3 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
4 for use with wrappers (in lieu of finding the user/group associated
5 with a virtual host - under Apache2 this isn't accomodated).
7 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
8 apr_proc_create() changes.
10 *) Log when invoking and restoring the restart backoff policy.
12 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
13 OVERLAPPED structure).
15 *) Eliminate need for dummy files for external servers under Apache2
17 *) Fix auth compatibility mode handling for access checker and authorizer
19 *) Fix HEAD request handling. Based on a patch by
20 Chris Lightfoot [chris@ex-parrot.com]
22 *) [*nix] When autoupdate is enabled touch the socket when restarting
23 the processes to prevent further requests.
24 Eckebrecht von Pappenheim [evp@heise.de]
26 *) Apache 2.0 support.
28 *) [WIN32] Don't read from a potentially closed named pipe.
29 Philip Gladstone [philip@okena.com]
31 *) Require the Apache version 1.3.6 or later to eliminate some signal
34 *) [WIN32] Use asyncronous io with named pipes instead of polled
35 nonblocking io. This should eliminate the last of the npipe issues.
37 *) Handle an application returning a complete and valid response without
38 having consumed all of the data sent to it.
40 *) Consume remaining client data (RESPONDERs only) if any.
42 *) Add support for backing off attempts to start applications that continuously
43 fail to start. Three new macros defined in mod_fastcgi.h control this
44 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
46 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
47 applications that do not (properly) support the shutdown event (this
48 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
49 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
50 interval between signaling a proper shutdown and wacking the process(s)
51 with a TerminateProcess().
53 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
54 setting it was just plain broken.
56 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
58 *) Fix 'FastCgiConfig -autoUpdate'.
60 *) Fix 'FastCgiConfig -flush'.
62 *) Prevent silly maxProcesses and processSlack combinations.
63 Dmitry Dorofeev [dima@yasp.com]
65 *) Properly handle the killing of idle processes when one takes a long time
66 to exit once signaled down (or the config is funky).
67 Dmitry Dorofeev [dima@yasp.com]
69 *) Always kill the youngest instance of an application. Suggested by
70 Dmitry Dorofeev [dima@yasp.com]
74 *) Delay the logging of write errors to the pm to account for shutdown/restart.
76 *) (Win32) An assortment of fixes.
78 *) Fix some broken casts that were likely the cause of an assert.
80 *) Win32. Eliminate forward slashes from the named pipe path name.
81 Gerald Richter [richter@ecos.de]
83 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
84 applications it spawns. [ryans@amazon.com]
86 *) Added support for the -flush argument to FastCgiConfig.
87 Eric Sit [esit@alum.mit.edu]
89 *) Change the "which call to module_init() is this" check to a more
90 reliable approach. Doru Petrescu [pdoru@kappa.ro]
92 *) Close the old pipe file descriptor in apache main on USR1/HUP
93 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
95 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
96 [tetsuya@secom-sis.co.jp]
98 *) Eliminate the use of locks to assist in the clean shutdown of
99 applications. Instead, it is assumed that applications handle
100 termination signals properly (this is now embedded in the C
103 *) Fix Win32 process termination. Proper operation requires the use of
104 an updated application lib (termination is now signalled with an
105 Event and handled by specialized thread).
107 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
109 *) Added code so if the last instance of a dynamic application died without
110 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
111 configuration allows the last instance to be killed, then allow it to die).
112 Andrew Benham [adsb@bigfoot.com]
114 *) Fix the loadFactor calculation used to determine when dyanmic
115 applications could be killed off due to low demand [adsb@bigfoot.com].
117 *) Fix a deadlock condition that could occur with Win32 named
120 *) Fix a potential deadlock condition when FastCGI application
121 sent responses while still reading the client request (POST data).
125 *) Allow absolute pathnames in the -socket argument. Suggested by
126 Christian Jaeger [christian.jaeger@sl.ethz.ch].
128 *) Don't invoke suexec when the user/group for the fastcgi application
129 is the same as the apache main server. This is consistent with
130 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
132 *) Reset the apache drop dead timer upon successful read or writes
133 to/from the client. This eliminates timeouts that were occuring
134 during the large file transfers to/from slow clients.
136 *) Support generic wrappers such as cgiwrap by eliminating dependencies
137 on Apache's SUEXEC, renaming the FastCgiSuexec directive
138 FastCgiWrapper and eliminating any checks regarding the target
139 application (this is the repsonibility of the wrapper).
141 *) Fix a nasty bug that occurred when a client aborted a POST request
142 before the connection to a dynamic FastCGI application was opened.
143 The application's lock file descriptor wasn't setup but was being
144 closed which resulted in FD0 being closed. Normally this is open to
145 /dev/null and should pose no problems except that because the FD
146 was available it was being returned by Apache's accept(). This
147 caused it to be registered for pool cleanup. mod_cgi though moves
148 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
149 cleanup. Problem identified by checksum@163.net.
151 Changes with mod_fastcgi 2.2.8
153 *) Eliminate the concept of disabled applications. If a failure
154 occurs trying to setup an application (e.g. bind() error) its
155 tried repeatedly every init-start-delay seconds.
157 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
159 Changes with mod_fastcgi 2.2.6
161 *) Shutdown the PM when Apache appears to have disappeared.
163 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
165 *) (Win32) More dynamic fixes.
167 *) (Win32) Eliminate the per application named pipe mutex. Its now
168 per process to keep the lib happy (the use _FCGI_MUTEX_ should
169 removed from the lib). This allows pipe-based applications to
170 now handle multiple simoultanous requests (one per process).
172 *) Increase the number of open FDs we look to close when spawning apps.
174 *) Prevent an assert from popping unnecessarily.
176 *) (Win32) Add support for interpretter scripts.
178 *) (Win32) Fix named pipe handling (problems with large responses).
180 *) (Win32) Remove the "can exec" check.
182 Changes with mod_fastcgi 2.2.4
184 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
185 Rob Saccoccio [robs@chelsea.net]
187 *) Remove request type restriction (GET and POST only) in order to support
188 Web DAV requests. Scott Robertson [sroberts@codeit.com]
190 *) Allow requests for URLs such as /server/some/other/qualifier to match a
191 FastCGI server defined as /server. This was done primarily for Java
192 Servlets, but is generically useful.
194 *) Change the comm between the PM and the request handlers from a regular
195 file to a pipe and an assortment of other dynamic fixes.
197 *) Log dynamic process termination scheduling and the resulting
198 process exit notification.
200 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
201 application process from being killed off due to low demand.
203 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
205 *) Fix bugs that prevented dynamic processes from being shutdown when
206 the load subsided. Lars Heete [heete@do.isst.fhg.de]
208 *) Prevent dynamic processes from being scheduled to be started before
209 the init-start-delay or restart-delay period expires. This prevents
210 the queuing of process start requests while an application which has
211 a long initialization period starts up.
213 *) When suexec is enabled, allow the process manager to become root again in
214 order to signal applications it spawned.
216 *) Change the "server started" log message level from INFO to WARN.
218 *) Fix dynamic server "Connection Refused" handling.
220 *) Fix demand-based dynamic process spawing (uses a more portable approach).
222 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
223 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
224 longer than this period. It applies to the initiation of connections as
225 well (and thus is similar to appConnTimeout). Default is 30 seconds.
227 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
228 is more platform portable/predicable.
230 *) Leave STDOUT and STDERR open to the main server error_log. This should
231 eliminate a pile of protocol errors and having to track down 3rd party
232 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
233 be fixed in the application to use the FastCGI I/O, it just allows these
234 applications to run when previously they'd crash and burn.
236 *) Add the pass-header arg to server directives.
238 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
239 is done for CGI) - at least until PR2580 is resolved.
241 *) Initialize the default (empty) environment in a more socially acceptable
244 *) Miscellaneous doc improvements based on notes I've collected up over the
247 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
248 that resulted in not or improperly terminated strings.
250 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
251 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
253 Changes with mod_fastcgi 2.2.2
255 *) Added support for blocking connect()s by setting appConnTimeout to 0.
256 Non-blocking connect()s (the default) can be troublesome on some platforms.
257 Its expected that blocking connect()s will become the default (and
258 non-blocking connect()s will become optional) in the next release.
260 *) Dump a compile time error if the version of Apache is too old.
262 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
263 to Apache releases prior to 1.3.6.
265 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
269 Always restart a failed dynamic application if it is the last instance.
270 This means there once started there will always be at least one process
271 instance of a dynamic application.
273 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
274 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
275 the PM to start (another) application instance may help empty it faster.
277 Change two sleep() calls to select() based snoozes because alarm() is in
278 effect and sleep() and alarm() don't always play nice together.
280 Fixed a couple of error messages.
282 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
283 was ignored and the default was always used.
285 *) Allow the -initial-env argument to be used to pass variables from the
286 Apache process environment to the FastCGI server (by specifying a
287 variable name without the "=" or a value). [suggested by
288 Martin Lichtin <lichtin@oanda.com>]
290 *) Cleanup some debug macros.
292 *) Improved script stderr handling. [based on suggestions by
293 David Birnbaum <davidb@chelsea.net>]
295 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
296 updates to the INSTALL doc.
298 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
299 5 to 100. Its still configurable with the -listen-queue-depth option.
300 This should help eliminate the FAQ - Why do I see "Connection Refused"
303 *) Fix a bug in FastCgiExternalServer that broke support for external servers
304 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
306 Changes with mod_fastcgi 2.2.1
308 *) Updates to the INSTALL doc to describe building as a DSO.
310 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
311 Apache was running ("apachectl restart" does this), the contents of the
312 dynamic directory (dynamic sockets and the mbox) would be blown away
315 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
316 Apache header file) to prevent a compile error on systems without
319 *) During auth requests, the subprocess_env table was left holding variables
320 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
321 were passed to other processing phases (such as CGI/FastCGI). The
322 subprocess_env is now restored to its pre-auth condition.
324 *) Added a FastCGI Authorizer Role compatibility mode which implements the
325 specification to the tee. Use -compat arg with any of the Auth server
326 directives. This is intended for new and existing FastCGI authorizer
327 applications that require compatibility with other server implementations.
329 *) Fix logging from the process manager. Previously, all calls to the
330 logging routines used NULL for the server_rec. This works fine in a dev
331 environment, but results in most of these messages being tossed by
332 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
334 *) Always setup fr->header so that in the event the FastCGI server
335 doesn't get a valid FastCGI protocol header over the wire, we
336 can still print our error message to the log from do_work().
338 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
339 defined (Don Locrasto [locrasto@iceminer.com]).
341 Changes with mod_fastcgi 2.2.0
343 *) Serious rewrite of mod_fastcgi.html. An example conf will have
344 to wait until another time - I'm burned out on docs updates.
346 *) Add to the mod_fastcgi to the server version string.
348 *) SCO doesn't like the const in the arg to inet_addr() and
351 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
353 *) Deleted redundant header files from fcgi.h.
355 *) Full path names are no longer required for directives which take
356 file names as arguments. Like other Apache directives you can
357 specify paths from server_root (don't start with a "/").
359 *) The fcgiKillMgr is gone. This intermediate process sat between
360 the Apache parent and the FastCGI Process Manager under Apache
361 1.2. The process manager tries to name itself fcgi_pm (it used
364 *) Authentication, Authorization, and Access phases are now
365 supported. I've bent the FastCgi spec a bit: as many of the standard
366 environment variables (that were easy) are sent (the spec says don't
367 send some by name), all headers (except Status) sent by the auth
368 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
369 as environment variables rather than just those prefixed by
370 "Variable-". Custom responses for auth failures aren't supported
371 (yet.. we'll see what the demand for this is like). In addition to
372 the FastCGI protocol defined environment variable "FCGI_ROLE" being
373 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
374 set indicating which of the three phases is being processed. See
375 the docs for info on the new per-directory directives:
377 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
378 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
379 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
381 *) The code has been broken into logical chunks making figuring it
382 out much easier. All of the #defines a user may want to change are
383 now in mod_fastcgi.h.
385 *) All the logging has been converted to the "supported" Apache
386 logging routines, ap_log_error() and ap_log_rerror(). Log entries
387 for request specific errors should now be directed to the correct
388 server errorlog (if your running more than one). Some attempt has
389 been made at setting appropriate log levels (comments of course are
390 welcome) and printing errno info when its of value. All of log
391 messages now have "FastCGI" in them (nice for grep).
393 *) All of the calls to Apache routines have been updated to the ap_
394 convention, eliminating the need for the compat header.
396 *) By default we no longer flush every piece of data we get from the
397 FastCGI server to the client. This allows the FastCGI server to
398 release without having to wait for the client flush to complete.
399 The old behaviour can revived with the -flush argument to AppClass.
401 *) I renamed some of the directives for consistency (the old names
404 AppClass -> FastCgiServer,
405 FCGIConfig -> FastCgiConfig,
406 ExternalAppClass -> FastCgiExternalServer
408 *) Directive arguments are no longer case sensitive.
410 *) The module now uses hard_timeout() rather than soft_timeout. This
411 means that the module will longjump out of the request if the
412 drop-dead timeout expires (set with Apache's Timeout directive, the
413 default is 5min) or if the client closes the connection (SIGPIPE).
414 This is more typical for a Apache modules and I'm not convinced we
415 properly handle all of the error cases well enough to use
416 soft_timeout (I can think of one place we don't anyway). This means
417 your FastCGI application can see SIGPIPE.
419 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
420 Apache. Without this the default behaviour is exit().
422 *) It should (I think) run now under DSO now.
424 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
426 *) The broken -initial-env bug ins 2.1b1 is fixed.
428 *) Maybe some other stuff.
432 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
433 commit notes and stick 'em in here.. when I get some free time. ;)
438 *** Originally from docs/README..
440 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
441 From: Stanley Gambarin <stanleyg@cs.bu.edu>
443 *) Yet more changes to the source distribution. Separated out
444 the Tcl dynamic string and buffer libraries into separate files.
445 Did the same for the OS library, however, it is not really an
446 abstraction... more like a bunch of wrappers. Added to DEVNOTES
447 a note about the problem of Apache not allowing including header
450 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
451 From: Stanley Gambarin <stanleyg@cs.bu.edu>
453 *) More changes to the installation scripts. Theoretically we
454 should be able to to use the same Makefile/installation script
455 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
456 layer to use the same source code for both 1.2.x and 1.3.x sources.
458 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
459 From: Stanley Gambarin <stanleyg@cs.bu.edu>
461 *) Created Makefile to compile the module into the archive library,
462 just like the proxy module. Also initial draft of the installation
463 script, which should work for both 1.2 and 1.3 sources of Apache.
465 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
466 From: Stanley Gambarin <stanleyg@cs.bu.edu>
468 *) More source reorg. Separated out some header files. Need much
469 more work on this. Updated the TODO file. Added fcgivers.h and
470 CHANGES file to track the history.
472 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
473 From: Stanley Gambarin <stanleyg@cs.bu.edu>
475 *) Source reorganization. This is done to provide a basis for later
476 functionality implementation. See TODO file for more information
477 of what possible future enhancements are possible/desired.
479 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
480 From: David MacKenzie <djm@va.pubnix.com>
482 *) Create the "dynamic" dir and "mbox" as the user specified in the
483 User and Group directives instead of as root. They are created with
484 restrictive permissions, and then the module checks to see whether the
485 user specified in the User and Group directives has read, write, and
486 execute permissions on them. Those permissions had been explicitly
487 denied by creating them as "root".
489 *) Some documentation was garbled or incomplete.
491 *) When a FastCGI application can't be execl'd, the code used a
492 value of errno that may have been stomped on by intervening system
493 calls. That happened on BSD/OS 2.1, where the error_log was
494 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
495 An intervening stdio call was setting errno as a side-effect, which
496 it has the right to do.
498 *) Erroneous fprintf arguments, a missing return value, an unused
499 function, missing declarations, and other problems detected by gcc
502 *) Duplicated code to create the lock file name merged into a single
503 function, closing several memory leaks.
505 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
506 FCGI app in a given dynamic class would live forever, even if it
507 hadn't been requested for weeks.
509 *) The test for whether to keep looking for dynamic app victims to
510 kill was backwards, so no dynamic apps were ever selected as
513 *) The killInterval and updateInterval were being ignored; the
514 sigsuspend() forced recalculations only when a child died,
515 instead of at the intervals specified.
517 *) When a dynamic app couldn't be started, the program SEGV'd when
518 trying to free the ipcAddrPtr twice.
520 *) Remove the dead lock file and socket when cleaning up after a
521 server whose last child has been killed, so the FastCGIHandler
522 isn't fooled into thinking there is still a process serving
525 *) Make file locking robust in the presence of signals.
527 *) Rename some badly named variables.
529 *) Fix typos in many comments.
531 *) Fix some memory and file descriptor leaks.
533 *) Make the blocking kill of a server closer to working.
535 *) More fixes to calculations in dynamic application management.
538 *** Originally from docs/README.OMI..
540 What's New: Version 2.0b1, 16 Apr 1997
542 *) Implemented a mechanism by which FastCGI applications are started by
543 the web server on the first request and continue running. Also, a
544 heuristics have been implemented to allow for dynamic killing of the
545 running FastCGI apps. The configuration options are supplied via
546 the new FCGIConfig directive, that is described in the
549 *) When performing internal redirect, since the original request's body
550 has already been read, do not allow the redirected request to think
551 that is has one [body].
553 *) More conditional compilation for OS/2.
555 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
557 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
559 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
560 force some output to be written in the period of inactivity. This
561 circumvents the Apache's buffering during the server-push.
563 *) Added a header-parser function placeholder for compatibility with an
566 *) Additional checks have been placed with regard to permissions of the
567 FastCGI processes. These include the existence of the file,
568 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
570 *) Implemented a restart cleanup, so that no parentlesses FastCGI
571 applications are left after Apache is restarted and/or terminated.
572 This was accomplished via two independent mechanisms, where the
573 first one prevents a start of the process manager on the first
574 reading of the configuration files. The second mechanism deals away
575 from Apache's implementation of fork() (via spawn_child()) and
576 implements its own forking, thus removing any dependencies on the
577 Apache to cleanup processes during termination phase.
579 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
580 itself in the file conf.h
582 What's New: Version 1.4.3, 15 Jan 1997
584 *) Fixed compilation warnings for various platforms, as well as
585 conditional compilation for OS/2.
587 What's New: Version 1.4.2, 12 Dec 1996
589 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
590 will proceed under this version of Apache web server.
592 *) As the result of porting, the "include virtual" construct of SSI
593 will now work correctly using either <Location> or <Directory>
596 What's New: Version 1.4.1, 4 Dec 1996
598 *) Checks have been removed from the ScanCGIHeaders that provided
599 for the presence of both Status and Location headers as being
600 an error. Contradictory to CGI/1.1 Internet Draft, both of
601 these headers are used by the current CGI applications.
603 What's New: Version 1.4, 22 Nov 1996
605 *) Added the -port option to AppClass, allowing TCP/IP communication.
606 Added the -socket option to AppClass, allowing Unix domain
607 communication via a configurable pathname.
609 *) Added the ExternalAppClass directive, allowing TCP/IP
610 communication with remote FastCGI applications.
612 *) The handler had its own code for generating HTTP response headers;
613 now it uses Apache's. This reduces the size of the module.
614 More importantly, it fixes the bug in which "include virtual"
615 sees the HTTP response headers.
617 *) The response header parser performed very little checking.
618 Now the parser enforces the guidelines in the CGI/1.1
619 Internet-Draft: Status and Location are mutually exclusive,
620 Location can only be a response to GET or HEAD,
621 CGI response headers can't be repeated, etc. (The CGI response
622 headers are the ones the handler interprets: Status, Location,
625 *) The response header parser used to miss CGI headers
626 with no whitespace after the colon, e.g. "Status:200 OK".
628 *) The response header parser sometimes interpreted the first line
629 of content as an RFC822 continuation line.
631 *) The handler implemented a nonstandard version of Location
632 which never used internal redirects. The handler
633 now attempts to implement Location as specified in the
634 CGI/1.1 Internet-Draft. The module documentation explains
637 *) Error log entries from the response header parser were
638 pretty uninformative; they are better now. If a header is
639 malformed, the log entry includes it. If the headers are
640 unterminated rather than malformed, the log entry says that,
641 and says how many bytes were received from the app.
643 *) When the application manager forked a new process, and that
644 process ran into trouble before executing the first instruction
645 of application code, the process used to exit with status =
646 errno. This made certain configuration problems (e.g. incorrect
647 file permissions when server parent is root) quite difficult
648 to pin down. Now the failing child process opens up the error log
649 and writes an informative entry before exiting.
651 *) The module now correctly handles a slash at the end of the
652 DocumentRoot directive. This was a one-line fix.
654 What's New: Version 1.3.3, 17 Oct 1996
656 *) The module now registers its request handler under the name
657 fastcgi-script in addition to the name application/x-httpd-fcgi.
658 This was a one-line addition to the module, but had quite
659 a large impact on the documentation and sample configuration.
661 What's New: Version 1.3.2, 27 Sept 1996
663 *) On some systems (SunOS, Linux?), fopen for append has a bug
664 that strikes when two processes append to the same file.
665 This bug causes the process manager to corrupt the error log.
666 Work around the bug by calling open, then fdopen. A patch
667 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
668 (Reported by Bob Ramstad.)
670 What's New: Version 1.3.1, 17 Sept 1996
672 *) On Linux and SunOS, the Apache default user_id (-1) is not a
673 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
674 Scott Langley, others.)
676 *) On some systems (Linux, some Solaris?, Irix?), connect requires
677 write access to a Unix Domain socket, so provide it. (Reported
678 by Scott Langley, freeform@wired.com, others.)
680 *) If you hit a .fcg file, but there's no AppClass defined, the
681 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
684 What's New: Version 1.3, 4 Sept 1996
686 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
688 *) Module violated the Apache buff abstraction.
689 Fixed. Might now work with SSL (not tested).
690 As part of the fix, eliminated support for
691 Apache 1.0x versions.
693 *) Module failed to chmod the Unix domain listening sockets
694 it created, so protections were set according to the current
697 *) Module forked too many process manager processes: One
698 per AppClass. When the Apache parent ran as root, these
699 process manager processes ran as root. New process
700 manager is a single process (only started if AppClass is
701 used) and runs with same privileges as other children.
702 The new process manager doesn't do any polling, so
703 there's less system overhead than before.
705 *) AppClass insisted on getting at least two arguments. Fixed.
707 *) Module used setjmp/longjmp, causing compiler warnings
708 on some platforms (e.g. Linux.) The module no longer
711 *) Module created listening sockets in /tmp, where they were
712 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
713 directive to give control over the location of listening
716 *) Module wrote error log entries without a timestamp. Fixed.
718 *) AppClass directive wrote error messages to stderr in addition
719 to returning a char * message to the Apache core. Fixed.
721 What's New: Version 1.2, 3 June 1996
723 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
726 Add version string: APACHEVERSION macro to mod_fastcgi.c
728 *) chown FastCGI socket to user_id and setuid to user_id for app
731 *) Modify GetFromStream() by having it call an OS dependent
732 function GetStreamSize(FILE *) which uses FILE internal data member.
733 Linux users might need to modify GetStreamSize().
735 What's New: Version 1.1, 10 May 1996
737 *) If you specify a non-existent executable in the AppClass directive,
738 or if the file exists and it does not have execute permission you
739 get a constant stream of error messages telling you that "program
740 terminated due to a signal".
742 *) The mod_fastcgi module should use the standard Apache error logging
743 facility instead of writing to stderr.
745 Version 1.0, 30 April 1996