1 APACHE 2.1 STATUS: -*-text-*-
2 Last modified at [$Date$]
4 The current version of this file can be found at:
6 * http://svn.apache.org/repos/asf/httpd/httpd/trunk/STATUS
8 Documentation status is maintained seperately and can be found at:
10 * docs/STATUS in this source tree, or
11 * http://svn.apache.org/repos/asf/httpd/httpd/trunk/docs/STATUS
13 Consult the following STATUS files for information on related projects:
15 * http://svn.apache.org/repos/asf/apr/apr/trunk/STATUS
16 * http://svn.apache.org/repos/asf/apr/apr-util/trunk/STATUS
18 Patches considered for backport are noted in their branches' STATUS:
20 * http://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x/STATUS
21 * http://svn.apache.org/repos/asf/httpd/httpd/branches/2.0.x/STATUS
25 [NOTE that only Alpha/Beta releases occur in 2.1 development]
27 2.1.9 : in development
28 2.1.8 : Released on October 1, 2005 as beta.
29 2.1.7 : Released on September 12, 2005 as beta.
30 2.1.6 : Released on June 27, 2005 as alpha.
31 2.1.5 : Tagged on June 17, 2005.
33 2.1.3 : Released on February 22, 2005 as alpha.
34 2.1.2 : Released on December 8, 2004 as alpha.
35 2.1.1 : Released on November 19, 2004 as alpha.
39 Contributors looking for a mission:
41 * Just do an egrep on "TODO" or "XXX" in the source.
43 * Review the bug database at: http://issues.apache.org/bugzilla/
45 * Review the "PatchAvailable" bugs in the bug database:
47 http://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2.0&keywords=PatchAvailable
49 After testing, you can append a comment saying "Reviewed and tested".
51 * Open bugs in the bug database.
54 CURRENT RELEASE NOTES:
59 * Handling of non-trailing / config by non-default handler is broken
60 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105451701628081&w=2
61 jerenkrantz asks: Why should this block a release?
62 wsanchez agrees: this may be a change in behavior, but isn't
63 clearly wrong, and even if so, it doesn't seem like a
66 * the edge connection filter cannot be removed
67 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105366252619530&w=2
69 jerenkrantz asks: Why should this block a release?
71 stas replies: because it requires a rewrite of the filters stack
72 implementation (you have suggested that) and once 2.2 is
73 released you can't do that anymore.
78 * httpd-std.conf and friends
80 a) httpd-std.conf should be tailored by install (from src or
81 binbuild) even if user has existing httpd.conf
82 +1: trawick, slive, gregames, ianh, Ken, wrowe, jwoolley, jim, nd,
84 wrowe - prefer httpd.default.conf to avoid ambiguity with cvs
86 b) tailored httpd-std.conf should be copied by install to
90 c) tailored httpd-std.conf should be installed to
91 sysconfdir/examples or manualdir/exampleconf/
92 +1: slive, trawick, Ken, nd (prefer the latter), erikabele
93 +1: wsanchez (propose sysconfdir/examples/<version> for diffiness)
95 d) Installing a set of default config files when upgrading a server
96 doesn't make ANY sense at all.
97 +1: ianh - medium/big sites don't use 'standard config' anyway, as it
98 usually needs major customizations
99 -1: Ken, wrowe, jwoolley, jim, nd, erikabele
100 wrowe - diff is wonderful when comparing old/new default configs,
101 even for customized sites that ianh mentions
102 jim - ... assuming that the default configs have been updated
103 with the required inline docs to explain the
106 * If the parent process dies, should the remaining child processes
107 "gracefully" self-terminate. Or maybe we should make it a runtime
108 option, or have a concept of 2 parent processes (one being a
110 See: Message-ID: <3C58232C.FE91F19F@Golux.Com>
112 Self-destruct: Ken, Martin, Lars
113 Not self-destruct: BrianP, Ian, Cliff, BillS
114 Make it runtime configurable: Aaron, jim, Justin, wrowe, rederpj, nd
116 /* The below was a concept on *how* to handle the problem */
117 Have 2 parents: +1: jim
118 -1: Justin, wrowe, rederpj, nd
119 +0: Lars, Martin (while standing by, could it do
122 * Make the worker MPM the default MPM for threaded Unix boxes.
123 +1: Justin, Ian, Cliff, BillS, striker, wrowe, nd
124 +0: BrianP, Aaron (mutex contention is looking better with the
125 latest code, let's continue tuning and testing), rederpj, jim
128 pquerna: Do we want to change this for 2.2?
131 RELEASE NON-SHOWSTOPPERS BUT WOULD BE REAL NICE TO WRAP THESE UP:
133 * Patches submitted to the bug database:
134 http://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2.0&keywords=PatchAvailable
136 * Filter stacks and subrequests, redirects and fast redirects.
137 There's at least one PR that suffers from the current unclean behaviour
138 (which lets the server send garbage): PR 17629
139 nd says: Every subrequest should get its own filter stack with the
140 subreq_core filter as bottom-most. That filter does two things:
141 - swallow EOS buckets
142 - redirect the data stream to the upper request's (rr->main)
143 filter chain directly after the subrequest's starting
145 Once we have a clean solution, we can try to optimize
146 it, so that the server won't be slow down too much.
148 * RFC 2616 violations.
150 Open PRs: 15852, 15859, 15861, 15864, 15865, 15866, 15868, 15869,
151 15870, 16120, 16125, 16126, 16133, 16135, 16136, 16137,
152 16138, 16139, 16140, 16142, 16518, 16520, 16521,
153 jerenkrantz says: need to decide how many we need to backport and/or
154 if these rise to showstopper status.
155 wrowe suggests: it would be nice to see "MUST" v.s. "SHOULD" v.s. "MAY"
156 out of this list, without reviewing them individually.
158 * There is a bug in how we sort some hooks, at least the pre-config
159 hook. The first time we call the hooks, they are in the correct
160 order, but the second time, we don't sort them correctly. Currently,
161 the modules/http/config.m4 file has been renamed to
162 modules/http/config2.m4 to work around this problem, it should moved
163 back when this is fixed.
165 OtherBill offers that this is a SERIOUS problem. We do not sort
166 correctly by the ordering arguments passed to the register hook
167 functions. This was proven when I reordered the open_logs hook
168 to attempt to open the error logs prior to the access logs. Possibly
169 the entire sorting code needs to be refactored.
171 * pipes deadlock on all platforms with limited pipe buffers (e.g. both
172 Linux and Win32, as opposed to only Win32 on 1.3). The right solution
173 is either GStein's proposal for a "CGI Brigade", or OtherBill's proposal
174 for "Poll Buckets" for "Polling Filter Chains". Or maybe both :-)
176 * All handlers should always send content down even if r->header_only
177 is set. If not, it means that the HEAD requests don't generate the
178 same headers as a GET which is wrong.
180 * exec cmd and suexec arg-passing enhancements
181 Status: Patches proposed
182 Message-ID: <20020526041748.A29148@prodigy.Redbrick.DCU.IE>
183 (see the "proc.patch" and "suexec-shell.patch" links in this message)
185 * The 2.0.36 worker MPM graceless shutdown changes work but are
186 a bit clunky on some platforms; eg, on Linux, the loop to
187 join each worker thread seems to hang, and the parent ends up
188 killing off the child with SIGKILL. But at least it shuts down.
190 * --enable-mods-shared="foo1 foo2" is busted on Darwin. Pier
191 posted a patch (Message-ID: <B8DBBE8D.575A%pier@betaversion.org>).
193 * We do not properly substitute the prefix-variables in the configuration
194 scripts or generated-configs. (i.e. if sysconfdir is etc,
195 httpd-std.conf points to conf.)
197 * If any request gets through ap_process_request_internal() and is
198 scheduled to be served by the core handler, without a flag that this
199 r->filename was tested by dir/file_walk, we need to 500 at the very
200 end of the ap_process_request_internal() processing so sub_req-esters
201 know this request cannot be run. This provides authors of older
202 modules better compatibility, while still improving the security and
205 Status: still need to decide where this goes, OtherBill comments...
206 Message-ID: <065701c14526$495203b0$96c0b0d0@roweclan.net>
207 [Deleted comments regarding the ap_run_handler phase, as irrelevant
208 as BillS points out that "common case will be caught in
209 default_handler already (with the r->finfo.filetype == 0 check)"
210 and the issue is detecting this -before- we try to run the req.]
212 gregames says: can this happen somehow without a broken module
213 being involved? If not, why waste cycles trying to defend against
214 potential broken modules? It seems futile.
215 wrowe counters: no, it shouldn't happen unless the module is broken.
216 But the right answer is to fail the request up-front in dir/file
217 walk if the path was entirely invalid; and we can't do that either
218 UNTIL 2.1 or we break modules that haven't hooked map_to_storage.
220 * With AP_MODE_EXHAUSTIVE in the core, it is finally clear to me
221 how the Perchild MPM should be re-written. It hasn't worked
222 correctly since filters were added because it wasn't possible to
223 get the content that had already been written and the socket at
224 the same time. This mode lets us do that, so the MPM can be
227 * Can a static httpd be built reliably?
228 Message-ID: <20020207142751.T31582@clove.org>
230 * Usage of APR_BRIGADE_NORMALIZE in core_input_filter should be
232 Message-ID: <Pine.LNX.4.33.0201202232430.318-100000@deepthought.cs.virginia.edu>
233 Jeff wonders if we still care about this. It is no longer an
234 API issue but simply an extra trip through the brigade.
236 * Get perchild to work on platforms other than Linux. This
237 will require a portable mechanism to pass data and file/socket
238 descriptors between vhost child groups. An API was proposed
240 Message-ID: <20020111115006.K1529@clove.org>
242 * Try to get libtool inter-library dependency code working on AIX.
243 Message-ID: <cm3n10lx555.fsf@rdu163-40-092.nc.rr.com>
245 Justin says: If we get it working on AIX, we can enable this
246 on all platforms and clean up our build system
248 Jeff says: I thought I tested a patch for you sometime in
249 January that you were going to commit within a few
252 * Handling of %2f in URIs. Currently both 1.3 and 2.0
253 completely disallow %2f in the request URI path (see
254 ap_unescape_url() in util.c). It's permitted and passed
255 through in the query string, however. Roy says the
256 original reason for disallowing it, from five years ago,
257 was to protect CGI scripts that applied PATH_INFO to
258 a filesystem location and which might be tricked by
259 ..%2f..%2f(...). We *should* allow path-info of the
260 form 'http://foo.com/index.cgi/path/to/path%2finfo'.
261 Since we've revamped a lot of our processing of path
262 segments, it would be nice to allow this, or at least
263 allow it conditionally with a directive.
265 OtherBill adds that %2f as the SECOND character of a multibyte
266 sequence causes the request to fail! This happens notably in
269 * FreeBSD, threads, and worker MPM. All seems to work fine
270 if you only have one worker process with many threads. Add
271 a second worker process and the accept lock seems to be
272 lost. This might be an APR issue with how it deals with
273 the child_init hook (i.e. the fcntl lock needs to be resynced).
274 More examination and analysis is required.
275 Status: Works with FreeBSD 5.3. Does not work in previous versions.
276 This has also been reported on Cygwin.
278 * There is increasing demand from module writers for an API
279 that will allow them to control the server à la apachectl.
280 Reasons include sole-function servers that need to die if
281 an external dependency (e.g., a database) fails, et cetera.
282 Perhaps something in the (ever more abused) scoreboard?
284 On the other hand, we already have a pipe that goes between parent
285 and child for graceful shutdown events, along with an API that
286 can be used to send a message down that pipe. In threaded MPMs,
287 it is easy enough to make that one pipe be used for graceful
288 and graceless events, and it is also easy to open that pipe
289 to both parent and child for writing. Then we just need to
290 figure out how to do graceless on non-threaded MPMs.
292 * Allow the DocumentRoot directive within <Location > scopes? This
293 allows the beloved (crusty) Alias /foo/ /somepath/foo/ followed
294 by a <Directory /somepath/foo> to become simply
295 <Location /foo/> DocumentRoot /somefile/foo (IMHO a bit more legible
296 and in-your-face.) DocumentRoot unset would be accepted [and would
297 not permit content to be served, only virtual resources such as
298 server-info or server-status.
299 This proposed change would _not_ depricate Alias.
300 striker: See the thread starting with Message-ID:
301 JLEGKKNELMHCJPNMOKHOGEEJFBAA.striker@apache.org.
303 * Win32: Rotatelogs sometimes is not terminated when Apache
304 goes down hard. FirstBill was looking at possibly tracking the
305 child's-child processes in the parent process.
306 stoddard: Shared scoreboard might offer a good way for the parent
307 to keep track of 'other child' processes and whack them if the child
309 Other thoughts on walking the process chain using the NT kernel
310 have also been proposed on APR.
312 * Eliminate unnecessary creation of pipes in mod_cgid
314 * Combine log_child and piped_log_spawn. Clean up http_log.c.
317 * Platforms that do not support fork (primarily Win32 and AS/400)
318 Architect start-up code that avoids initializing all the modules
319 in the parent process on platforms that do not support fork.
321 * There are still a number of places in the code where we are
322 losing error status (i.e. throwing away the error returned by a
323 system call and replacing it with a generic error code)
325 * Mass vhosting version of suEXEC.
327 * All DBMs suffer from confusion in support/dbmmanage (perl script) since
328 the dbmmanage employs the first-matched dbm format. This is not
329 necessarily the library that Apache was built with. Aught to
330 rewrite dbmmanage upon installation to bin/ with the proper library
331 for predictable mod_auth_dbm administration.
332 Questions; htdbm exists, time to kill dbmmanage, or does it remain
333 useful as a perl dbm management example? If we keep it,
334 do we address the issue above?
337 Some additional items remaining:
338 - case_preserved_filename stuff
339 (use the new canonical name stuff?)
340 - find a new home for ap_text(_header)
341 - is it possible to remove the DAV: namespace stuff from util_xml?
343 * ap_core_translate() and its use by mod_mmap_static and mod_file_cache
344 are a bit wonky. The function should probably be exposed as a utility
345 function (such as ap_translate_url2fs() or ap_validate_fs_url() or
346 something). Another approach would be a new hook phase after
347 "translate" which would allow the module to munge what the
348 translation has decided to do.
349 Status: Greg +1 (volunteers)
351 * Explore use of a post-config hook for the code in http_main.c which
352 calls ap_fixup_virutal_hosts(), ap_fini_vhost_config(), and
353 ap_sort_hooks() [to reduce the logic in main()]
355 * read the config tree just once, and process N times (as necessary)
357 * (possibly) use UUIDs in mod_unique_id and/or mod_usertrack
359 * (possibly) port the bug fix for PR 6942 (segv when LoadModule is put
360 into a VirtualHost container) to 2.0.
362 * shift stuff to mod_core.h
364 * callers of ap_run_create_request() should check the return value
365 for failure (Doug volunteers)
367 * Win32: Get Apache working on Windows 95/98. The following work
368 (at least) needs to be done:
369 - Document warning that OSR2 is required (for Crypt functions, in
370 rand.c, at least.) This could be resolved with an SSL library, or
371 randomization in APR itself.
372 - Bring the Win9xConHook.dll from 1.3 into 2.0 (no sense till it
373 actually works) and add in a splash of Win9x service code.
375 * Fix the worker MPM to use POD to kill child processes instead
376 of ap_os_killpg, regardless of how they should die.
378 * Scoreboard structures could be changed in the future such that
379 proper alignment is not maintained, leading to segfaults on
380 some systems. Cliff posted a patch to deal with this issue but
381 later recanted. See this message to dev@apr.apache.org:
382 Message-ID: <Pine.LNX.4.44.0203011354090.16457-200000@deepthought
385 * APXS either needs to be fixed completely for use when apr is out of tree,
386 or it should drop query mode altogether, and we just grow an
387 httpd-config or similar arrangement.
388 To quote a discussion in STATUS earlier:
390 thommay: this doesn't fix all the problems with apxs and out of
391 tree apr/apr-util, but it's a good start. There's still the
392 query cases; but I'm beginning to think that in these cases
393 the app should be querying ap{r,u}-config directly
394 gstein: agreed. apxs should deprecate the -q flag
395 pquerna: I vote for a httpd-config, and to deprecate the -q flag.
396 minfrin: +1 for httpd-config, and to deprecate -q.
399 TODO ISSUES REMAINING IN MOD_SSL:
401 * In order to use a DSO version of mod_ssl we have to link with
402 -lssl and -lcrypto. A workaround is in place right now where the
403 entire EXTRA_LIBS macro is being appended to the objects list, but
404 this is a hack. We should either revamp the APACHE_CHECK_SSL_TOOLKIT
405 autoconf function or come up with some other autoconf checks to
406 search for libssl and libcrypto and properly add them to mod_ssl's
409 * SSL renegotiations in combination with POST request
411 * Port or dispose all code inside #if 0...#endif blocks that remain
412 from the porting effort.
414 * Do we need SSL_set_read_ahead()?
416 * the ssl_expr api is NOT THREAD SAFE. race conditions exist:
417 -in ssl_expr_comp() if SSLRequire is used in .htaccess
418 (ssl_expr_info is global)
419 -is ssl_expr_eval() if there is an error
420 (ssl_expr_error is global)
422 * SSLRequire directive (parsing of) leaks memory
424 * Diffie-Hellman-Parameters for temporary keys are hardcoded in
425 ssl_engine_dh.c, while the comment in ssl_engine_kernel.c says:
426 "it is suggested that keys be changed daily or every 500
427 transactions, and more often if possible."
429 * ssl_var_lookup could be rewritten to be MUCH faster
431 * CRL callback should be pluggable
433 * session cache store should be pluggable
435 * init functions should return status code rather than ssl_die()
437 * ssl_engine_pphrase.c needs to be reworked so it is generic enough
438 to also decrypt proxy keys
440 * the shmcb code should just align its memory segment rather than
441 jumping through all the "safe" memcpy and memset hoops
445 * mod_proxy: Ability to run SSL over proxy gateway connections,
446 encrypting (or reencrypting) at the proxy.
448 * mod_cache: Handle ESI tags.
450 * mod_cache: Resolve issue of how to cache page fragements (or perhaps
451 -if- we want to cache page fragements). Today, mod_cache/mod_mem_cache
452 will cache #include 'virtual' requests (but not #include 'file'
453 requests). This was accomplished by making CACHE_IN a
454 CONTENT_SET-1 filter to force it to run before the SUBREQ_CORE
455 filter. But now responses cannot be cached that include the
456 effects of having been run through CONTENT_SET filters
457 (mod_deflate, mod_expires, etc). We could rerun all the
458 CONTENT_SET filters on the cached response, but this will not
459 work in all cases. For example, mod_expires relies on installing
460 the EXPIRATION filter during fixups. Contents served out of
461 mod_cache (out of the quick_handler) bypass -all- the request
462 line server hooks (Ryan really hated this. It is great for
463 performance, but bad because of the complications listed above).
465 mod_cache/mod_mem_cache/mod_disk_cache:
467 * mod_mem_cache: Consider adding a RevalidateTimeout directive to
468 specify time at which local cached content is to be revalidated
469 (ie, underlying file stat'ed to see if it has changed).
471 * mod_cache: CacheEnable/CacheDisable should accept regular expressions.
472 jerenkrantz says: Too slow. Get regexs away from speedy caches by
473 default. Introduce a new CacheEnableRegex if you want.
475 * mod_mem_cache/mod_disk_cache: Need to be able to query cache
476 status (num of entries, cache object properties, etc.).
477 mod_status could be extended to query optional hooks defined
478 by modules for the purpose of reporting module status.
479 mod_cache (et. al.) could define optional hooks that are called
480 to collect status. Status should be queryable by
482 jerenkrantz says: Yawn. Who cares.
484 * MaxRequestsPerChild measures connections, not requests.
485 Until someone has a better way, we'll probably just rename it
486 "MaxConnectionsPerChild".
488 * Regex containers don't work in an intutive way
489 Status: No one has come up with an efficient way to fix this
490 behavior. Dean has suggested getting rid of regex containers
492 OtherBill suggests: We at least seem to agree on eliminating
493 the <Container ~ foo> forms, and using only
494 <ContainerMatch foo> semantics.
496 * orig_ct in the byterange/multipart handling may not be
497 needed. Apache 1.3 just never stashed "multipart" into
498 r->content_type. We should probably follow suit since the
499 byterange stuff doesn't want the rest of the code to see the
500 multipart content-type; the other code should still think it is
501 dealing with the <orig_ct> stuff.
502 Status: Greg volunteers to investigate (esp. since he was most
503 likely the one to break it :-)
505 EXPERIMENTAL MODULES:
507 Experimental modules should eventually be be promoted to fully supported
508 status or removed from the repository entirely (ie, the
509 'experiment' failed). This section tracks what needs to happen to
510 get the modules promoted to fully supported status.