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