1 APACHE 2.3 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
22 * http://svn.apache.org/repos/asf/httpd/httpd/branches/2.2.x/STATUS
26 [NOTE that x.{odd}.z versions are strictly Alpha/Beta releases,
27 while x.{even}.z versions are Stable/GA releases.]
29 2.3.2 : in development
30 2.3.2 : Tagged on March 23, 2009.
31 2.3.1 : Tagged on January 2, 2009.
32 2.3.0 : Tagged on December 6, 2008, not released.
34 Contributors looking for a mission:
36 * Just do an egrep on "TODO" or "XXX" in the source.
38 * Review the bug database at: http://issues.apache.org/bugzilla/
40 * Review the "PatchAvailable" bugs in the bug database:
42 https://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2&keywords=PatchAvailable
44 After testing, you can append a comment saying "Reviewed and tested".
46 * Open bugs in the bug database.
49 CURRENT RELEASE NOTES:
54 * Not all MPMs are updated to set conn_rec::current_thread correctly.
55 (Prefork, Worker, Event, Simple are updated).
57 * Handling of non-trailing / config by non-default handler is broken
58 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105451701628081&w=2
59 jerenkrantz asks: Why should this block a release?
60 wsanchez agrees: this may be a change in behavior, but isn't
61 clearly wrong, and even if so, it doesn't seem like a
64 * the edge connection filter cannot be removed
65 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105366252619530&w=2
67 jerenkrantz asks: Why should this block a release?
69 stas replies: because it requires a rewrite of the filters stack
70 implementation (you have suggested that) and once 2.2 is
71 released you can't do that anymore.
73 pgollucci: this affects mod_perl I'm pretty sure.
77 * If the parent process dies, should the remaining child processes
78 "gracefully" self-terminate. Or maybe we should make it a runtime
79 option, or have a concept of 2 parent processes (one being a
81 See: Message-ID: <3C58232C.FE91F19F@Golux.Com>
83 Self-destruct: Ken, Martin, Lars
84 Not self-destruct: BrianP, Ian, Cliff, BillS
85 Make it runtime configurable: Aaron, jim, Justin, wrowe, rederpj, nd
87 /* The below was a concept on *how* to handle the problem */
88 Have 2 parents: +1: jim
89 -1: Justin, wrowe, rederpj, nd, pgollucci
90 +0: Lars, Martin (while standing by, could it do
93 * Make the worker MPM the default MPM for threaded Unix boxes.
94 +1: Justin, Ian, Cliff, BillS, striker, wrowe, nd, pgollucci
95 +0: BrianP, Aaron (mutex contention is looking better with the
96 latest code, let's continue tuning and testing), rederpj, jim
99 pquerna: Do we want to change this for *2.4*?
102 RELEASE NON-SHOWSTOPPERS BUT WOULD BE REAL NICE TO WRAP THESE UP:
104 * Patches submitted to the bug database:
105 http://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2&keywords=PatchAvailable
107 * Filter stacks and subrequests, redirects and fast redirects.
108 There's at least one PR that suffers from the current unclean behaviour
109 (which lets the server send garbage): PR 17629
110 nd says: Every subrequest should get its own filter stack with the
111 subreq_core filter as bottom-most. That filter does two things:
112 - swallow EOS buckets
113 - redirect the data stream to the upper request's (rr->main)
114 filter chain directly after the subrequest's starting
116 Once we have a clean solution, we can try to optimize
117 it, so that the server won't be slow down too much.
119 * RFC 2616 violations.
121 Open PRs: 15852, 15859, 15861, 15864, 15865, 15866, 15868, 15869,
122 15870, 16120, 16125, 16126, 16133, 16135, 16136, 16137,
123 16138, 16139, 16140, 16142, 16518, 16520, 16521,
124 jerenkrantz says: need to decide how many we need to backport and/or
125 if these rise to showstopper status.
126 wrowe suggests: it would be nice to see "MUST" v.s. "SHOULD" v.s. "MAY"
127 out of this list, without reviewing them individually.
129 * There is a bug in how we sort some hooks, at least the pre-config
130 hook. The first time we call the hooks, they are in the correct
131 order, but the second time, we don't sort them correctly. Currently,
132 the modules/http/config.m4 file has been renamed to
133 modules/http/config2.m4 to work around this problem, it should moved
134 back when this is fixed.
136 OtherBill offers that this is a SERIOUS problem. We do not sort
137 correctly by the ordering arguments passed to the register hook
138 functions. This was proven when I reordered the open_logs hook
139 to attempt to open the error logs prior to the access logs. Possibly
140 the entire sorting code needs to be refactored.
142 * pipes deadlock on all platforms with limited pipe buffers (e.g. both
143 Linux and Win32, as opposed to only Win32 on 1.3). The right solution
144 is either GStein's proposal for a "CGI Brigade", or OtherBill's proposal
145 for "Poll Buckets" for "Polling Filter Chains". Or maybe both :-)
147 * All handlers should always send content down even if r->header_only
148 is set. If not, it means that the HEAD requests don't generate the
149 same headers as a GET which is wrong.
151 * exec cmd and suexec arg-passing enhancements
152 Status: Patches proposed
153 Message-ID: <20020526041748.A29148@prodigy.Redbrick.DCU.IE>
154 (see the "proc.patch" and "suexec-shell.patch" links in this message)
156 * The 2.0.36 worker MPM graceless shutdown changes work but are
157 a bit clunky on some platforms; eg, on Linux, the loop to
158 join each worker thread seems to hang, and the parent ends up
159 killing off the child with SIGKILL. But at least it shuts down.
161 chrisd: Has this been fixed by the changes for PR 38737?
163 * --enable-mods-shared="foo1 foo2" is busted on Darwin. Pier
164 posted a patch (Message-ID: <B8DBBE8D.575A%pier@betaversion.org>).
166 * We do not properly substitute the prefix-variables in the configuration
167 scripts or generated-configs. (i.e. if sysconfdir is etc,
168 httpd-std.conf points to conf.)
170 * If any request gets through ap_process_request_internal() and is
171 scheduled to be served by the core handler, without a flag that this
172 r->filename was tested by dir/file_walk, we need to 500 at the very
173 end of the ap_process_request_internal() processing so sub_req-esters
174 know this request cannot be run. This provides authors of older
175 modules better compatibility, while still improving the security and
178 Status: still need to decide where this goes, OtherBill comments...
179 Message-ID: <065701c14526$495203b0$96c0b0d0@roweclan.net>
180 [Deleted comments regarding the ap_run_handler phase, as irrelevant
181 as BillS points out that "common case will be caught in
182 default_handler already (with the r->finfo.filetype == 0 check)"
183 and the issue is detecting this -before- we try to run the req.]
185 gregames says: can this happen somehow without a broken module
186 being involved? If not, why waste cycles trying to defend against
187 potential broken modules? It seems futile.
188 wrowe counters: no, it shouldn't happen unless the module is broken.
189 But the right answer is to fail the request up-front in dir/file
190 walk if the path was entirely invalid; and we can't do that either
191 UNTIL 2.1 or we break modules that haven't hooked map_to_storage.
193 * With AP_MODE_EXHAUSTIVE in the core, it is finally clear to me
194 how the Perchild MPM should be re-written. It hasn't worked
195 correctly since filters were added because it wasn't possible to
196 get the content that had already been written and the socket at
197 the same time. This mode lets us do that, so the MPM can be
200 * Can a static httpd be built reliably?
201 Message-ID: <20020207142751.T31582@clove.org>
203 * Usage of APR_BRIGADE_NORMALIZE in core_input_filter should be
206 <Pine.LNX.4.33.0201202232430.318-100000@deepthought.cs.virginia.edu>
207 Jeff wonders if we still care about this. It is no longer an
208 API issue but simply an extra trip through the brigade.
210 * Get perchild to work on platforms other than Linux. This
211 will require a portable mechanism to pass data and file/socket
212 descriptors between vhost child groups. An API was proposed
214 Message-ID: <20020111115006.K1529@clove.org>
216 * Try to get libtool inter-library dependency code working on AIX.
217 Message-ID: <cm3n10lx555.fsf@rdu163-40-092.nc.rr.com>
219 Justin says: If we get it working on AIX, we can enable this
220 on all platforms and clean up our build system somewhat.
221 Jeff says: I thought I tested a patch for you sometime in
222 January that you were going to commit within a few days.
224 * Handling of %2f in URIs. Currently both 1.3 and 2.0
225 completely disallow %2f in the request URI path (see
226 ap_unescape_url() in util.c). It's permitted and passed
227 through in the query string, however. Roy says the
228 original reason for disallowing it, from five years ago,
229 was to protect CGI scripts that applied PATH_INFO to
230 a filesystem location and which might be tricked by
231 ..%2f..%2f(...). We *should* allow path-info of the
232 form 'http://foo.com/index.cgi/path/to/path%2finfo'.
233 Since we've revamped a lot of our processing of path
234 segments, it would be nice to allow this, or at least
235 allow it conditionally with a directive.
237 OtherBill adds that %2f as the SECOND character of a multibyte
238 sequence causes the request to fail! This happens notably in
241 * There is increasing demand from module writers for an API
242 that will allow them to control the server à la apachectl.
243 Reasons include sole-function servers that need to die if
244 an external dependency (e.g., a database) fails, et cetera.
245 Perhaps something in the (ever more abused) scoreboard?
247 On the other hand, we already have a pipe that goes between parent
248 and child for graceful shutdown events, along with an API that
249 can be used to send a message down that pipe. In threaded MPMs,
250 it is easy enough to make that one pipe be used for graceful
251 and graceless events, and it is also easy to open that pipe
252 to both parent and child for writing. Then we just need to
253 figure out how to do graceless on non-threaded MPMs.
255 * Allow the DocumentRoot directive within <Location > scopes? This
256 allows the beloved (crusty) Alias /foo/ /somepath/foo/ followed
257 by a <Directory /somepath/foo> to become simply
258 <Location /foo/> DocumentRoot /somefile/foo (IMHO a bit more legible
259 and in-your-face.) DocumentRoot unset would be accepted [and would
260 not permit content to be served, only virtual resources such as
261 server-info or server-status.
262 This proposed change would _not_ depricate Alias.
263 striker: See the thread starting with Message-ID:
264 JLEGKKNELMHCJPNMOKHOGEEJFBAA.striker@apache.org.
266 * Win32: Rotatelogs sometimes is not terminated when Apache
267 goes down hard. FirstBill was looking at possibly tracking the
268 child's-child processes in the parent process.
269 stoddard: Shared scoreboard might offer a good way for the parent
270 to keep track of 'other child' processes and whack them if the child
272 Other thoughts on walking the process chain using the NT kernel
273 have also been proposed on APR.
275 * Eliminate unnecessary creation of pipes in mod_cgid
277 * Combine log_child and piped_log_spawn. Clean up http_log.c.
280 * Platforms that do not support fork (primarily Win32 and AS/400)
281 Architect start-up code that avoids initializing all the modules
282 in the parent process on platforms that do not support fork.
284 * There are still a number of places in the code where we are
285 losing error status (i.e. throwing away the error returned by a
286 system call and replacing it with a generic error code)
288 * Mass vhosting version of suEXEC.
290 * All DBMs suffer from confusion in support/dbmmanage (perl script) since
291 the dbmmanage employs the first-matched dbm format. This is not
292 necessarily the library that Apache was built with. Aught to
293 rewrite dbmmanage upon installation to bin/ with the proper library
294 for predictable mod_auth_dbm administration.
295 Questions; htdbm exists, time to kill dbmmanage, or does it remain
296 useful as a perl dbm management example? If we keep it,
297 do we address the issue above?
300 Some additional items remaining:
301 - case_preserved_filename stuff
302 (use the new canonical name stuff?)
303 - find a new home for ap_text(_header)
304 - is it possible to remove the DAV: namespace stuff from util_xml?
306 * ap_core_translate() and its use by mod_mmap_static and mod_file_cache
307 are a bit wonky. The function should probably be exposed as a utility
308 function (such as ap_translate_url2fs() or ap_validate_fs_url() or
309 something). Another approach would be a new hook phase after
310 "translate" which would allow the module to munge what the
311 translation has decided to do.
312 Status: Greg +1 (volunteers)
314 * Explore use of a post-config hook for the code in http_main.c which
315 calls ap_fixup_virutal_hosts(), ap_fini_vhost_config(), and
316 ap_sort_hooks() [to reduce the logic in main()]
318 * read the config tree just once, and process N times (as necessary)
320 * (possibly) use UUIDs in mod_unique_id and/or mod_usertrack
322 * (possibly) port the bug fix for PR 6942 (segv when LoadModule is put
323 into a VirtualHost container) to 2.0.
325 * shift stuff to mod_core.h
327 * callers of ap_run_create_request() should check the return value
328 for failure (Doug volunteers)
330 * Fix the worker MPM to use POD to kill child processes instead
331 of ap_os_killpg, regardless of how they should die.
333 chrisd: Is this done, by any chance? See r92598 and r93358.
335 * Scoreboard structures could be changed in the future such that
336 proper alignment is not maintained, leading to segfaults on
337 some systems. Cliff posted a patch to deal with this issue but
338 later recanted. See this message to dev@apr.apache.org:
340 <Pine.LNX.4.44.0203011354090.16457-200000@deepthought.cs.virginia.edu>
342 * APXS either needs to be fixed completely for use when apr is out of tree,
343 or it should drop query mode altogether, and we just grow an
344 httpd-config or similar arrangement.
345 To quote a discussion in STATUS earlier:
347 thommay: this doesn't fix all the problems with apxs and out of
348 tree apr/apr-util, but it's a good start. There's still the
349 query cases; but I'm beginning to think that in these cases
350 the app should be querying ap{r,u}-config directly
351 deprecate -q: add htpd-config: gstein, pquerna, minfrin, pgollucci
354 TODO ISSUES REMAINING IN MOD_SSL:
356 * Do we need SSL_set_read_ahead()?
358 * the ssl_expr api is NOT THREAD SAFE. race conditions exist:
359 -in ssl_expr_comp() if SSLRequire is used in .htaccess
360 (ssl_expr_info is global)
361 -is ssl_expr_eval() if there is an error
362 (ssl_expr_error is global)
364 * SSLRequire directive (parsing of) leaks memory
366 * Diffie-Hellman-Parameters for temporary keys are hardcoded in
367 ssl_engine_dh.c, while the comment in ssl_engine_kernel.c says:
368 "it is suggested that keys be changed daily or every 500
369 transactions, and more often if possible."
371 * ssl_var_lookup could be rewritten to be MUCH faster
373 * CRL callback should be pluggable
375 * session cache store should be pluggable
377 * init functions should return status code rather than ssl_die()
379 * ssl_engine_pphrase.c needs to be reworked so it is generic enough
380 to also decrypt proxy keys
382 * output warning when allowing SSL v2.0 ? its so old
385 * mod_proxy: Ability to run SSL over proxy gateway connections,
386 encrypting (or reencrypting) at the proxy.
388 * mod_cache: Handle ESI tags.
390 * mod_cache: Resolve issue of how to cache page fragements (or perhaps
391 -if- we want to cache page fragements). Today, mod_cache/mod_mem_cache
392 will cache #include 'virtual' requests (but not #include 'file'
393 requests). This was accomplished by making CACHE_IN a
394 CONTENT_SET-1 filter to force it to run before the SUBREQ_CORE
395 filter. But now responses cannot be cached that include the
396 effects of having been run through CONTENT_SET filters
397 (mod_deflate, mod_expires, etc). We could rerun all the
398 CONTENT_SET filters on the cached response, but this will not
399 work in all cases. For example, mod_expires relies on installing
400 the EXPIRATION filter during fixups. Contents served out of
401 mod_cache (out of the quick_handler) bypass -all- the request
402 line server hooks (Ryan really hated this. It is great for
403 performance, but bad because of the complications listed above).
405 mod_cache/mod_mem_cache/mod_disk_cache:
407 * mod_mem_cache: Consider adding a RevalidateTimeout directive to
408 specify time at which local cached content is to be revalidated
409 (ie, underlying file stat'ed to see if it has changed).
411 * mod_cache: CacheEnable/CacheDisable should accept regular expressions.
412 jerenkrantz says: Too slow. Get regexs away from speedy caches by
413 default. Introduce a new CacheEnableRegex if you want.
415 * mod_mem_cache/mod_disk_cache: Need to be able to query cache
416 status (num of entries, cache object properties, etc.).
417 mod_status could be extended to query optional hooks defined
418 by modules for the purpose of reporting module status.
419 mod_cache (et. al.) could define optional hooks that are called
420 to collect status. Status should be queryable by
422 jerenkrantz says: Yawn. Who cares.
424 * MaxRequestsPerChild measures connections, not requests.
425 Until someone has a better way, we'll probably just rename it
426 "MaxConnectionsPerChild".
428 * Regex containers don't work in an intutive way
429 Status: No one has come up with an efficient way to fix this
430 behavior. Dean has suggested getting rid of regex containers
432 OtherBill suggests: We at least seem to agree on eliminating
433 the <Container ~ foo> forms, and using only
434 <ContainerMatch foo> semantics.
436 * orig_ct in the byterange/multipart handling may not be
437 needed. Apache 1.3 just never stashed "multipart" into
438 r->content_type. We should probably follow suit since the
439 byterange stuff doesn't want the rest of the code to see the
440 multipart content-type; the other code should still think it is
441 dealing with the <orig_ct> stuff.
442 Status: Greg volunteers to investigate (esp. since he was most
443 likely the one to break it :-)
445 EXPERIMENTAL MODULES:
447 Experimental modules should eventually be be promoted to fully supported
448 status or removed from the repository entirely (ie, the
449 'experiment' failed). This section tracks what needs to happen to
450 get the modules promoted to fully supported status.