConfig -> FastCgiConfig,
394 ExternalAppClass -> FastCgiExternalServer
396 *) Directive arguments are no longer case sensitive.
398 *) The module now uses hard_timeout() rather than soft_timeout. This
399 means that the module will longjump out of the request if the
400 drop-dead timeout expires (set with Apache's Timeout directive, the
401 default is 5min) or if the client closes the connection (SIGPIPE).
402 This is more typical for a Apache modules and I'm not convinced we
403 properly handle all of the error cases well enough to use
404 soft_timeout (I can think of one place we don't anyway). This means
405 your FastCGI application can see SIGPIPE.
407 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
408 Apache. Without this the default behaviour is exit().
410 *) It should (I think) run now under DSO now.
412 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
414 *) The broken -initial-env bug ins 2.1b1 is fixed.
416 *) Maybe some other stuff.
420 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
421 commit notes and stick 'em in here.. when I get some free time. ;)
426 *** Originally from docs/README..
428 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
429 From: Stanley Gambarin <stanleyg@cs.bu.edu>
431 *) Yet more changes to the source distribution. Separated out
432 the Tcl dynamic string and buffer libraries into separate files.
433 Did the same for the OS library, however, it is not really an
434 abstraction... more like a bunch of wrappers. Added to DEVNOTES
435 a note about the problem of Apache not allowing including header
438 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
439 From: Stanley Gambarin <stanleyg@cs.bu.edu>
441 *) More changes to the installation scripts. Theoretically we
442 should be able to to use the same Makefile/installation script
443 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
444 layer to use the same source code for both 1.2.x and 1.3.x sources.
446 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
447 From: Stanley Gambarin <stanleyg@cs.bu.edu>
449 *) Created Makefile to compile the module into the archive library,
450 just like the proxy module. Also initial draft of the installation
451 script, which should work for both 1.2 and 1.3 sources of Apache.
453 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
454 From: Stanley Gambarin <stanleyg@cs.bu.edu>
456 *) More source reorg. Separated out some header files. Need much
457 more work on this. Updated the TODO file. Added fcgivers.h and
458 CHANGES file to track the history.
460 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
461 From: Stanley Gambarin <stanleyg@cs.bu.edu>
463 *) Source reorganization. This is done to provide a basis for later
464 functionality implementation. See TODO file for more information
465 of what possible future enhancements are possible/desired.
467 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
468 From: David MacKenzie <djm@va.pubnix.com>
470 *) Create the "dynamic" dir and "mbox" as the user specified in the
471 User and Group directives instead of as root. They are created with
472 restrictive permissions, and then the module checks to see whether the
473 user specified in the User and Group directives has read, write, and
474 execute permissions on them. Those permissions had been explicitly
475 denied by creating them as "root".
477 *) Some documentation was garbled or incomplete.
479 *) When a FastCGI application can't be execl'd, the code used a
480 value of errno that may have been stomped on by intervening system
481 calls. That happened on BSD/OS 2.1, where the error_log was
482 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
483 An intervening stdio call was setting errno as a side-effect, which
484 it has the right to do.
486 *) Erroneous fprintf arguments, a missing return value, an unused
487 function, missing declarations, and other problems detected by gcc
490 *) Duplicated code to create the lock file name merged into a single
491 function, closing several memory leaks.
493 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
494 FCGI app in a given dynamic class would live forever, even if it
495 hadn't been requested for weeks.
497 *) The test for whether to keep looking for dynamic app victims to
498 kill was backwards, so no dynamic apps were ever selected as
501 *) The killInterval and updateInterval were being ignored; the
502 sigsuspend() forced recalculations only when a child died,
503 instead of at the intervals specified.
505 *) When a dynamic app couldn't be started, the program SEGV'd when
506 trying to free the ipcAddrPtr twice.
508 *) Remove the dead lock file and socket when cleaning up after a
509 server whose last child has been killed, so the FastCGIHandler
510 isn't fooled into thinking there is still a process serving
513 *) Make file locking robust in the presence of signals.
515 *) Rename some badly named variables.
517 *) Fix typos in many comments.
519 *) Fix some memory and file descriptor leaks.
521 *) Make the blocking kill of a server closer to working.
523 *) More fixes to calculations in dynamic application management.
526 *** Originally from docs/README.OMI..
528 What's New: Version 2.0b1, 16 Apr 1997
530 *) Implemented a mechanism by which FastCGI applications are started by
531 the web server on the first request and continue running. Also, a
532 heuristics have been implemented to allow for dynamic killing of the
533 running FastCGI apps. The configuration options are supplied via
534 the new FCGIConfig directive, that is described in the
537 *) When performing internal redirect, since the original request's body
538 has already been read, do not allow the redirected request to think
539 that is has one [body].
541 *) More conditional compilation for OS/2.
543 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
545 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
547 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
548 force some output to be written in the period of inactivity. This
549 circumvents the Apache's buffering during the server-push.
551 *) Added a header-parser function placeholder for compatibility with an
554 *) Additional checks have been placed with regard to permissions of the
555 FastCGI processes. These include the existence of the file,
556 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
558 *) Implemented a restart cleanup, so that no parentlesses FastCGI
559 applications are left after Apache is restarted and/or terminated.
560 This was accomplished via two independent mechanisms, where the
561 first one prevents a start of the process manager on the first
562 reading of the configuration files. The second mechanism deals away
563 from Apache's implementation of fork() (via spawn_child()) and
564 implements its own forking, thus removing any dependencies on the
565 Apache to cleanup processes during termination phase.
567 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
568 itself in the file conf.h
570 What's New: Version 1.4.3, 15 Jan 1997
572 *) Fixed compilation warnings for various platforms, as well as
573 conditional compilation for OS/2.
575 What's New: Version 1.4.2, 12 Dec 1996
577 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
578 will proceed under this version of Apache web server.
580 *) As the result of porting, the "include virtual" construct of SSI
581 will now work correctly using either <Location> or <Directory>
584 What's New: Version 1.4.1, 4 Dec 1996
586 *) Checks have been removed from the ScanCGIHeaders that provided
587 for the presence of both Status and Location headers as being
588 an error. Contradictory to CGI/1.1 Internet Draft, both of
589 these headers are used by the current CGI applications.
591 What's New: Version 1.4, 22 Nov 1996
593 *) Added the -port option to AppClass, allowing TCP/IP communication.
594 Added the -socket option to AppClass, allowing Unix domain
595 communication via a configurable pathname.
597 *) Added the ExternalAppClass directive, allowing TCP/IP
598 communication with remote FastCGI applications.
600 *) The handler had its own code for generating HTTP response headers;
601 now it uses Apache's. This reduces the size of the module.
602 More importantly, it fixes the bug in which "include virtual"
603 sees the HTTP response headers.
605 *) The response header parser performed very little checking.
606 Now the parser enforces the guidelines in the CGI/1.1
607 Internet-Draft: Status and Location are mutually exclusive,
608 Location can only be a response to GET or HEAD,
609 CGI response headers can't be repeated, etc. (The CGI response
610 headers are the ones the handler interprets: Status, Location,
613 *) The response header parser used to miss CGI headers
614 with no whitespace after the colon, e.g. "Status:200 OK".
616 *) The response header parser sometimes interpreted the first line
617 of content as an RFC822 continuation line.
619 *) The handler implemented a nonstandard version of Location
620 which never used internal redirects. The handler
621 now attempts to implement Location as specified in the
622 CGI/1.1 Internet-Draft. The module documentation explains
625 *) Error log entries from the response header parser were
626 pretty uninformative; they are better now. If a header is
627 malformed, the log entry includes it. If the headers are
628 unterminated rather than malformed, the log entry says that,
629 and says how many bytes were received from the app.
631 *) When the application manager forked a new process, and that
632 process ran into trouble before executing the first instruction
633 of application code, the process used to exit with status =
634 errno. This made certain configuration problems (e.g. incorrect
635 file permissions when server parent is root) quite difficult
636 to pin down. Now the failing child process opens up the error log
637 and writes an informative entry before exiting.
639 *) The module now correctly handles a slash at the end of the
640 DocumentRoot directive. This was a one-line fix.
642 What's New: Version 1.3.3, 17 Oct 1996
644 *) The module now registers its request handler under the name
645 fastcgi-script in addition to the name application/x-httpd-fcgi.
646 This was a one-line addition to the module, but had quite
647 a large impact on the documentation and sample configuration.
649 What's New: Version 1.3.2, 27 Sept 1996
651 *) On some systems (SunOS, Linux?), fopen for append has a bug
652 that strikes when two processes append to the same file.
653 This bug causes the process manager to corrupt the error log.
654 Work around the bug by calling open, then fdopen. A patch
655 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
656 (Reported by Bob Ramstad.)
658 What's New: Version 1.3.1, 17 Sept 1996
660 *) On Linux and SunOS, the Apache default user_id (-1) is not a
661 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
662 Scott Langley, others.)
664 *) On some systems (Linux, some Solaris?, Irix?), connect requires
665 write access to a Unix Domain socket, so provide it. (Reported
666 by Scott Langley, freeform@wired.com, others.)
668 *) If you hit a .fcg file, but there's no AppClass defined, the
669 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
672 What's New: Version 1.3, 4 Sept 1996
674 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
676 *) Module violated the Apache buff abstraction.
677 Fixed. Might now work with SSL (not tested).
678 As part of the fix, eliminated support for
679 Apache 1.0x versions.
681 *) Module failed to chmod the Unix domain listening sockets
682 it created, so protections were set according to the current
685 *) Module forked too many process manager processes: One
686 per AppClass. When the Apache parent ran as root, these
687 process manager processes ran as root. New process
688 manager is a single process (only started if AppClass is
689 used) and runs with same privileges as other children.
690 The new process manager doesn't do any polling, so
691 there's less system overhead than before.
693 *) AppClass insisted on getting at least two arguments. Fixed.
695 *) Module used setjmp/longjmp, causing compiler warnings
696 on some platforms (e.g. Linux.) The module no longer
699 *) Module created listening sockets in /tmp, where they were
700 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
701 directive to give control over the location of listening
704 *) Module wrote error log entries without a timestamp. Fixed.
706 *) AppClass directive wrote error messages to stderr in addition
707 to returning a char * message to the Apache core. Fixed.
709 What's New: Version 1.2, 3 June 1996
711 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
714 Add version string: APACHEVERSION macro to mod_fastcgi.c
716 *) chown FastCGI socket to user_id and setuid to user_id for app
719 *) Modify GetFromStream() by having it call an OS dependent
720 function GetStreamSize(FILE *) which uses FILE internal data member.
721 Linux users might need to modify GetStreamSize().
723 What's New: Version 1.1, 10 May 1996
725 *) If you specify a non-existent executable in the AppClass directive,
726 or if the file exists and it does not have execute permission you
727 get a constant stream of error messages telling you that "program
728 terminated due to a signal".
730 *) The mod_fastcgi module should use the standard Apache error logging
731 facility instead of writing to stderr.
733 Version 1.0, 30 April 1996