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