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