1 APACHE 2.1 STATUS: -*-text-*-
2 Last modified at [$Date$]
4 The current version of this file can be found at:
5 http://svn.apache.org/repos/asf/httpd/httpd/trunk/STATUS
8 [NOTE that only Alpha/Beta releases occur in 2.1 development]
10 2.1.5 : in development
12 2.1.3 : Released on 2/22/2005 as alpha.
13 2.1.2 : Released on 12/08/2004 as alpha.
14 2.1.1 : Released on 11/19/2004 as alpha.
17 Please consult the following STATUS files for information on related projects:
19 * http://svn.apache.org/repos/asf/apr/apr/trunk/STATUS
20 * http://svn.apache.org/repos/asf/apr/apr-util/trunk/STATUS
21 * http://svn.apache.org/repos/asf/httpd/httpd/trunk/docs/STATUS
23 Contributors looking for a mission:
25 * Just do an egrep on "TODO" or "XXX" in the source.
27 * Review the bug database at: http://issues.apache.org/bugzilla/
29 * Review the "PatchAvailable" bugs in the bug database:
31 http://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2.0&keywords=PatchAvailable
33 After testing, you can append a comment saying "Reviewed and tested".
35 * Open bugs in the bug database.
37 CURRENT RELEASE NOTES:
41 * Handling of non-trailing / config by non-default handler is broken
42 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105451701628081&w=2
43 jerenkrantz asks: Why should this block a release?
45 * the edge connection filter cannot be removed
46 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105366252619530&w=2
47 jerenkrantz asks: Why should this block a release?
48 stas replies: because it requires a rewrite of the filters stack
49 implementation (you have suggested that) and once 2.2 is
50 released you can't do that anymore.
54 * httpd-std.conf and friends
56 a) httpd-std.conf should be tailored by install (from src or
57 binbuild) even if user has existing httpd.conf
58 +1: trawick, slive, gregames, ianh, Ken, wrowe, jwoolley, jim, nd,
60 wrowe - prefer httpd.default.conf to avoid ambiguity with cvs
62 b) tailored httpd-std.conf should be copied by install to
66 c) tailored httpd-std.conf should be installed to
67 sysconfdir/examples or manualdir/exampleconf/
68 +1: slive, trawick, Ken, nd (prefer the latter), erikabele
70 d) Installing a set of default config files when upgrading a server
71 doesn't make ANY sense at all.
72 +1: ianh - medium/big sites don't use 'standard config' anyway, as it
73 usually needs major customizations
74 -1: Ken, wrowe, jwoolley, jim, nd, erikabele
75 wrowe - diff is wonderful when comparing old/new default configs,
76 even for customized sites that ianh mentions
77 jim - ... assuming that the default configs have been updated
78 with the required inline docs to explain the
81 * If the parent process dies, should the remaining child processes
82 "gracefully" self-terminate. Or maybe we should make it a runtime
83 option, or have a concept of 2 parent processes (one being a
85 See: Message-ID: <3C58232C.FE91F19F@Golux.Com>
87 Self-destruct: Ken, Martin, Lars
88 Not self-destruct: BrianP, Ian, Cliff, BillS
89 Make it runtime configurable: Aaron, jim, Justin, wrowe, rederpj, nd
91 /* The below was a concept on *how* to handle the problem */
92 Have 2 parents: +1: jim
93 -1: Justin, wrowe, rederpj, nd
94 +0: Lars, Martin (while standing by, could it do
97 * Make the worker MPM the default MPM for threaded Unix boxes.
98 +1: Justin, Ian, Cliff, BillS, striker, wrowe, nd
99 +0: BrianP, Aaron (mutex contention is looking better with the
100 latest code, let's continue tuning and testing), rederpj, jim
103 pquerna: Do we want to change this for 2.2?
105 RELEASE NON-SHOWSTOPPERS BUT WOULD BE REAL NICE TO WRAP THESE UP:
107 * Patches submitted to the bug database:
108 http://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2.0&keywords=PatchAvailable
110 * The Event MPM does not work on Solaris 10. PR 34040.
112 * Filter stacks and subrequests, redirects and fast redirects.
113 There's at least one PR that suffers from the current unclean behaviour
114 (which lets the server send garbage): PR 17629
115 nd says: Every subrequest should get its own filter stack with the
116 subreq_core filter as bottom-most. That filter does two things:
117 - swallow EOS buckets
118 - redirect the data stream to the upper request's (rr->main)
119 filter chain directly after the subrequest's starting
121 Once we have a clean solution, we can try to optimize
122 it, so that the server won't be slow down too much.
124 * RFC 2616 violations.
126 Open PRs: 15852, 15859, 15861, 15864, 15865, 15866, 15868, 15869,
127 15870, 16120, 16125, 16126, 16133, 16135, 16136, 16137,
128 16138, 16139, 16140, 16142, 16518, 16520, 16521,
129 jerenkrantz says: need to decide how many we need to backport and/or
130 if these rise to showstopper status.
131 wrowe suggests: it would be nice to see "MUST" v.s. "SHOULD" v.s. "MAY"
132 out of this list, without reviewing them individually.
134 * There is a bug in how we sort some hooks, at least the pre-config
135 hook. The first time we call the hooks, they are in the correct
136 order, but the second time, we don't sort them correctly. Currently,
137 the modules/http/config.m4 file has been renamed to
138 modules/http/config2.m4 to work around this problem, it should moved
139 back when this is fixed.
141 OtherBill offers that this is a SERIOUS problem. We do not sort
142 correctly by the ordering arguments passed to the register hook
143 functions. This was proven when I reordered the open_logs hook
144 to attempt to open the error logs prior to the access logs. Possibly
145 the entire sorting code needs to be refactored.
147 * pipes deadlock on all platforms with limited pipe buffers (e.g. both
148 Linux and Win32, as opposed to only Win32 on 1.3). The right solution
149 is either GStein's proposal for a "CGI Brigade", or OtherBill's proposal
150 for "Poll Buckets" for "Polling Filter Chains". Or maybe both :-)
152 * All handlers should always send content down even if r->header_only
153 is set. If not, it means that the HEAD requests don't generate the
154 same headers as a GET which is wrong.
156 * exec cmd and suexec arg-passing enhancements
157 Status: Patches proposed
158 Message-ID: <20020526041748.A29148@prodigy.Redbrick.DCU.IE>
159 (see the "proc.patch" and "suexec-shell.patch" links in this message)
161 * The 2.0.36 worker MPM graceless shutdown changes work but are
162 a bit clunky on some platforms; eg, on Linux, the loop to
163 join each worker thread seems to hang, and the parent ends up
164 killing off the child with SIGKILL. But at least it shuts down.
166 * --enable-mods-shared="foo1 foo2" is busted on Darwin. Pier
167 posted a patch (Message-ID: <B8DBBE8D.575A%pier@betaversion.org>).
169 * We do not properly substitute the prefix-variables in the configuration
170 scripts or generated-configs. (i.e. if sysconfdir is etc,
171 httpd-std.conf points to conf.)
173 * If any request gets through ap_process_request_internal() and is
174 scheduled to be served by the core handler, without a flag that this
175 r->filename was tested by dir/file_walk, we need to 500 at the very
176 end of the ap_process_request_internal() processing so sub_req-esters
177 know this request cannot be run. This provides authors of older
178 modules better compatibility, while still improving the security and
181 Status: still need to decide where this goes, OtherBill comments...
182 Message-ID: <065701c14526$495203b0$96c0b0d0@roweclan.net>
183 [Deleted comments regarding the ap_run_handler phase, as irrelevant
184 as BillS points out that "common case will be caught in
185 default_handler already (with the r->finfo.filetype == 0 check)"
186 and the issue is detecting this -before- we try to run the req.]
188 gregames says: can this happen somehow without a broken module
189 being involved? If not, why waste cycles trying to defend against
190 potential broken modules? It seems futile.
191 wrowe counters: no, it shouldn't happen unless the module is broken.
192 But the right answer is to fail the request up-front in dir/file
193 walk if the path was entirely invalid; and we can't do that either
194 UNTIL 2.1 or we break modules that haven't hooked map_to_storage.
196 * With AP_MODE_EXHAUSTIVE in the core, it is finally clear to me
197 how the Perchild MPM should be re-written. It hasn't worked
198 correctly since filters were added because it wasn't possible to
199 get the content that had already been written and the socket at
200 the same time. This mode lets us do that, so the MPM can be
203 * Can a static httpd be built reliably?
204 Message-ID: <20020207142751.T31582@clove.org>
206 * Usage of APR_BRIGADE_NORMALIZE in core_input_filter should be
208 Message-ID: <Pine.LNX.4.33.0201202232430.318-100000@deepthought.cs.virginia.edu>
209 Jeff wonders if we still care about this. It is no longer an
210 API issue but simply an extra trip through the brigade.
212 * Get perchild to work on platforms other than Linux. This
213 will require a portable mechanism to pass data and file/socket
214 descriptors between vhost child groups. An API was proposed
216 Message-ID: <20020111115006.K1529@clove.org>
218 * Try to get libtool inter-library dependency code working on AIX.
219 Message-ID: <cm3n10lx555.fsf@rdu163-40-092.nc.rr.com>
221 Justin says: If we get it working on AIX, we can enable this
222 on all platforms and clean up our build system
224 Jeff says: I thought I tested a patch for you sometime in
225 January that you were going to commit within a few
228 * Handling of %2f in URIs. Currently both 1.3 and 2.0
229 completely disallow %2f in the request URI path (see
230 ap_unescape_url() in util.c). It's permitted and passed
231 through in the query string, however. Roy says the
232 original reason for disallowing it, from five years ago,
233 was to protect CGI scripts that applied PATH_INFO to
234 a filesystem location and which might be tricked by
235 ..%2f..%2f(...). We *should* allow path-info of the
236 form 'http://foo.com/index.cgi/path/to/path%2finfo'.
237 Since we've revamped a lot of our processing of path
238 segments, it would be nice to allow this, or at least
239 allow it conditionally with a directive.
241 OtherBill adds that %2f as the SECOND character of a multibyte
242 sequence causes the request to fail! This happens notably in
245 * FreeBSD, threads, and worker MPM. All seems to work fine
246 if you only have one worker process with many threads. Add
247 a second worker process and the accept lock seems to be
248 lost. This might be an APR issue with how it deals with
249 the child_init hook (i.e. the fcntl lock needs to be resynced).
250 More examination and analysis is required.
251 Status: Works with FreeBSD 5.3. Does not work in previous versions.
252 This has also been reported on Cygwin.
254 * There is increasing demand from module writers for an API
255 that will allow them to control the server à la apachectl.
256 Reasons include sole-function servers that need to die if
257 an external dependency (e.g., a database) fails, et cetera.
258 Perhaps something in the (ever more abused) scoreboard?
260 On the other hand, we already have a pipe that goes between parent
261 and child for graceful shutdown events, along with an API that
262 can be used to send a message down that pipe. In threaded MPMs,
263 it is easy enough to make that one pipe be used for graceful
264 and graceless events, and it is also easy to open that pipe
265 to both parent and child for writing. Then we just need to
266 figure out how to do graceless on non-threaded MPMs.
268 * Allow the DocumentRoot directive within <Location > scopes? This
269 allows the beloved (crusty) Alias /foo/ /somepath/foo/ followed
270 by a <Directory /somepath/foo> to become simply
271 <Location /foo/> DocumentRoot /somefile/foo (IMHO a bit more legible
272 and in-your-face.) DocumentRoot unset would be accepted [and would
273 not permit content to be served, only virtual resources such as
274 server-info or server-status.
275 This proposed change would _not_ depricate Alias.
276 striker: See the thread starting with Message-ID:
277 JLEGKKNELMHCJPNMOKHOGEEJFBAA.striker@apache.org.
279 * Win32: Rotatelogs sometimes is not terminated when Apache
280 goes down hard. FirstBill was looking at possibly tracking the
281 child's-child processes in the parent process.
282 stoddard: Shared scoreboard might offer a good way for the parent
283 to keep track of 'other child' processes and whack them if the child
285 Other thoughts on walking the process chain using the NT kernel
286 have also been proposed on APR.
288 * Eliminate unnecessary creation of pipes in mod_cgid
290 * Combine log_child and piped_log_spawn. Clean up http_log.c.
293 * Platforms that do not support fork (primarily Win32 and AS/400)
294 Architect start-up code that avoids initializing all the modules
295 in the parent process on platforms that do not support fork.
297 * There are still a number of places in the code where we are
298 losing error status (i.e. throwing away the error returned by a
299 system call and replacing it with a generic error code)
301 * Mass vhosting version of suEXEC.
303 * All DBMs suffer from confusion in support/dbmmanage (perl script) since
304 the dbmmanage employs the first-matched dbm format. This is not
305 necessarily the library that Apache was built with. Aught to
306 rewrite dbmmanage upon installation to bin/ with the proper library
307 for predictable mod_auth_dbm administration.
308 Questions; htdbm exists, time to kill dbmmanage, or does it remain
309 useful as a perl dbm management example? If we keep it,
310 do we address the issue above?
313 Some additional items remaining:
314 - case_preserved_filename stuff
315 (use the new canonical name stuff?)
316 - find a new home for ap_text(_header)
317 - is it possible to remove the DAV: namespace stuff from util_xml?
319 * ap_core_translate() and its use by mod_mmap_static and mod_file_cache
320 are a bit wonky. The function should probably be exposed as a utility
321 function (such as ap_translate_url2fs() or ap_validate_fs_url() or
322 something). Another approach would be a new hook phase after
323 "translate" which would allow the module to munge what the
324 translation has decided to do.
325 Status: Greg +1 (volunteers)
327 * Explore use of a post-config hook for the code in http_main.c which
328 calls ap_fixup_virutal_hosts(), ap_fini_vhost_config(), and
329 ap_sort_hooks() [to reduce the logic in main()]
331 * read the config tree just once, and process N times (as necessary)
333 * (possibly) use UUIDs in mod_unique_id and/or mod_usertrack
335 * (possibly) port the bug fix for PR 6942 (segv when LoadModule is put
336 into a VirtualHost container) to 2.0.
338 * shift stuff to mod_core.h
340 * callers of ap_run_create_request() should check the return value
341 for failure (Doug volunteers)
343 * Win32: Get Apache working on Windows 95/98. The following work
344 (at least) needs to be done:
345 - Document warning that OSR2 is required (for Crypt functions, in
346 rand.c, at least.) This could be resolved with an SSL library, or
347 randomization in APR itself.
348 - Bring the Win9xConHook.dll from 1.3 into 2.0 (no sense till it
349 actually works) and add in a splash of Win9x service code.
351 * Fix the worker MPM to use POD to kill child processes instead
352 of ap_os_killpg, regardless of how they should die.
354 * Scoreboard structures could be changed in the future such that
355 proper alignment is not maintained, leading to segfaults on
356 some systems. Cliff posted a patch to deal with this issue but
357 later recanted. See this message to dev@apr.apache.org:
358 Message-ID: <Pine.LNX.4.44.0203011354090.16457-200000@deepthought
361 * When sufficiently tested, the AllowEncodedSlashes/%2f patch
362 needs to be backported to 2.0 and 1.3.
364 * APXS either needs to be fixed completely for use when apr is out of tree,
365 or it should drop query mode altogether, and we just grow an
366 httpd-config or similar arrangement.
367 To quote a discussion in STATUS earlier:
369 thommay: this doesn't fix all the problems with apxs and out of
370 tree apr/apr-util, but it's a good start. There's still the
371 query cases; but I'm beginning to think that in these cases
372 the app should be querying ap{r,u}-config directly
373 gstein: agreed. apxs should deprecate the -q flag
375 TODO ISSUES REMAINING IN MOD_SSL:
377 * In order to use a DSO version of mod_ssl we have to link with
378 -lssl and -lcrypto. A workaround is in place right now where the
379 entire EXTRA_LIBS macro is being appended to the objects list, but
380 this is a hack. We should either revamp the APACHE_CHECK_SSL_TOOLKIT
381 autoconf function or come up with some other autoconf checks to
382 search for libssl and libcrypto and properly add them to mod_ssl's
385 * SSL renegotiations in combination with POST request
387 * Port or dispose all code inside #if 0...#endif blocks that remain
388 from the porting effort.
390 * Do we need SSL_set_read_ahead()?
392 * the ssl_expr api is NOT THREAD SAFE. race conditions exist:
393 -in ssl_expr_comp() if SSLRequire is used in .htaccess
394 (ssl_expr_info is global)
395 -is ssl_expr_eval() if there is an error
396 (ssl_expr_error is global)
398 * SSLRequire directive (parsing of) leaks memory
400 * Diffie-Hellman-Parameters for temporary keys are hardcoded in
401 ssl_engine_dh.c, while the comment in ssl_engine_kernel.c says:
402 "it is suggested that keys be changed daily or every 500
403 transactions, and more often if possible."
405 * ssl_var_lookup could be rewritten to be MUCH faster
407 * CRL callback should be pluggable
409 * session cache store should be pluggable
411 * init functions should return status code rather than ssl_die()
413 * ssl_engine_pphrase.c needs to be reworked so it is generic enough
414 to also decrypt proxy keys
416 * the shmcb code should just align its memory segment rather than
417 jumping through all the "safe" memcpy and memset hoops
420 * mod_proxy: Ability to run SSL over proxy gateway connections,
421 encrypting (or reencrypting) at the proxy.
423 * mod_cache: Handle ESI tags.
425 * mod_cache: Resolve issue of how to cache page fragements (or perhaps
426 -if- we want to cache page fragements). Today, mod_cache/mod_mem_cache
427 will cache #include 'virtual' requests (but not #include 'file'
428 requests). This was accomplished by making CACHE_IN a
429 CONTENT_SET-1 filter to force it to run before the SUBREQ_CORE
430 filter. But now responses cannot be cached that include the
431 effects of having been run through CONTENT_SET filters
432 (mod_deflate, mod_expires, etc). We could rerun all the
433 CONTENT_SET filters on the cached response, but this will not
434 work in all cases. For example, mod_expires relies on installing
435 the EXPIRATION filter during fixups. Contents served out of
436 mod_cache (out of the quick_handler) bypass -all- the request
437 line server hooks (Ryan really hated this. It is great for
438 performance, but bad because of the complications listed above).
440 mod_cache/mod_mem_cache/mod_disk_cache:
442 * mod_mem_cache: Consider adding a RevalidateTimeout directive to
443 specify time at which local cached content is to be revalidated
444 (ie, underlying file stat'ed to see if it has changed).
446 * mod_cache: CacheEnable/CacheDisable should accept regular expressions.
447 jerenkrantz says: Too slow. Get regexs away from speedy caches by
448 default. Introduce a new CacheEnableRegex if you want.
450 * mod_mem_cache/mod_disk_cache: Need to be able to query cache
451 status (num of entries, cache object properties, etc.).
452 mod_status could be extended to query optional hooks defined
453 by modules for the purpose of reporting module status.
454 mod_cache (et. al.) could define optional hooks that are called
455 to collect status. Status should be queryable by
457 jerenkrantz says: Yawn. Who cares.
459 EXPERIMENTAL MODULES:
461 Experimental modules should eventually be be promoted to fully supported
462 status or removed from the repository entirely (ie, the
463 'experiment' failed). This section tracks what needs to happen to
464 get the modules promoted to fully supported status.
467 Other bugs that need fixing:
469 * MaxRequestsPerChild measures connections, not requests.
470 Until someone has a better way, we'll probably just rename it
471 "MaxConnectionsPerChild".
473 * Regex containers don't work in an intutive way
474 Status: No one has come up with an efficient way to fix this
475 behavior. Dean has suggested getting rid of regex containers
477 OtherBill suggests: We at least seem to agree on eliminating
478 the <Container ~ foo> forms, and using only
479 <ContainerMatch foo> semantics.
481 * orig_ct in the byterange/multipart handling may not be
482 needed. Apache 1.3 just never stashed "multipart" into
483 r->content_type. We should probably follow suit since the
484 byterange stuff doesn't want the rest of the code to see the
485 multipart content-type; the other code should still think it is
486 dealing with the <orig_ct> stuff.
487 Status: Greg volunteers to investigate (esp. since he was most
488 likely the one to break it :-)
490 Binaries (probably not till beta):
492 Platform Avail. Volunteer
493 ------------------------------------------------------------------
494 AIX 4.3.3 no Bill Stoddard
497 hppa2.0w-hp-hpux11.00 no Cliff Woolley
498 i386-pc-solaris2.8 no Aaron Bannert
499 i386-unknown-freebsd4.5 no
500 i386-unknown-freebsd4.6 no Cliff Woolley
501 i686-pc-linux-gnu-slackware81 no Cliff Woolley
502 i686-pc-linux-gnu-rh70 no Aaron Bannert
503 i686-pc-linux-gnu-rh73 no Cliff Woolley
505 powerpc-apple-darwin5.5 no Aaron Bannert
506 powerpc-unknown-linux-gnu no Graham Leggett
507 s390-ibm-linux no Greg Ames
508 sparc-sun-solaris2.8 no Jim Jagielski
509 NetWare no Brad Nicholes
512 Win32-x86 no William Rowe