3 *) Require the Apache version 1.3.6 or later to eliminate some signal
6 *) [WIN32] Use asyncronous io with named pipes instead of polled
7 nonblocking io. This should eliminate the last of the npipe issues.
9 *) Handle an application returning a complete and valid response without
10 having consumed all of the data sent to it.
12 *) Consume remaining client data (RESPONDERs only) if any.
14 *) Add support for backing off attempts to start applications that continuously
15 fail to start. Three new macros defined in mod_fastcgi.h control this
16 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
18 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
19 applications that do not (properly) support the shutdown event (this
20 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
21 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
22 interval between signaling a proper shutdown and wacking the process(s)
23 with a TerminateProcess().
25 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
26 setting it was just plain broken.
28 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
30 *) Fix 'FastCgiConfig -autoUpdate'.
32 *) Fix 'FastCgiConfig -flush'.
34 *) Prevent silly maxProcesses and processSlack combinations.
35 Dmitry Dorofeev [dima@yasp.com]
37 *) Properly handle the killing of idle processes when one takes a long time
38 to exit once signaled down (or the config is funky).
39 Dmitry Dorofeev [dima@yasp.com]
41 *) Always kill the youngest instance of an application. Suggested by
42 Dmitry Dorofeev [dima@yasp.com]
46 *) Delay the logging of write errors to the pm to account for shutdown/restart.
48 *) (Win32) An assortment of fixes.
50 *) Fix some broken casts that were likely the cause of an assert.
52 *) Win32. Eliminate forward slashes from the named pipe path name.
53 Gerald Richter [richter@ecos.de]
55 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
56 applications it spawns. [ryans@amazon.com]
58 *) Added support for the -flush argument to FastCgiConfig.
59 Eric Sit [esit@alum.mit.edu]
61 *) Change the "which call to module_init() is this" check to a more
62 reliable approach. Doru Petrescu [pdoru@kappa.ro]
64 *) Close the old pipe file descriptor in apache main on USR1/HUP
65 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
67 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
68 [tetsuya@secom-sis.co.jp]
70 *) Eliminate the use of locks to assist in the clean shutdown of
71 applications. Instead, it is assumed that applications handle
72 termination signals properly (this is now embedded in the C
75 *) Fix Win32 process termination. Proper operation requires the use of
76 an updated application lib (termination is now signalled with an
77 Event and handled by specialized thread).
79 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
81 *) Added code so if the last instance of a dynamic application died without
82 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
83 configuration allows the last instance to be killed, then allow it to die).
84 Andrew Benham [adsb@bigfoot.com]
86 *) Fix the loadFactor calculation used to determine when dyanmic
87 applications could be killed off due to low demand [adsb@bigfoot.com].
89 *) Fix a deadlock condition that could occur with Win32 named
92 *) Fix a potential deadlock condition when FastCGI application
93 sent responses while still reading the client request (POST data).
97 *) Allow absolute pathnames in the -socket argument. Suggested by
98 Christian Jaeger [christian.jaeger@sl.ethz.ch].
100 *) Don't invoke suexec when the user/group for the fastcgi application
101 is the same as the apache main server. This is consistent with
102 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
104 *) Reset the apache drop dead timer upon successful read or writes
105 to/from the client. This eliminates timeouts that were occuring
106 during the large file transfers to/from slow clients.
108 *) Support generic wrappers such as cgiwrap by eliminating dependencies
109 on Apache's SUEXEC, renaming the FastCgiSuexec directive
110 FastCgiWrapper and eliminating any checks regarding the target
111 application (this is the repsonibility of the wrapper).
113 *) Fix a nasty bug that occurred when a client aborted a POST request
114 before the connection to a dynamic FastCGI application was opened.
115 The application's lock file descriptor wasn't setup but was being
116 closed which resulted in FD0 being closed. Normally this is open to
117 /dev/null and should pose no problems except that because the FD
118 was available it was being returned by Apache's accept(). This
119 caused it to be registered for pool cleanup. mod_cgi though moves
120 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
121 cleanup. Problem identified by checksum@163.net.
123 Changes with mod_fastcgi 2.2.8
125 *) Eliminate the concept of disabled applications. If a failure
126 occurs trying to setup an application (e.g. bind() error) its
127 tried repeatedly every init-start-delay seconds.
129 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
131 Changes with mod_fastcgi 2.2.6
133 *) Shutdown the PM when Apache appears to have disappeared.
135 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
137 *) (Win32) More dynamic fixes.
139 *) (Win32) Eliminate the per application named pipe mutex. Its now
140 per process to keep the lib happy (the use _FCGI_MUTEX_ should
141 removed from the lib). This allows pipe-based applications to
142 now handle multiple simoultanous requests (one per process).
144 *) Increase the number of open FDs we look to close when spawning apps.
146 *) Prevent an assert from popping unnecessarily.
148 *) (Win32) Add support for interpretter scripts.
150 *) (Win32) Fix named pipe handling (problems with large responses).
152 *) (Win32) Remove the "can exec" check.
154 Changes with mod_fastcgi 2.2.4
156 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
157 Rob Saccoccio [robs@chelsea.net]
159 *) Remove request type restriction (GET and POST only) in order to support
160 Web DAV requests. Scott Robertson [sroberts@codeit.com]
162 *) Allow requests for URLs such as /server/some/other/qualifier to match a
163 FastCGI server defined as /server. This was done primarily for Java
164 Servlets, but is generically useful.
166 *) Change the comm between the PM and the request handlers from a regular
167 file to a pipe and an assortment of other dynamic fixes.
169 *) Log dynamic process termination scheduling and the resulting
170 process exit notification.
172 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
173 application process from being killed off due to low demand.
175 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
177 *) Fix bugs that prevented dynamic processes from being shutdown when
178 the load subsided. Lars Heete [heete@do.isst.fhg.de]
180 *) Prevent dynamic processes from being scheduled to be started before
181 the init-start-delay or restart-delay period expires. This prevents
182 the queuing of process start requests while an application which has
183 a long initialization period starts up.
185 *) When suexec is enabled, allow the process manager to become root again in
186 order to signal applications it spawned.
188 *) Change the "server started" log message level from INFO to WARN.
190 *) Fix dynamic server "Connection Refused" handling.
192 *) Fix demand-based dynamic process spawing (uses a more portable approach).
194 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
195 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
196 longer than this period. It applies to the initiation of connections as
197 well (and thus is similar to appConnTimeout). Default is 30 seconds.
199 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
200 is more platform portable/predicable.
202 *) Leave STDOUT and STDERR open to the main server error_log. This should
203 eliminate a pile of protocol errors and having to track down 3rd party
204 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
205 be fixed in the application to use the FastCGI I/O, it just allows these
206 applications to run when previously they'd crash and burn.
208 *) Add the pass-header arg to server directives.
210 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
211 is done for CGI) - at least until PR2580 is resolved.
213 *) Initialize the default (empty) environment in a more socially acceptable
216 *) Miscellaneous doc improvements based on notes I've collected up over the
219 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
220 that resulted in not or improperly terminated strings.
222 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
223 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
225 Changes with mod_fastcgi 2.2.2
227 *) Added support for blocking connect()s by setting appConnTimeout to 0.
228 Non-blocking connect()s (the default) can be troublesome on some platforms.
229 Its expected that blocking connect()s will become the default (and
230 non-blocking connect()s will become optional) in the next release.
232 *) Dump a compile time error if the version of Apache is too old.
234 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
235 to Apache releases prior to 1.3.6.
237 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
241 Always restart a failed dynamic application if it is the last instance.
242 This means there once started there will always be at least one process
243 instance of a dynamic application.
245 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
246 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
247 the PM to start (another) application instance may help empty it faster.
249 Change two sleep() calls to select() based snoozes because alarm() is in
250 effect and sleep() and alarm() don't always play nice together.
252 Fixed a couple of error messages.
254 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
255 was ignored and the default was always used.
257 *) Allow the -initial-env argument to be used to pass variables from the
258 Apache process environment to the FastCGI server (by specifying a
259 variable name without the "=" or a value). [suggested by
260 Martin Lichtin <lichtin@oanda.com>]
262 *) Cleanup some debug macros.
264 *) Improved script stderr handling. [based on suggestions by
265 David Birnbaum <davidb@chelsea.net>]
267 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
268 updates to the INSTALL doc.
270 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
271 5 to 100. Its still configurable with the -listen-queue-depth option.
272 This should help eliminate the FAQ - Why do I see "Connection Refused"
275 *) Fix a bug in FastCgiExternalServer that broke support for external servers
276 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
278 Changes with mod_fastcgi 2.2.1
280 *) Updates to the INSTALL doc to describe building as a DSO.
282 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
283 Apache was running ("apachectl restart" does this), the contents of the
284 dynamic directory (dynamic sockets and the mbox) would be blown away
287 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
288 Apache header file) to prevent a compile error on systems without
291 *) During auth requests, the subprocess_env table was left holding variables
292 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
293 were passed to other processing phases (such as CGI/FastCGI). The
294 subprocess_env is now restored to its pre-auth condition.
296 *) Added a FastCGI Authorizer Role compatibility mode which implements the
297 specification to the tee. Use -compat arg with any of the Auth server
298 directives. This is intended for new and existing FastCGI authorizer
299 applications that require compatibility with other server implementations.
301 *) Fix logging from the process manager. Previously, all calls to the
302 logging routines used NULL for the server_rec. This works fine in a dev
303 environment, but results in most of these messages being tossed by
304 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
306 *) Always setup fr->header so that in the event the FastCGI server
307 doesn't get a valid FastCGI protocol header over the wire, we
308 can still print our error message to the log from do_work().
310 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
311 defined (Don Locrasto [locrasto@iceminer.com]).
313 Changes with mod_fastcgi 2.2.0
315 *) Serious rewrite of mod_fastcgi.html. An example conf will have
316 to wait until another time - I'm burned out on docs updates.
318 *) Add to the mod_fastcgi to the server version string.
320 *) SCO doesn't like the const in the arg to inet_addr() and
323 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
325 *) Deleted redundant header files from fcgi.h.
327 *) Full path names are no longer required for directives which take
328 file names as arguments. Like other Apache directives you can
329 specify paths from server_root (don't start with a "/").
331 *) The fcgiKillMgr is gone. This intermediate process sat between
332 the Apache parent and the FastCGI Process Manager under Apache
333 1.2. The process manager tries to name itself fcgi_pm (it used
336 *) Authentication, Authorization, and Access phases are now
337 supported. I've bent the FastCgi spec a bit: as many of the standard
338 environment variables (that were easy) are sent (the spec says don't
339 send some by name), all headers (except Status) sent by the auth
340 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
341 as environment variables rather than just those prefixed by
342 "Variable-". Custom responses for auth failures aren't supported
343 (yet.. we'll see what the demand for this is like). In addition to
344 the FastCGI protocol defined environment variable "FCGI_ROLE" being
345 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
346 set indicating which of the three phases is being processed. See
347 the docs for info on the new per-directory directives:
349 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
350 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
351 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
353 *) The code has been broken into logical chunks making figuring it
354 out much easier. All of the #defines a user may want to change are
355 now in mod_fastcgi.h.
357 *) All the logging has been converted to the "supported" Apache
358 logging routines, ap_log_error() and ap_log_rerror(). Log entries
359 for request specific errors should now be directed to the correct
360 server errorlog (if your running more than one). Some attempt has
361 been made at setting appropriate log levels (comments of course are
362 welcome) and printing errno info when its of value. All of log
363 messages now have "FastCGI" in them (nice for grep).
365 *) All of the calls to Apache routines have been updated to the ap_
366 convention, eliminating the need for the compat header.
368 *) By default we no longer flush every piece of data we get from the
369 FastCGI server to the client. This allows the FastCGI server to
370 release without having to wait for the client flush to complete.
371 The old behaviour can revived with the -flush argument to AppClass.
373 *) I renamed some of the directives for consistency (the old names
376 AppClass -> FastCgiServer,
377 FCGIConfig -> FastCgiConfig,
378 ExternalAppClass -> FastCgiExternalServer
380 *) Directive arguments are no longer case sensitive.
382 *) The module now uses hard_timeout() rather than soft_timeout. This
383 means that the module will longjump out of the request if the
384 drop-dead timeout expires (set with Apache's Timeout directive, the
385 default is 5min) or if the client closes the connection (SIGPIPE).
386 This is more typical for a Apache modules and I'm not convinced we
387 properly handle all of the error cases well enough to use
388 soft_timeout (I can think of one place we don't anyway). This means
389 your FastCGI application can see SIGPIPE.
391 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
392 Apache. Without this the default behaviour is exit().
394 *) It should (I think) run now under DSO now.
396 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
398 *) The broken -initial-env bug ins 2.1b1 is fixed.
400 *) Maybe some other stuff.
404 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
405 commit notes and stick 'em in here.. when I get some free time. ;)
410 *** Originally from docs/README..
412 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
413 From: Stanley Gambarin <stanleyg@cs.bu.edu>
415 *) Yet more changes to the source distribution. Separated out
416 the Tcl dynamic string and buffer libraries into separate files.
417 Did the same for the OS library, however, it is not really an
418 abstraction... more like a bunch of wrappers. Added to DEVNOTES
419 a note about the problem of Apache not allowing including header
422 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
423 From: Stanley Gambarin <stanleyg@cs.bu.edu>
425 *) More changes to the installation scripts. Theoretically we
426 should be able to to use the same Makefile/installation script
427 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
428 layer to use the same source code for both 1.2.x and 1.3.x sources.
430 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
431 From: Stanley Gambarin <stanleyg@cs.bu.edu>
433 *) Created Makefile to compile the module into the archive library,
434 just like the proxy module. Also initial draft of the installation
435 script, which should work for both 1.2 and 1.3 sources of Apache.
437 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
438 From: Stanley Gambarin <stanleyg@cs.bu.edu>
440 *) More source reorg. Separated out some header files. Need much
441 more work on this. Updated the TODO file. Added fcgivers.h and
442 CHANGES file to track the history.
444 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
445 From: Stanley Gambarin <stanleyg@cs.bu.edu>
447 *) Source reorganization. This is done to provide a basis for later
448 functionality implementation. See TODO file for more information
449 of what possible future enhancements are possible/desired.
451 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
452 From: David MacKenzie <djm@va.pubnix.com>
454 *) Create the "dynamic" dir and "mbox" as the user specified in the
455 User and Group directives instead of as root. They are created with
456 restrictive permissions, and then the module checks to see whether the
457 user specified in the User and Group directives has read, write, and
458 execute permissions on them. Those permissions had been explicitly
459 denied by creating them as "root".
461 *) Some documentation was garbled or incomplete.
463 *) When a FastCGI application can't be execl'd, the code used a
464 value of errno that may have been stomped on by intervening system
465 calls. That happened on BSD/OS 2.1, where the error_log was
466 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
467 An intervening stdio call was setting errno as a side-effect, which
468 it has the right to do.
470 *) Erroneous fprintf arguments, a missing return value, an unused
471 function, missing declarations, and other problems detected by gcc
474 *) Duplicated code to create the lock file name merged into a single
475 function, closing several memory leaks.
477 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
478 FCGI app in a given dynamic class would live forever, even if it
479 hadn't been requested for weeks.
481 *) The test for whether to keep looking for dynamic app victims to
482 kill was backwards, so no dynamic apps were ever selected as
485 *) The killInterval and updateInterval were being ignored; the
486 sigsuspend() forced recalculations only when a child died,
487 instead of at the intervals specified.
489 *) When a dynamic app couldn't be started, the program SEGV'd when
490 trying to free the ipcAddrPtr twice.
492 *) Remove the dead lock file and socket when cleaning up after a
493 server whose last child has been killed, so the FastCGIHandler
494 isn't fooled into thinking there is still a process serving
497 *) Make file locking robust in the presence of signals.
499 *) Rename some badly named variables.
501 *) Fix typos in many comments.
503 *) Fix some memory and file descriptor leaks.
505 *) Make the blocking kill of a server closer to working.
507 *) More fixes to calculations in dynamic application management.
510 *** Originally from docs/README.OMI..
512 What's New: Version 2.0b1, 16 Apr 1997
514 *) Implemented a mechanism by which FastCGI applications are started by
515 the web server on the first request and continue running. Also, a
516 heuristics have been implemented to allow for dynamic killing of the
517 running FastCGI apps. The configuration options are supplied via
518 the new FCGIConfig directive, that is described in the
521 *) When performing internal redirect, since the original request's body
522 has already been read, do not allow the redirected request to think
523 that is has one [body].
525 *) More conditional compilation for OS/2.
527 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
529 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
531 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
532 force some output to be written in the period of inactivity. This
533 circumvents the Apache's buffering during the server-push.
535 *) Added a header-parser function placeholder for compatibility with an
538 *) Additional checks have been placed with regard to permissions of the
539 FastCGI processes. These include the existence of the file,
540 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
542 *) Implemented a restart cleanup, so that no parentlesses FastCGI
543 applications are left after Apache is restarted and/or terminated.
544 This was accomplished via two independent mechanisms, where the
545 first one prevents a start of the process manager on the first
546 reading of the configuration files. The second mechanism deals away
547 from Apache's implementation of fork() (via spawn_child()) and
548 implements its own forking, thus removing any dependencies on the
549 Apache to cleanup processes during termination phase.
551 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
552 itself in the file conf.h
554 What's New: Version 1.4.3, 15 Jan 1997
556 *) Fixed compilation warnings for various platforms, as well as
557 conditional compilation for OS/2.
559 What's New: Version 1.4.2, 12 Dec 1996
561 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
562 will proceed under this version of Apache web server.
564 *) As the result of porting, the "include virtual" construct of SSI
565 will now work correctly using either <Location> or <Directory>
568 What's New: Version 1.4.1, 4 Dec 1996
570 *) Checks have been removed from the ScanCGIHeaders that provided
571 for the presence of both Status and Location headers as being
572 an error. Contradictory to CGI/1.1 Internet Draft, both of
573 these headers are used by the current CGI applications.
575 What's New: Version 1.4, 22 Nov 1996
577 *) Added the -port option to AppClass, allowing TCP/IP communication.
578 Added the -socket option to AppClass, allowing Unix domain
579 communication via a configurable pathname.
581 *) Added the ExternalAppClass directive, allowing TCP/IP
582 communication with remote FastCGI applications.
584 *) The handler had its own code for generating HTTP response headers;
585 now it uses Apache's. This reduces the size of the module.
586 More importantly, it fixes the bug in which "include virtual"
587 sees the HTTP response headers.
589 *) The response header parser performed very little checking.
590 Now the parser enforces the guidelines in the CGI/1.1
591 Internet-Draft: Status and Location are mutually exclusive,
592 Location can only be a response to GET or HEAD,
593 CGI response headers can't be repeated, etc. (The CGI response
594 headers are the ones the handler interprets: Status, Location,
597 *) The response header parser used to miss CGI headers
598 with no whitespace after the colon, e.g. "Status:200 OK".
600 *) The response header parser sometimes interpreted the first line
601 of content as an RFC822 continuation line.
603 *) The handler implemented a nonstandard version of Location
604 which never used internal redirects. The handler
605 now attempts to implement Location as specified in the
606 CGI/1.1 Internet-Draft. The module documentation explains
609 *) Error log entries from the response header parser were
610 pretty uninformative; they are better now. If a header is
611 malformed, the log entry includes it. If the headers are
612 unterminated rather than malformed, the log entry says that,
613 and says how many bytes were received from the app.
615 *) When the application manager forked a new process, and that
616 process ran into trouble before executing the first instruction
617 of application code, the process used to exit with status =
618 errno. This made certain configuration problems (e.g. incorrect
619 file permissions when server parent is root) quite difficult
620 to pin down. Now the failing child process opens up the error log
621 and writes an informative entry before exiting.
623 *) The module now correctly handles a slash at the end of the
624 DocumentRoot directive. This was a one-line fix.
626 What's New: Version 1.3.3, 17 Oct 1996
628 *) The module now registers its request handler under the name
629 fastcgi-script in addition to the name application/x-httpd-fcgi.
630 This was a one-line addition to the module, but had quite
631 a large impact on the documentation and sample configuration.
633 What's New: Version 1.3.2, 27 Sept 1996
635 *) On some systems (SunOS, Linux?), fopen for append has a bug
636 that strikes when two processes append to the same file.
637 This bug causes the process manager to corrupt the error log.
638 Work around the bug by calling open, then fdopen. A patch
639 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
640 (Reported by Bob Ramstad.)
642 What's New: Version 1.3.1, 17 Sept 1996
644 *) On Linux and SunOS, the Apache default user_id (-1) is not a
645 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
646 Scott Langley, others.)
648 *) On some systems (Linux, some Solaris?, Irix?), connect requires
649 write access to a Unix Domain socket, so provide it. (Reported
650 by Scott Langley, freeform@wired.com, others.)
652 *) If you hit a .fcg file, but there's no AppClass defined, the
653 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
656 What's New: Version 1.3, 4 Sept 1996
658 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
660 *) Module violated the Apache buff abstraction.
661 Fixed. Might now work with SSL (not tested).
662 As part of the fix, eliminated support for
663 Apache 1.0x versions.
665 *) Module failed to chmod the Unix domain listening sockets
666 it created, so protections were set according to the current
669 *) Module forked too many process manager processes: One
670 per AppClass. When the Apache parent ran as root, these
671 process manager processes ran as root. New process
672 manager is a single process (only started if AppClass is
673 used) and runs with same privileges as other children.
674 The new process manager doesn't do any polling, so
675 there's less system overhead than before.
677 *) AppClass insisted on getting at least two arguments. Fixed.
679 *) Module used setjmp/longjmp, causing compiler warnings
680 on some platforms (e.g. Linux.) The module no longer
683 *) Module created listening sockets in /tmp, where they were
684 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
685 directive to give control over the location of listening
688 *) Module wrote error log entries without a timestamp. Fixed.
690 *) AppClass directive wrote error messages to stderr in addition
691 to returning a char * message to the Apache core. Fixed.
693 What's New: Version 1.2, 3 June 1996
695 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
698 Add version string: APACHEVERSION macro to mod_fastcgi.c
700 *) chown FastCGI socket to user_id and setuid to user_id for app
703 *) Modify GetFromStream() by having it call an OS dependent
704 function GetStreamSize(FILE *) which uses FILE internal data member.
705 Linux users might need to modify GetStreamSize().
707 What's New: Version 1.1, 10 May 1996
709 *) If you specify a non-existent executable in the AppClass directive,
710 or if the file exists and it does not have execute permission you
711 get a constant stream of error messages telling you that "program
712 terminated due to a signal".
714 *) The mod_fastcgi module should use the standard Apache error logging
715 facility instead of writing to stderr.
717 Version 1.0, 30 April 1996