3 *) [*nix/AP2] Prevent the module from being initalized twice at startup
4 (resulting in confusing error messages to the log).
6 *) Eliminate the need for SetHandler or AddHandler with static or
9 *) Limit PM requests to start a dynamic application
10 to 5sec to prevent endless spinning (this is a drop-dead
11 limit that should only occur if the socket/named_pipe directory
12 is removed out from under a running server).
14 *) [*nix] Change the default socket directory from /tmp/fcgi to:
16 Apache2 - RUNTIMEDIR/fastcgi
18 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
19 for use with wrappers (in lieu of finding the user/group associated
20 with a virtual host - under Apache2 this isn't accomodated).
22 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
23 apr_proc_create() changes.
25 *) Log when invoking and restoring the restart backoff policy.
27 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
28 OVERLAPPED structure).
30 *) Eliminate need for dummy files for external servers under Apache2
32 *) Fix auth compatibility mode handling for access checker and authorizer
34 *) Fix HEAD request handling. Based on a patch by
35 Chris Lightfoot [chris@ex-parrot.com]
37 *) [*nix] When autoupdate is enabled touch the socket when restarting
38 the processes to prevent further requests.
39 Eckebrecht von Pappenheim [evp@heise.de]
41 *) Apache 2.0 support.
43 *) [WIN32] Don't read from a potentially closed named pipe.
44 Philip Gladstone [philip@okena.com]
46 *) Require the Apache version 1.3.6 or later to eliminate some signal
49 *) [WIN32] Use asyncronous io with named pipes instead of polled
50 nonblocking io. This should eliminate the last of the npipe issues.
52 *) Handle an application returning a complete and valid response without
53 having consumed all of the data sent to it.
55 *) Consume remaining client data (RESPONDERs only) if any.
57 *) Add support for backing off attempts to start applications that continuously
58 fail to start. Three new macros defined in mod_fastcgi.h control this
59 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
61 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
62 applications that do not (properly) support the shutdown event (this
63 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
64 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
65 interval between signaling a proper shutdown and wacking the process(s)
66 with a TerminateProcess().
68 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
69 setting it was just plain broken.
71 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
73 *) Fix 'FastCgiConfig -autoUpdate'.
75 *) Fix 'FastCgiConfig -flush'.
77 *) Prevent silly maxProcesses and processSlack combinations.
78 Dmitry Dorofeev [dima@yasp.com]
80 *) Properly handle the killing of idle processes when one takes a long time
81 to exit once signaled down (or the config is funky).
82 Dmitry Dorofeev [dima@yasp.com]
84 *) Always kill the youngest instance of an application. Suggested by
85 Dmitry Dorofeev [dima@yasp.com]
89 *) Delay the logging of write errors to the pm to account for shutdown/restart.
91 *) (Win32) An assortment of fixes.
93 *) Fix some broken casts that were likely the cause of an assert.
95 *) Win32. Eliminate forward slashes from the named pipe path name.
96 Gerald Richter [richter@ecos.de]
98 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
99 applications it spawns. [ryans@amazon.com]
101 *) Added support for the -flush argument to FastCgiConfig.
102 Eric Sit [esit@alum.mit.edu]
104 *) Change the "which call to module_init() is this" check to a more
105 reliable approach. Doru Petrescu [pdoru@kappa.ro]
107 *) Close the old pipe file descriptor in apache main on USR1/HUP
108 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
110 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
111 [tetsuya@secom-sis.co.jp]
113 *) Eliminate the use of locks to assist in the clean shutdown of
114 applications. Instead, it is assumed that applications handle
115 termination signals properly (this is now embedded in the C
118 *) Fix Win32 process termination. Proper operation requires the use of
119 an updated application lib (termination is now signalled with an
120 Event and handled by specialized thread).
122 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
124 *) Added code so if the last instance of a dynamic application died without
125 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
126 configuration allows the last instance to be killed, then allow it to die).
127 Andrew Benham [adsb@bigfoot.com]
129 *) Fix the loadFactor calculation used to determine when dyanmic
130 applications could be killed off due to low demand [adsb@bigfoot.com].
132 *) Fix a deadlock condition that could occur with Win32 named
135 *) Fix a potential deadlock condition when FastCGI application
136 sent responses while still reading the client request (POST data).
140 *) Allow absolute pathnames in the -socket argument. Suggested by
141 Christian Jaeger [christian.jaeger@sl.ethz.ch].
143 *) Don't invoke suexec when the user/group for the fastcgi application
144 is the same as the apache main server. This is consistent with
145 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
147 *) Reset the apache drop dead timer upon successful read or writes
148 to/from the client. This eliminates timeouts that were occuring
149 during the large file transfers to/from slow clients.
151 *) Support generic wrappers such as cgiwrap by eliminating dependencies
152 on Apache's SUEXEC, renaming the FastCgiSuexec directive
153 FastCgiWrapper and eliminating any checks regarding the target
154 application (this is the repsonibility of the wrapper).
156 *) Fix a nasty bug that occurred when a client aborted a POST request
157 before the connection to a dynamic FastCGI application was opened.
158 The application's lock file descriptor wasn't setup but was being
159 closed which resulted in FD0 being closed. Normally this is open to
160 /dev/null and should pose no problems except that because the FD
161 was available it was being returned by Apache's accept(). This
162 caused it to be registered for pool cleanup. mod_cgi though moves
163 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
164 cleanup. Problem identified by checksum@163.net.
166 Changes with mod_fastcgi 2.2.8
168 *) Eliminate the concept of disabled applications. If a failure
169 occurs trying to setup an application (e.g. bind() error) its
170 tried repeatedly every init-start-delay seconds.
172 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
174 Changes with mod_fastcgi 2.2.6
176 *) Shutdown the PM when Apache appears to have disappeared.
178 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
180 *) (Win32) More dynamic fixes.
182 *) (Win32) Eliminate the per application named pipe mutex. Its now
183 per process to keep the lib happy (the use _FCGI_MUTEX_ should
184 removed from the lib). This allows pipe-based applications to
185 now handle multiple simoultanous requests (one per process).
187 *) Increase the number of open FDs we look to close when spawning apps.
189 *) Prevent an assert from popping unnecessarily.
191 *) (Win32) Add support for interpretter scripts.
193 *) (Win32) Fix named pipe handling (problems with large responses).
195 *) (Win32) Remove the "can exec" check.
197 Changes with mod_fastcgi 2.2.4
199 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
200 Rob Saccoccio [robs@chelsea.net]
202 *) Remove request type restriction (GET and POST only) in order to support
203 Web DAV requests. Scott Robertson [sroberts@codeit.com]
205 *) Allow requests for URLs such as /server/some/other/qualifier to match a
206 FastCGI server defined as /server. This was done primarily for Java
207 Servlets, but is generically useful.
209 *) Change the comm between the PM and the request handlers from a regular
210 file to a pipe and an assortment of other dynamic fixes.
212 *) Log dynamic process termination scheduling and the resulting
213 process exit notification.
215 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
216 application process from being killed off due to low demand.
218 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
220 *) Fix bugs that prevented dynamic processes from being shutdown when
221 the load subsided. Lars Heete [heete@do.isst.fhg.de]
223 *) Prevent dynamic processes from being scheduled to be started before
224 the init-start-delay or restart-delay period expires. This prevents
225 the queuing of process start requests while an application which has
226 a long initialization period starts up.
228 *) When suexec is enabled, allow the process manager to become root again in
229 order to signal applications it spawned.
231 *) Change the "server started" log message level from INFO to WARN.
233 *) Fix dynamic server "Connection Refused" handling.
235 *) Fix demand-based dynamic process spawing (uses a more portable approach).
237 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
238 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
239 longer than this period. It applies to the initiation of connections as
240 well (and thus is similar to appConnTimeout). Default is 30 seconds.
242 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
243 is more platform portable/predicable.
245 *) Leave STDOUT and STDERR open to the main server error_log. This should
246 eliminate a pile of protocol errors and having to track down 3rd party
247 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
248 be fixed in the application to use the FastCGI I/O, it just allows these
249 applications to run when previously they'd crash and burn.
251 *) Add the pass-header arg to server directives.
253 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
254 is done for CGI) - at least until PR2580 is resolved.
256 *) Initialize the default (empty) environment in a more socially acceptable
259 *) Miscellaneous doc improvements based on notes I've collected up over the
262 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
263 that resulted in not or improperly terminated strings.
265 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
266 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
268 Changes with mod_fastcgi 2.2.2
270 *) Added support for blocking connect()s by setting appConnTimeout to 0.
271 Non-blocking connect()s (the default) can be troublesome on some platforms.
272 Its expected that blocking connect()s will become the default (and
273 non-blocking connect()s will become optional) in the next release.
275 *) Dump a compile time error if the version of Apache is too old.
277 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
278 to Apache releases prior to 1.3.6.
280 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
284 Always restart a failed dynamic application if it is the last instance.
285 This means there once started there will always be at least one process
286 instance of a dynamic application.
288 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
289 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
290 the PM to start (another) application instance may help empty it faster.
292 Change two sleep() calls to select() based snoozes because alarm() is in
293 effect and sleep() and alarm() don't always play nice together.
295 Fixed a couple of error messages.
297 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
298 was ignored and the default was always used.
300 *) Allow the -initial-env argument to be used to pass variables from the
301 Apache process environment to the FastCGI server (by specifying a
302 variable name without the "=" or a value). [suggested by
303 Martin Lichtin <lichtin@oanda.com>]
305 *) Cleanup some debug macros.
307 *) Improved script stderr handling. [based on suggestions by
308 David Birnbaum <davidb@chelsea.net>]
310 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
311 updates to the INSTALL doc.
313 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
314 5 to 100. Its still configurable with the -listen-queue-depth option.
315 This should help eliminate the FAQ - Why do I see "Connection Refused"
318 *) Fix a bug in FastCgiExternalServer that broke support for external servers
319 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
321 Changes with mod_fastcgi 2.2.1
323 *) Updates to the INSTALL doc to describe building as a DSO.
325 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
326 Apache was running ("apachectl restart" does this), the contents of the
327 dynamic directory (dynamic sockets and the mbox) would be blown away
330 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
331 Apache header file) to prevent a compile error on systems without
334 *) During auth requests, the subprocess_env table was left holding variables
335 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
336 were passed to other processing phases (such as CGI/FastCGI). The
337 subprocess_env is now restored to its pre-auth condition.
339 *) Added a FastCGI Authorizer Role compatibility mode which implements the
340 specification to the tee. Use -compat arg with any of the Auth server
341 directives. This is intended for new and existing FastCGI authorizer
342 applications that require compatibility with other server implementations.
344 *) Fix logging from the process manager. Previously, all calls to the
345 logging routines used NULL for the server_rec. This works fine in a dev
346 environment, but results in most of these messages being tossed by
347 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
349 *) Always setup fr->header so that in the event the FastCGI server
350 doesn't get a valid FastCGI protocol header over the wire, we
351 can still print our error message to the log from do_work().
353 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
354 defined (Don Locrasto [locrasto@iceminer.com]).
356 Changes with mod_fastcgi 2.2.0
358 *) Serious rewrite of mod_fastcgi.html. An example conf will have
359 to wait until another time - I'm burned out on docs updates.
361 *) Add to the mod_fastcgi to the server version string.
363 *) SCO doesn't like the const in the arg to inet_addr() and
366 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
368 *) Deleted redundant header files from fcgi.h.
370 *) Full path names are no longer required for directives which take
371 file names as arguments. Like other Apache directives you can
372 specify paths from server_root (don't start with a "/").
374 *) The fcgiKillMgr is gone. This intermediate process sat between
375 the Apache parent and the FastCGI Process Manager under Apache
376 1.2. The process manager tries to name itself fcgi_pm (it used
379 *) Authentication, Authorization, and Access phases are now
380 supported. I've bent the FastCgi spec a bit: as many of the standard
381 environment variables (that were easy) are sent (the spec says don't
382 send some by name), all headers (except Status) sent by the auth
383 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
384 as environment variables rather than just those prefixed by
385 "Variable-". Custom responses for auth failures aren't supported
386 (yet.. we'll see what the demand for this is like). In addition to
387 the FastCGI protocol defined environment variable "FCGI_ROLE" being
388 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
389 set indicating which of the three phases is being processed. See
390 the docs for info on the new per-directory directives:
392 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
393 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
394 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
396 *) The code has been broken into logical chunks making figuring it
397 out much easier. All of the #defines a user may want to change are
398 now in mod_fastcgi.h.
400 *) All the logging has been converted to the "supported" Apache
401 logging routines, ap_log_error() and ap_log_rerror(). Log entries
402 for request specific errors should now be directed to the correct
403 server errorlog (if your running more than one). Some attempt has
404 been made at setting appropriate log levels (comments of course are
405 welcome) and printing errno info when its of value. All of log
406 messages now have "FastCGI" in them (nice for grep).
408 *) All of the calls to Apache routines have been updated to the ap_
409 convention, eliminating the need for the compat header.
411 *) By default we no longer flush every piece of data we get from the
412 FastCGI server to the client. This allows the FastCGI server to
413 release without having to wait for the client flush to complete.
414 The old behaviour can revived with the -flush argument to AppClass.
416 *) I renamed some of the directives for consistency (the old names
419 AppClass -> FastCgiServer,
420 FCGIConfig -> FastCgiConfig,
421 ExternalAppClass -> FastCgiExternalServer
423 *) Directive arguments are no longer case sensitive.
425 *) The module now uses hard_timeout() rather than soft_timeout. This
426 means that the module will longjump out of the request if the
427 drop-dead timeout expires (set with Apache's Timeout directive, the
428 default is 5min) or if the client closes the connection (SIGPIPE).
429 This is more typical for a Apache modules and I'm not convinced we
430 properly handle all of the error cases well enough to use
431 soft_timeout (I can think of one place we don't anyway). This means
432 your FastCGI application can see SIGPIPE.
434 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
435 Apache. Without this the default behaviour is exit().
437 *) It should (I think) run now under DSO now.
439 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
441 *) The broken -initial-env bug ins 2.1b1 is fixed.
443 *) Maybe some other stuff.
447 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
448 commit notes and stick 'em in here.. when I get some free time. ;)
453 *** Originally from docs/README..
455 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
456 From: Stanley Gambarin <stanleyg@cs.bu.edu>
458 *) Yet more changes to the source distribution. Separated out
459 the Tcl dynamic string and buffer libraries into separate files.
460 Did the same for the OS library, however, it is not really an
461 abstraction... more like a bunch of wrappers. Added to DEVNOTES
462 a note about the problem of Apache not allowing including header
465 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
466 From: Stanley Gambarin <stanleyg@cs.bu.edu>
468 *) More changes to the installation scripts. Theoretically we
469 should be able to to use the same Makefile/installation script
470 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
471 layer to use the same source code for both 1.2.x and 1.3.x sources.
473 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
474 From: Stanley Gambarin <stanleyg@cs.bu.edu>
476 *) Created Makefile to compile the module into the archive library,
477 just like the proxy module. Also initial draft of the installation
478 script, which should work for both 1.2 and 1.3 sources of Apache.
480 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
481 From: Stanley Gambarin <stanleyg@cs.bu.edu>
483 *) More source reorg. Separated out some header files. Need much
484 more work on this. Updated the TODO file. Added fcgivers.h and
485 CHANGES file to track the history.
487 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
488 From: Stanley Gambarin <stanleyg@cs.bu.edu>
490 *) Source reorganization. This is done to provide a basis for later
491 functionality implementation. See TODO file for more information
492 of what possible future enhancements are possible/desired.
494 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
495 From: David MacKenzie <djm@va.pubnix.com>
497 *) Create the "dynamic" dir and "mbox" as the user specified in the
498 User and Group directives instead of as root. They are created with
499 restrictive permissions, and then the module checks to see whether the
500 user specified in the User and Group directives has read, write, and
501 execute permissions on them. Those permissions had been explicitly
502 denied by creating them as "root".
504 *) Some documentation was garbled or incomplete.
506 *) When a FastCGI application can't be execl'd, the code used a
507 value of errno that may have been stomped on by intervening system
508 calls. That happened on BSD/OS 2.1, where the error_log was
509 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
510 An intervening stdio call was setting errno as a side-effect, which
511 it has the right to do.
513 *) Erroneous fprintf arguments, a missing return value, an unused
514 function, missing declarations, and other problems detected by gcc
517 *) Duplicated code to create the lock file name merged into a single
518 function, closing several memory leaks.
520 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
521 FCGI app in a given dynamic class would live forever, even if it
522 hadn't been requested for weeks.
524 *) The test for whether to keep looking for dynamic app victims to
525 kill was backwards, so no dynamic apps were ever selected as
528 *) The killInterval and updateInterval were being ignored; the
529 sigsuspend() forced recalculations only when a child died,
530 instead of at the intervals specified.
532 *) When a dynamic app couldn't be started, the program SEGV'd when
533 trying to free the ipcAddrPtr twice.
535 *) Remove the dead lock file and socket when cleaning up after a
536 server whose last child has been killed, so the FastCGIHandler
537 isn't fooled into thinking there is still a process serving
540 *) Make file locking robust in the presence of signals.
542 *) Rename some badly named variables.
544 *) Fix typos in many comments.
546 *) Fix some memory and file descriptor leaks.
548 *) Make the blocking kill of a server closer to working.
550 *) More fixes to calculations in dynamic application management.
553 *** Originally from docs/README.OMI..
555 What's New: Version 2.0b1, 16 Apr 1997
557 *) Implemented a mechanism by which FastCGI applications are started by
558 the web server on the first request and continue running. Also, a
559 heuristics have been implemented to allow for dynamic killing of the
560 running FastCGI apps. The configuration options are supplied via
561 the new FCGIConfig directive, that is described in the
564 *) When performing internal redirect, since the original request's body
565 has already been read, do not allow the redirected request to think
566 that is has one [body].
568 *) More conditional compilation for OS/2.
570 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
572 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
574 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
575 force some output to be written in the period of inactivity. This
576 circumvents the Apache's buffering during the server-push.
578 *) Added a header-parser function placeholder for compatibility with an
581 *) Additional checks have been placed with regard to permissions of the
582 FastCGI processes. These include the existence of the file,
583 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
585 *) Implemented a restart cleanup, so that no parentlesses FastCGI
586 applications are left after Apache is restarted and/or terminated.
587 This was accomplished via two independent mechanisms, where the
588 first one prevents a start of the process manager on the first
589 reading of the configuration files. The second mechanism deals away
590 from Apache's implementation of fork() (via spawn_child()) and
591 implements its own forking, thus removing any dependencies on the
592 Apache to cleanup processes during termination phase.
594 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
595 itself in the file conf.h
597 What's New: Version 1.4.3, 15 Jan 1997
599 *) Fixed compilation warnings for various platforms, as well as
600 conditional compilation for OS/2.
602 What's New: Version 1.4.2, 12 Dec 1996
604 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
605 will proceed under this version of Apache web server.
607 *) As the result of porting, the "include virtual" construct of SSI
608 will now work correctly using either <Location> or <Directory>
611 What's New: Version 1.4.1, 4 Dec 1996
613 *) Checks have been removed from the ScanCGIHeaders that provided
614 for the presence of both Status and Location headers as being
615 an error. Contradictory to CGI/1.1 Internet Draft, both of
616 these headers are used by the current CGI applications.
618 What's New: Version 1.4, 22 Nov 1996
620 *) Added the -port option to AppClass, allowing TCP/IP communication.
621 Added the -socket option to AppClass, allowing Unix domain
622 communication via a configurable pathname.
624 *) Added the ExternalAppClass directive, allowing TCP/IP
625 communication with remote FastCGI applications.
627 *) The handler had its own code for generating HTTP response headers;
628 now it uses Apache's. This reduces the size of the module.
629 More importantly, it fixes the bug in which "include virtual"
630 sees the HTTP response headers.
632 *) The response header parser performed very little checking.
633 Now the parser enforces the guidelines in the CGI/1.1
634 Internet-Draft: Status and Location are mutually exclusive,
635 Location can only be a response to GET or HEAD,
636 CGI response headers can't be repeated, etc. (The CGI response
637 headers are the ones the handler interprets: Status, Location,
640 *) The response header parser used to miss CGI headers
641 with no whitespace after the colon, e.g. "Status:200 OK".
643 *) The response header parser sometimes interpreted the first line
644 of content as an RFC822 continuation line.
646 *) The handler implemented a nonstandard version of Location
647 which never used internal redirects. The handler
648 now attempts to implement Location as specified in the
649 CGI/1.1 Internet-Draft. The module documentation explains
652 *) Error log entries from the response header parser were
653 pretty uninformative; they are better now. If a header is
654 malformed, the log entry includes it. If the headers are
655 unterminated rather than malformed, the log entry says that,
656 and says how many bytes were received from the app.
658 *) When the application manager forked a new process, and that
659 process ran into trouble before executing the first instruction
660 of application code, the process used to exit with status =
661 errno. This made certain configuration problems (e.g. incorrect
662 file permissions when server parent is root) quite difficult
663 to pin down. Now the failing child process opens up the error log
664 and writes an informative entry before exiting.
666 *) The module now correctly handles a slash at the end of the
667 DocumentRoot directive. This was a one-line fix.
669 What's New: Version 1.3.3, 17 Oct 1996
671 *) The module now registers its request handler under the name
672 fastcgi-script in addition to the name application/x-httpd-fcgi.
673 This was a one-line addition to the module, but had quite
674 a large impact on the documentation and sample configuration.
676 What's New: Version 1.3.2, 27 Sept 1996
678 *) On some systems (SunOS, Linux?), fopen for append has a bug
679 that strikes when two processes append to the same file.
680 This bug causes the process manager to corrupt the error log.
681 Work around the bug by calling open, then fdopen. A patch
682 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
683 (Reported by Bob Ramstad.)
685 What's New: Version 1.3.1, 17 Sept 1996
687 *) On Linux and SunOS, the Apache default user_id (-1) is not a
688 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
689 Scott Langley, others.)
691 *) On some systems (Linux, some Solaris?, Irix?), connect requires
692 write access to a Unix Domain socket, so provide it. (Reported
693 by Scott Langley, freeform@wired.com, others.)
695 *) If you hit a .fcg file, but there's no AppClass defined, the
696 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
699 What's New: Version 1.3, 4 Sept 1996
701 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
703 *) Module violated the Apache buff abstraction.
704 Fixed. Might now work with SSL (not tested).
705 As part of the fix, eliminated support for
706 Apache 1.0x versions.
708 *) Module failed to chmod the Unix domain listening sockets
709 it created, so protections were set according to the current
712 *) Module forked too many process manager processes: One
713 per AppClass. When the Apache parent ran as root, these
714 process manager processes ran as root. New process
715 manager is a single process (only started if AppClass is
716 used) and runs with same privileges as other children.
717 The new process manager doesn't do any polling, so
718 there's less system overhead than before.
720 *) AppClass insisted on getting at least two arguments. Fixed.
722 *) Module used setjmp/longjmp, causing compiler warnings
723 on some platforms (e.g. Linux.) The module no longer
726 *) Module created listening sockets in /tmp, where they were
727 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
728 directive to give control over the location of listening
731 *) Module wrote error log entries without a timestamp. Fixed.
733 *) AppClass directive wrote error messages to stderr in addition
734 to returning a char * message to the Apache core. Fixed.
736 What's New: Version 1.2, 3 June 1996
738 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
741 Add version string: APACHEVERSION macro to mod_fastcgi.c
743 *) chown FastCGI socket to user_id and setuid to user_id for app
746 *) Modify GetFromStream() by having it call an OS dependent
747 function GetStreamSize(FILE *) which uses FILE internal data member.
748 Linux users might need to modify GetStreamSize().
750 What's New: Version 1.1, 10 May 1996
752 *) If you specify a non-existent executable in the AppClass directive,
753 or if the file exists and it does not have execute permission you
754 get a constant stream of error messages telling you that "program
755 terminated due to a signal".
757 *) The mod_fastcgi module should use the standard Apache error logging
758 facility instead of writing to stderr.
760 Version 1.0, 30 April 1996