3 *) Eliminate the need for SetHandler or AddHandler with static or
6 *) Limit PM requests to start a dynamic application
7 to 5sec to prevent endless spinning (this is a drop-dead
8 limit that should only occur if the socket/named_pipe directory
9 is removed out from under a running server).
11 *) [*nix] Change the default socket directory from /tmp/fcgi to:
13 Apache2 - RUNTIMEDIR/fastcgi
15 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
16 for use with wrappers (in lieu of finding the user/group associated
17 with a virtual host - under Apache2 this isn't accomodated).
19 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
20 apr_proc_create() changes.
22 *) Log when invoking and restoring the restart backoff policy.
24 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
25 OVERLAPPED structure).
27 *) Eliminate need for dummy files for external servers under Apache2
29 *) Fix auth compatibility mode handling for access checker and authorizer
31 *) Fix HEAD request handling. Based on a patch by
32 Chris Lightfoot [chris@ex-parrot.com]
34 *) [*nix] When autoupdate is enabled touch the socket when restarting
35 the processes to prevent further requests.
36 Eckebrecht von Pappenheim [evp@heise.de]
38 *) Apache 2.0 support.
40 *) [WIN32] Don't read from a potentially closed named pipe.
41 Philip Gladstone [philip@okena.com]
43 *) Require the Apache version 1.3.6 or later to eliminate some signal
46 *) [WIN32] Use asyncronous io with named pipes instead of polled
47 nonblocking io. This should eliminate the last of the npipe issues.
49 *) Handle an application returning a complete and valid response without
50 having consumed all of the data sent to it.
52 *) Consume remaining client data (RESPONDERs only) if any.
54 *) Add support for backing off attempts to start applications that continuously
55 fail to start. Three new macros defined in mod_fastcgi.h control this
56 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
58 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
59 applications that do not (properly) support the shutdown event (this
60 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
61 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
62 interval between signaling a proper shutdown and wacking the process(s)
63 with a TerminateProcess().
65 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
66 setting it was just plain broken.
68 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
70 *) Fix 'FastCgiConfig -autoUpdate'.
72 *) Fix 'FastCgiConfig -flush'.
74 *) Prevent silly maxProcesses and processSlack combinations.
75 Dmitry Dorofeev [dima@yasp.com]
77 *) Properly handle the killing of idle processes when one takes a long time
78 to exit once signaled down (or the config is funky).
79 Dmitry Dorofeev [dima@yasp.com]
81 *) Always kill the youngest instance of an application. Suggested by
82 Dmitry Dorofeev [dima@yasp.com]
86 *) Delay the logging of write errors to the pm to account for shutdown/restart.
88 *) (Win32) An assortment of fixes.
90 *) Fix some broken casts that were likely the cause of an assert.
92 *) Win32. Eliminate forward slashes from the named pipe path name.
93 Gerald Richter [richter@ecos.de]
95 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
96 applications it spawns. [ryans@amazon.com]
98 *) Added support for the -flush argument to FastCgiConfig.
99 Eric Sit [esit@alum.mit.edu]
101 *) Change the "which call to module_init() is this" check to a more
102 reliable approach. Doru Petrescu [pdoru@kappa.ro]
104 *) Close the old pipe file descriptor in apache main on USR1/HUP
105 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
107 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
108 [tetsuya@secom-sis.co.jp]
110 *) Eliminate the use of locks to assist in the clean shutdown of
111 applications. Instead, it is assumed that applications handle
112 termination signals properly (this is now embedded in the C
115 *) Fix Win32 process termination. Proper operation requires the use of
116 an updated application lib (termination is now signalled with an
117 Event and handled by specialized thread).
119 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
121 *) Added code so if the last instance of a dynamic application died without
122 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
123 configuration allows the last instance to be killed, then allow it to die).
124 Andrew Benham [adsb@bigfoot.com]
126 *) Fix the loadFactor calculation used to determine when dyanmic
127 applications could be killed off due to low demand [adsb@bigfoot.com].
129 *) Fix a deadlock condition that could occur with Win32 named
132 *) Fix a potential deadlock condition when FastCGI application
133 sent responses while still reading the client request (POST data).
137 *) Allow absolute pathnames in the -socket argument. Suggested by
138 Christian Jaeger [christian.jaeger@sl.ethz.ch].
140 *) Don't invoke suexec when the user/group for the fastcgi application
141 is the same as the apache main server. This is consistent with
142 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
144 *) Reset the apache drop dead timer upon successful read or writes
145 to/from the client. This eliminates timeouts that were occuring
146 during the large file transfers to/from slow clients.
148 *) Support generic wrappers such as cgiwrap by eliminating dependencies
149 on Apache's SUEXEC, renaming the FastCgiSuexec directive
150 FastCgiWrapper and eliminating any checks regarding the target
151 application (this is the repsonibility of the wrapper).
153 *) Fix a nasty bug that occurred when a client aborted a POST request
154 before the connection to a dynamic FastCGI application was opened.
155 The application's lock file descriptor wasn't setup but was being
156 closed which resulted in FD0 being closed. Normally this is open to
157 /dev/null and should pose no problems except that because the FD
158 was available it was being returned by Apache's accept(). This
159 caused it to be registered for pool cleanup. mod_cgi though moves
160 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
161 cleanup. Problem identified by checksum@163.net.
163 Changes with mod_fastcgi 2.2.8
165 *) Eliminate the concept of disabled applications. If a failure
166 occurs trying to setup an application (e.g. bind() error) its
167 tried repeatedly every init-start-delay seconds.
169 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
171 Changes with mod_fastcgi 2.2.6
173 *) Shutdown the PM when Apache appears to have disappeared.
175 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
177 *) (Win32) More dynamic fixes.
179 *) (Win32) Eliminate the per application named pipe mutex. Its now
180 per process to keep the lib happy (the use _FCGI_MUTEX_ should
181 removed from the lib). This allows pipe-based applications to
182 now handle multiple simoultanous requests (one per process).
184 *) Increase the number of open FDs we look to close when spawning apps.
186 *) Prevent an assert from popping unnecessarily.
188 *) (Win32) Add support for interpretter scripts.
190 *) (Win32) Fix named pipe handling (problems with large responses).
192 *) (Win32) Remove the "can exec" check.
194 Changes with mod_fastcgi 2.2.4
196 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
197 Rob Saccoccio [robs@chelsea.net]
199 *) Remove request type restriction (GET and POST only) in order to support
200 Web DAV requests. Scott Robertson [sroberts@codeit.com]
202 *) Allow requests for URLs such as /server/some/other/qualifier to match a
203 FastCGI server defined as /server. This was done primarily for Java
204 Servlets, but is generically useful.
206 *) Change the comm between the PM and the request handlers from a regular
207 file to a pipe and an assortment of other dynamic fixes.
209 *) Log dynamic process termination scheduling and the resulting
210 process exit notification.
212 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
213 application process from being killed off due to low demand.
215 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
217 *) Fix bugs that prevented dynamic processes from being shutdown when
218 the load subsided. Lars Heete [heete@do.isst.fhg.de]
220 *) Prevent dynamic processes from being scheduled to be started before
221 the init-start-delay or restart-delay period expires. This prevents
222 the queuing of process start requests while an application which has
223 a long initialization period starts up.
225 *) When suexec is enabled, allow the process manager to become root again in
226 order to signal applications it spawned.
228 *) Change the "server started" log message level from INFO to WARN.
230 *) Fix dynamic server "Connection Refused" handling.
232 *) Fix demand-based dynamic process spawing (uses a more portable approach).
234 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
235 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
236 longer than this period. It applies to the initiation of connections as
237 well (and thus is similar to appConnTimeout). Default is 30 seconds.
239 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
240 is more platform portable/predicable.
242 *) Leave STDOUT and STDERR open to the main server error_log. This should
243 eliminate a pile of protocol errors and having to track down 3rd party
244 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
245 be fixed in the application to use the FastCGI I/O, it just allows these
246 applications to run when previously they'd crash and burn.
248 *) Add the pass-header arg to server directives.
250 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
251 is done for CGI) - at least until PR2580 is resolved.
253 *) Initialize the default (empty) environment in a more socially acceptable
256 *) Miscellaneous doc improvements based on notes I've collected up over the
259 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
260 that resulted in not or improperly terminated strings.
262 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
263 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
265 Changes with mod_fastcgi 2.2.2
267 *) Added support for blocking connect()s by setting appConnTimeout to 0.
268 Non-blocking connect()s (the default) can be troublesome on some platforms.
269 Its expected that blocking connect()s will become the default (and
270 non-blocking connect()s will become optional) in the next release.
272 *) Dump a compile time error if the version of Apache is too old.
274 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
275 to Apache releases prior to 1.3.6.
277 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
281 Always restart a failed dynamic application if it is the last instance.
282 This means there once started there will always be at least one process
283 instance of a dynamic application.
285 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
286 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
287 the PM to start (another) application instance may help empty it faster.
289 Change two sleep() calls to select() based snoozes because alarm() is in
290 effect and sleep() and alarm() don't always play nice together.
292 Fixed a couple of error messages.
294 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
295 was ignored and the default was always used.
297 *) Allow the -initial-env argument to be used to pass variables from the
298 Apache process environment to the FastCGI server (by specifying a
299 variable name without the "=" or a value). [suggested by
300 Martin Lichtin <lichtin@oanda.com>]
302 *) Cleanup some debug macros.
304 *) Improved script stderr handling. [based on suggestions by
305 David Birnbaum <davidb@chelsea.net>]
307 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
308 updates to the INSTALL doc.
310 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
311 5 to 100. Its still configurable with the -listen-queue-depth option.
312 This should help eliminate the FAQ - Why do I see "Connection Refused"
315 *) Fix a bug in FastCgiExternalServer that broke support for external servers
316 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
318 Changes with mod_fastcgi 2.2.1
320 *) Updates to the INSTALL doc to describe building as a DSO.
322 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
323 Apache was running ("apachectl restart" does this), the contents of the
324 dynamic directory (dynamic sockets and the mbox) would be blown away
327 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
328 Apache header file) to prevent a compile error on systems without
331 *) During auth requests, the subprocess_env table was left holding variables
332 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
333 were passed to other processing phases (such as CGI/FastCGI). The
334 subprocess_env is now restored to its pre-auth condition.
336 *) Added a FastCGI Authorizer Role compatibility mode which implements the
337 specification to the tee. Use -compat arg with any of the Auth server
338 directives. This is intended for new and existing FastCGI authorizer
339 applications that require compatibility with other server implementations.
341 *) Fix logging from the process manager. Previously, all calls to the
342 logging routines used NULL for the server_rec. This works fine in a dev
343 environment, but results in most of these messages being tossed by
344 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
346 *) Always setup fr->header so that in the event the FastCGI server
347 doesn't get a valid FastCGI protocol header over the wire, we
348 can still print our error message to the log from do_work().
350 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
351 defined (Don Locrasto [locrasto@iceminer.com]).
353 Changes with mod_fastcgi 2.2.0
355 *) Serious rewrite of mod_fastcgi.html. An example conf will have
356 to wait until another time - I'm burned out on docs updates.
358 *) Add to the mod_fastcgi to the server version string.
360 *) SCO doesn't like the const in the arg to inet_addr() and
363 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
365 *) Deleted redundant header files from fcgi.h.
367 *) Full path names are no longer required for directives which take
368 file names as arguments. Like other Apache directives you can
369 specify paths from server_root (don't start with a "/").
371 *) The fcgiKillMgr is gone. This intermediate process sat between
372 the Apache parent and the FastCGI Process Manager under Apache
373 1.2. The process manager tries to name itself fcgi_pm (it used
376 *) Authentication, Authorization, and Access phases are now
377 supported. I've bent the FastCgi spec a bit: as many of the standard
378 environment variables (that were easy) are sent (the spec says don't
379 send some by name), all headers (except Status) sent by the auth
380 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
381 as environment variables rather than just those prefixed by
382 "Variable-". Custom responses for auth failures aren't supported
383 (yet.. we'll see what the demand for this is like). In addition to
384 the FastCGI protocol defined environment variable "FCGI_ROLE" being
385 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
386 set indicating which of the three phases is being processed. See
387 the docs for info on the new per-directory directives:
389 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
390 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
391 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
393 *) The code has been broken into logical chunks making figuring it
394 out much easier. All of the #defines a user may want to change are
395 now in mod_fastcgi.h.
397 *) All the logging has been converted to the "supported" Apache
398 logging routines, ap_log_error() and ap_log_rerror(). Log entries
399 for request specific errors should now be directed to the correct
400 server errorlog (if your running more than one). Some attempt has
401 been made at setting appropriate log levels (comments of course are
402 welcome) and printing errno info when its of value. All of log
403 messages now have "FastCGI" in them (nice for grep).
405 *) All of the calls to Apache routines have been updated to the ap_
406 convention, eliminating the need for the compat header.
408 *) By default we no longer flush every piece of data we get from the
409 FastCGI server to the client. This allows the FastCGI server to
410 release without having to wait for the client flush to complete.
411 The old behaviour can revived with the -flush argument to AppClass.
413 *) I renamed some of the directives for consistency (the old names
416 AppClass -> FastCgiServer,
417 FCGIConfig -> FastCgiConfig,
418 ExternalAppClass -> FastCgiExternalServer
420 *) Directive arguments are no longer case sensitive.
422 *) The module now uses hard_timeout() rather than soft_timeout. This
423 means that the module will longjump out of the request if the
424 drop-dead timeout expires (set with Apache's Timeout directive, the
425 default is 5min) or if the client closes the connection (SIGPIPE).
426 This is more typical for a Apache modules and I'm not convinced we
427 properly handle all of the error cases well enough to use
428 soft_timeout (I can think of one place we don't anyway). This means
429 your FastCGI application can see SIGPIPE.
431 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
432 Apache. Without this the default behaviour is exit().
434 *) It should (I think) run now under DSO now.
436 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
438 *) The broken -initial-env bug ins 2.1b1 is fixed.
440 *) Maybe some other stuff.
444 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
445 commit notes and stick 'em in here.. when I get some free time. ;)
450 *** Originally from docs/README..
452 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
453 From: Stanley Gambarin <stanleyg@cs.bu.edu>
455 *) Yet more changes to the source distribution. Separated out
456 the Tcl dynamic string and buffer libraries into separate files.
457 Did the same for the OS library, however, it is not really an
458 abstraction... more like a bunch of wrappers. Added to DEVNOTES
459 a note about the problem of Apache not allowing including header
462 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
463 From: Stanley Gambarin <stanleyg@cs.bu.edu>
465 *) More changes to the installation scripts. Theoretically we
466 should be able to to use the same Makefile/installation script
467 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
468 layer to use the same source code for both 1.2.x and 1.3.x sources.
470 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
471 From: Stanley Gambarin <stanleyg@cs.bu.edu>
473 *) Created Makefile to compile the module into the archive library,
474 just like the proxy module. Also initial draft of the installation
475 script, which should work for both 1.2 and 1.3 sources of Apache.
477 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
478 From: Stanley Gambarin <stanleyg@cs.bu.edu>
480 *) More source reorg. Separated out some header files. Need much
481 more work on this. Updated the TODO file. Added fcgivers.h and
482 CHANGES file to track the history.
484 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
485 From: Stanley Gambarin <stanleyg@cs.bu.edu>
487 *) Source reorganization. This is done to provide a basis for later
488 functionality implementation. See TODO file for more information
489 of what possible future enhancements are possible/desired.
491 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
492 From: David MacKenzie <djm@va.pubnix.com>
494 *) Create the "dynamic" dir and "mbox" as the user specified in the
495 User and Group directives instead of as root. They are created with
496 restrictive permissions, and then the module checks to see whether the
497 user specified in the User and Group directives has read, write, and
498 execute permissions on them. Those permissions had been explicitly
499 denied by creating them as "root".
501 *) Some documentation was garbled or incomplete.
503 *) When a FastCGI application can't be execl'd, the code used a
504 value of errno that may have been stomped on by intervening system
505 calls. That happened on BSD/OS 2.1, where the error_log was
506 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
507 An intervening stdio call was setting errno as a side-effect, which
508 it has the right to do.
510 *) Erroneous fprintf arguments, a missing return value, an unused
511 function, missing declarations, and other problems detected by gcc
514 *) Duplicated code to create the lock file name merged into a single
515 function, closing several memory leaks.
517 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
518 FCGI app in a given dynamic class would live forever, even if it
519 hadn't been requested for weeks.
521 *) The test for whether to keep looking for dynamic app victims to
522 kill was backwards, so no dynamic apps were ever selected as
525 *) The killInterval and updateInterval were being ignored; the
526 sigsuspend() forced recalculations only when a child died,
527 instead of at the intervals specified.
529 *) When a dynamic app couldn't be started, the program SEGV'd when
530 trying to free the ipcAddrPtr twice.
532 *) Remove the dead lock file and socket when cleaning up after a
533 server whose last child has been killed, so the FastCGIHandler
534 isn't fooled into thinking there is still a process serving
537 *) Make file locking robust in the presence of signals.
539 *) Rename some badly named variables.
541 *) Fix typos in many comments.
543 *) Fix some memory and file descriptor leaks.
545 *) Make the blocking kill of a server closer to working.
547 *) More fixes to calculations in dynamic application management.
550 *** Originally from docs/README.OMI..
552 What's New: Version 2.0b1, 16 Apr 1997
554 *) Implemented a mechanism by which FastCGI applications are started by
555 the web server on the first request and continue running. Also, a
556 heuristics have been implemented to allow for dynamic killing of the
557 running FastCGI apps. The configuration options are supplied via
558 the new FCGIConfig directive, that is described in the
561 *) When performing internal redirect, since the original request's body
562 has already been read, do not allow the redirected request to think
563 that is has one [body].
565 *) More conditional compilation for OS/2.
567 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
569 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
571 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
572 force some output to be written in the period of inactivity. This
573 circumvents the Apache's buffering during the server-push.
575 *) Added a header-parser function placeholder for compatibility with an
578 *) Additional checks have been placed with regard to permissions of the
579 FastCGI processes. These include the existence of the file,
580 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
582 *) Implemented a restart cleanup, so that no parentlesses FastCGI
583 applications are left after Apache is restarted and/or terminated.
584 This was accomplished via two independent mechanisms, where the
585 first one prevents a start of the process manager on the first
586 reading of the configuration files. The second mechanism deals away
587 from Apache's implementation of fork() (via spawn_child()) and
588 implements its own forking, thus removing any dependencies on the
589 Apache to cleanup processes during termination phase.
591 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
592 itself in the file conf.h
594 What's New: Version 1.4.3, 15 Jan 1997
596 *) Fixed compilation warnings for various platforms, as well as
597 conditional compilation for OS/2.
599 What's New: Version 1.4.2, 12 Dec 1996
601 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
602 will proceed under this version of Apache web server.
604 *) As the result of porting, the "include virtual" construct of SSI
605 will now work correctly using either <Location> or <Directory>
608 What's New: Version 1.4.1, 4 Dec 1996
610 *) Checks have been removed from the ScanCGIHeaders that provided
611 for the presence of both Status and Location headers as being
612 an error. Contradictory to CGI/1.1 Internet Draft, both of
613 these headers are used by the current CGI applications.
615 What's New: Version 1.4, 22 Nov 1996
617 *) Added the -port option to AppClass, allowing TCP/IP communication.
618 Added the -socket option to AppClass, allowing Unix domain
619 communication via a configurable pathname.
621 *) Added the ExternalAppClass directive, allowing TCP/IP
622 communication with remote FastCGI applications.
624 *) The handler had its own code for generating HTTP response headers;
625 now it uses Apache's. This reduces the size of the module.
626 More importantly, it fixes the bug in which "include virtual"
627 sees the HTTP response headers.
629 *) The response header parser performed very little checking.
630 Now the parser enforces the guidelines in the CGI/1.1
631 Internet-Draft: Status and Location are mutually exclusive,
632 Location can only be a response to GET or HEAD,
633 CGI response headers can't be repeated, etc. (The CGI response
634 headers are the ones the handler interprets: Status, Location,
637 *) The response header parser used to miss CGI headers
638 with no whitespace after the colon, e.g. "Status:200 OK".
640 *) The response header parser sometimes interpreted the first line
641 of content as an RFC822 continuation line.
643 *) The handler implemented a nonstandard version of Location
644 which never used internal redirects. The handler
645 now attempts to implement Location as specified in the
646 CGI/1.1 Internet-Draft. The module documentation explains
649 *) Error log entries from the response header parser were
650 pretty uninformative; they are better now. If a header is
651 malformed, the log entry includes it. If the headers are
652 unterminated rather than malformed, the log entry says that,
653 and says how many bytes were received from the app.
655 *) When the application manager forked a new process, and that
656 process ran into trouble before executing the first instruction
657 of application code, the process used to exit with status =
658 errno. This made certain configuration problems (e.g. incorrect
659 file permissions when server parent is root) quite difficult
660 to pin down. Now the failing child process opens up the error log
661 and writes an informative entry before exiting.
663 *) The module now correctly handles a slash at the end of the
664 DocumentRoot directive. This was a one-line fix.
666 What's New: Version 1.3.3, 17 Oct 1996
668 *) The module now registers its request handler under the name
669 fastcgi-script in addition to the name application/x-httpd-fcgi.
670 This was a one-line addition to the module, but had quite
671 a large impact on the documentation and sample configuration.
673 What's New: Version 1.3.2, 27 Sept 1996
675 *) On some systems (SunOS, Linux?), fopen for append has a bug
676 that strikes when two processes append to the same file.
677 This bug causes the process manager to corrupt the error log.
678 Work around the bug by calling open, then fdopen. A patch
679 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
680 (Reported by Bob Ramstad.)
682 What's New: Version 1.3.1, 17 Sept 1996
684 *) On Linux and SunOS, the Apache default user_id (-1) is not a
685 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
686 Scott Langley, others.)
688 *) On some systems (Linux, some Solaris?, Irix?), connect requires
689 write access to a Unix Domain socket, so provide it. (Reported
690 by Scott Langley, freeform@wired.com, others.)
692 *) If you hit a .fcg file, but there's no AppClass defined, the
693 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
696 What's New: Version 1.3, 4 Sept 1996
698 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
700 *) Module violated the Apache buff abstraction.
701 Fixed. Might now work with SSL (not tested).
702 As part of the fix, eliminated support for
703 Apache 1.0x versions.
705 *) Module failed to chmod the Unix domain listening sockets
706 it created, so protections were set according to the current
709 *) Module forked too many process manager processes: One
710 per AppClass. When the Apache parent ran as root, these
711 process manager processes ran as root. New process
712 manager is a single process (only started if AppClass is
713 used) and runs with same privileges as other children.
714 The new process manager doesn't do any polling, so
715 there's less system overhead than before.
717 *) AppClass insisted on getting at least two arguments. Fixed.
719 *) Module used setjmp/longjmp, causing compiler warnings
720 on some platforms (e.g. Linux.) The module no longer
723 *) Module created listening sockets in /tmp, where they were
724 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
725 directive to give control over the location of listening
728 *) Module wrote error log entries without a timestamp. Fixed.
730 *) AppClass directive wrote error messages to stderr in addition
731 to returning a char * message to the Apache core. Fixed.
733 What's New: Version 1.2, 3 June 1996
735 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
738 Add version string: APACHEVERSION macro to mod_fastcgi.c
740 *) chown FastCGI socket to user_id and setuid to user_id for app
743 *) Modify GetFromStream() by having it call an OS dependent
744 function GetStreamSize(FILE *) which uses FILE internal data member.
745 Linux users might need to modify GetStreamSize().
747 What's New: Version 1.1, 10 May 1996
749 *) If you specify a non-existent executable in the AppClass directive,
750 or if the file exists and it does not have execute permission you
751 get a constant stream of error messages telling you that "program
752 terminated due to a signal".
754 *) The mod_fastcgi module should use the standard Apache error logging
755 facility instead of writing to stderr.
757 Version 1.0, 30 April 1996