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