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 2.3.0 : in development
32 Contributors looking for a mission:
34 * Just do an egrep on "TODO" or "XXX" in the source.
36 * Review the bug database at: http://issues.apache.org/bugzilla/
38 * Review the "PatchAvailable" bugs in the bug database:
40 https://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2&keywords=PatchAvailable
42 After testing, you can append a comment saying "Reviewed and tested".
44 * Open bugs in the bug database.
47 CURRENT RELEASE NOTES:
52 * Handling of non-trailing / config by non-default handler is broken
53 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105451701628081&w=2
54 jerenkrantz asks: Why should this block a release?
55 wsanchez agrees: this may be a change in behavior, but isn't
56 clearly wrong, and even if so, it doesn't seem like a
59 * the edge connection filter cannot be removed
60 http://marc.theaimsgroup.com/?l=apache-httpd-dev&m=105366252619530&w=2
62 jerenkrantz asks: Why should this block a release?
64 stas replies: because it requires a rewrite of the filters stack
65 implementation (you have suggested that) and once 2.2 is
66 released you can't do that anymore.
71 * If the parent process dies, should the remaining child processes
72 "gracefully" self-terminate. Or maybe we should make it a runtime
73 option, or have a concept of 2 parent processes (one being a
75 See: Message-ID: <3C58232C.FE91F19F@Golux.Com>
77 Self-destruct: Ken, Martin, Lars
78 Not self-destruct: BrianP, Ian, Cliff, BillS
79 Make it runtime configurable: Aaron, jim, Justin, wrowe, rederpj, nd
81 /* The below was a concept on *how* to handle the problem */
82 Have 2 parents: +1: jim
83 -1: Justin, wrowe, rederpj, nd
84 +0: Lars, Martin (while standing by, could it do
87 * Make the worker MPM the default MPM for threaded Unix boxes.
88 +1: Justin, Ian, Cliff, BillS, striker, wrowe, nd
89 +0: BrianP, Aaron (mutex contention is looking better with the
90 latest code, let's continue tuning and testing), rederpj, jim
93 pquerna: Do we want to change this for *2.4*?
96 RELEASE NON-SHOWSTOPPERS BUT WOULD BE REAL NICE TO WRAP THESE UP:
98 * Patches submitted to the bug database:
99 http://issues.apache.org/bugzilla/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&product=Apache+httpd-2&keywords=PatchAvailable
101 * Filter stacks and subrequests, redirects and fast redirects.
102 There's at least one PR that suffers from the current unclean behaviour
103 (which lets the server send garbage): PR 17629
104 nd says: Every subrequest should get its own filter stack with the
105 subreq_core filter as bottom-most. That filter does two things:
106 - swallow EOS buckets
107 - redirect the data stream to the upper request's (rr->main)
108 filter chain directly after the subrequest's starting
110 Once we have a clean solution, we can try to optimize
111 it, so that the server won't be slow down too much.
113 * RFC 2616 violations.
115 Open PRs: 15852, 15859, 15861, 15864, 15865, 15866, 15868, 15869,
116 15870, 16120, 16125, 16126, 16133, 16135, 16136, 16137,
117 16138, 16139, 16140, 16142, 16518, 16520, 16521,
118 jerenkrantz says: need to decide how many we need to backport and/or
119 if these rise to showstopper status.
120 wrowe suggests: it would be nice to see "MUST" v.s. "SHOULD" v.s. "MAY"
121 out of this list, without reviewing them individually.
123 * There is a bug in how we sort some hooks, at least the pre-config
124 hook. The first time we call the hooks, they are in the correct
125 order, but the second time, we don't sort them correctly. Currently,
126 the modules/http/config.m4 file has been renamed to
127 modules/http/config2.m4 to work around this problem, it should moved
128 back when this is fixed.
130 OtherBill offers that this is a SERIOUS problem. We do not sort
131 correctly by the ordering arguments passed to the register hook
132 functions. This was proven when I reordered the open_logs hook
133 to attempt to open the error logs prior to the access logs. Possibly
134 the entire sorting code needs to be refactored.
136 * pipes deadlock on all platforms with limited pipe buffers (e.g. both
137 Linux and Win32, as opposed to only Win32 on 1.3). The right solution
138 is either GStein's proposal for a "CGI Brigade", or OtherBill's proposal
139 for "Poll Buckets" for "Polling Filter Chains". Or maybe both :-)
141 * All handlers should always send content down even if r->header_only
142 is set. If not, it means that the HEAD requests don't generate the
143 same headers as a GET which is wrong.
145 * exec cmd and suexec arg-passing enhancements
146 Status: Patches proposed
147 Message-ID: <20020526041748.A29148@prodigy.Redbrick.DCU.IE>
148 (see the "proc.patch" and "suexec-shell.patch" links in this message)
150 * The 2.0.36 worker MPM graceless shutdown changes work but are
151 a bit clunky on some platforms; eg, on Linux, the loop to
152 join each worker thread seems to hang, and the parent ends up
153 killing off the child with SIGKILL. But at least it shuts down.
155 chrisd: Has this been fixed by the changes for PR 38737?
157 * --enable-mods-shared="foo1 foo2" is busted on Darwin. Pier
158 posted a patch (Message-ID: <B8DBBE8D.575A%pier@betaversion.org>).
160 * We do not properly substitute the prefix-variables in the configuration
161 scripts or generated-configs. (i.e. if sysconfdir is etc,
162 httpd-std.conf points to conf.)
164 * If any request gets through ap_process_request_internal() and is
165 scheduled to be served by the core handler, without a flag that this
166 r->filename was tested by dir/file_walk, we need to 500 at the very
167 end of the ap_process_request_internal() processing so sub_req-esters
168 know this request cannot be run. This provides authors of older
169 modules better compatibility, while still improving the security and
172 Status: still need to decide where this goes, OtherBill comments...
173 Message-ID: <065701c14526$495203b0$96c0b0d0@roweclan.net>
174 [Deleted comments regarding the ap_run_handler phase, as irrelevant
175 as BillS points out that "common case will be caught in
176 default_handler already (with the r->finfo.filetype == 0 check)"
177 and the issue is detecting this -before- we try to run the req.]
179 gregames says: can this happen somehow without a broken module
180 being involved? If not, why waste cycles trying to defend against
181 potential broken modules? It seems futile.
182 wrowe counters: no, it shouldn't happen unless the module is broken.
183 But the right answer is to fail the request up-front in dir/file
184 walk if the path was entirely invalid; and we can't do that either
185 UNTIL 2.1 or we break modules that haven't hooked map_to_storage.
187 * With AP_MODE_EXHAUSTIVE in the core, it is finally clear to me
188 how the Perchild MPM should be re-written. It hasn't worked
189 correctly since filters were added because it wasn't possible to
190 get the content that had already been written and the socket at
191 the same time. This mode lets us do that, so the MPM can be
194 * Can a static httpd be built reliably?
195 Message-ID: <20020207142751.T31582@clove.org>
197 * Usage of APR_BRIGADE_NORMALIZE in core_input_filter should be
200 <Pine.LNX.4.33.0201202232430.318-100000@deepthought.cs.virginia.edu>
201 Jeff wonders if we still care about this. It is no longer an
202 API issue but simply an extra trip through the brigade.
204 * Get perchild to work on platforms other than Linux. This
205 will require a portable mechanism to pass data and file/socket
206 descriptors between vhost child groups. An API was proposed
208 Message-ID: <20020111115006.K1529@clove.org>
210 * Try to get libtool inter-library dependency code working on AIX.
211 Message-ID: <cm3n10lx555.fsf@rdu163-40-092.nc.rr.com>
213 Justin says: If we get it working on AIX, we can enable this
214 on all platforms and clean up our build system somewhat.
215 Jeff says: I thought I tested a patch for you sometime in
216 January that you were going to commit within a few days.
218 * Handling of %2f in URIs. Currently both 1.3 and 2.0
219 completely disallow %2f in the request URI path (see
220 ap_unescape_url() in util.c). It's permitted and passed
221 through in the query string, however. Roy says the
222 original reason for disallowing it, from five years ago,
223 was to protect CGI scripts that applied PATH_INFO to
224 a filesystem location and which might be tricked by
225 ..%2f..%2f(...). We *should* allow path-info of the
226 form 'http://foo.com/index.cgi/path/to/path%2finfo'.
227 Since we've revamped a lot of our processing of path
228 segments, it would be nice to allow this, or at least
229 allow it conditionally with a directive.
231 OtherBill adds that %2f as the SECOND character of a multibyte
232 sequence causes the request to fail! This happens notably in
235 * FreeBSD, threads, and worker MPM. All seems to work fine
236 if you only have one worker process with many threads. Add
237 a second worker process and the accept lock seems to be
238 lost. This might be an APR issue with how it deals with
239 the child_init hook (i.e. the fcntl lock needs to be resynced).
240 More examination and analysis is required.
241 Status: Works with FreeBSD 5.3. Does not work in previous versions.
242 This has also been reported on Cygwin.
244 * There is increasing demand from module writers for an API
245 that will allow them to control the server à la apachectl.
246 Reasons include sole-function servers that need to die if
247 an external dependency (e.g., a database) fails, et cetera.
248 Perhaps something in the (ever more abused) scoreboard?
250 On the other hand, we already have a pipe that goes between parent
251 and child for graceful shutdown events, along with an API that
252 can be used to send a message down that pipe. In threaded MPMs,
253 it is easy enough to make that one pipe be used for graceful
254 and graceless events, and it is also easy to open that pipe
255 to both parent and child for writing. Then we just need to
256 figure out how to do graceless on non-threaded MPMs.
258 * Allow the DocumentRoot directive within <Location > scopes? This
259 allows the beloved (crusty) Alias /foo/ /somepath/foo/ followed
260 by a <Directory /somepath/foo> to become simply
261 <Location /foo/> DocumentRoot /somefile/foo (IMHO a bit more legible
262 and in-your-face.) DocumentRoot unset would be accepted [and would
263 not permit content to be served, only virtual resources such as
264 server-info or server-status.
265 This proposed change would _not_ depricate Alias.
266 striker: See the thread starting with Message-ID:
267 JLEGKKNELMHCJPNMOKHOGEEJFBAA.striker@apache.org.
269 * Win32: Rotatelogs sometimes is not terminated when Apache
270 goes down hard. FirstBill was looking at possibly tracking the
271 child's-child processes in the parent process.
272 stoddard: Shared scoreboard might offer a good way for the parent
273 to keep track of 'other child' processes and whack them if the child
275 Other thoughts on walking the process chain using the NT kernel
276 have also been proposed on APR.
278 * Eliminate unnecessary creation of pipes in mod_cgid
280 * Combine log_child and piped_log_spawn. Clean up http_log.c.
283 * Platforms that do not support fork (primarily Win32 and AS/400)
284 Architect start-up code that avoids initializing all the modules
285 in the parent process on platforms that do not support fork.
287 * There are still a number of places in the code where we are
288 losing error status (i.e. throwing away the error returned by a
289 system call and replacing it with a generic error code)
291 * Mass vhosting version of suEXEC.
293 * All DBMs suffer from confusion in support/dbmmanage (perl script) since
294 the dbmmanage employs the first-matched dbm format. This is not
295 necessarily the library that Apache was built with. Aught to
296 rewrite dbmmanage upon installation to bin/ with the proper library
297 for predictable mod_auth_dbm administration.
298 Questions; htdbm exists, time to kill dbmmanage, or does it remain
299 useful as a perl dbm management example? If we keep it,
300 do we address the issue above?
303 Some additional items remaining:
304 - case_preserved_filename stuff
305 (use the new canonical name stuff?)
306 - find a new home for ap_text(_header)
307 - is it possible to remove the DAV: namespace stuff from util_xml?
309 * ap_core_translate() and its use by mod_mmap_static and mod_file_cache
310 are a bit wonky. The function should probably be exposed as a utility
311 function (such as ap_translate_url2fs() or ap_validate_fs_url() or
312 something). Another approach would be a new hook phase after
313 "translate" which would allow the module to munge what the
314 translation has decided to do.
315 Status: Greg +1 (volunteers)
317 * Explore use of a post-config hook for the code in http_main.c which
318 calls ap_fixup_virutal_hosts(), ap_fini_vhost_config(), and
319 ap_sort_hooks() [to reduce the logic in main()]
321 * read the config tree just once, and process N times (as necessary)
323 * (possibly) use UUIDs in mod_unique_id and/or mod_usertrack
325 * (possibly) port the bug fix for PR 6942 (segv when LoadModule is put
326 into a VirtualHost container) to 2.0.
328 * shift stuff to mod_core.h
330 * callers of ap_run_create_request() should check the return value
331 for failure (Doug volunteers)
333 * Fix the worker MPM to use POD to kill child processes instead
334 of ap_os_killpg, regardless of how they should die.
336 chrisd: Is this done, by any chance? See r92598 and r93358.
338 * Scoreboard structures could be changed in the future such that
339 proper alignment is not maintained, leading to segfaults on
340 some systems. Cliff posted a patch to deal with this issue but
341 later recanted. See this message to dev@apr.apache.org:
343 <Pine.LNX.4.44.0203011354090.16457-200000@deepthought.cs.virginia.edu>
345 * APXS either needs to be fixed completely for use when apr is out of tree,
346 or it should drop query mode altogether, and we just grow an
347 httpd-config or similar arrangement.
348 To quote a discussion in STATUS earlier:
350 thommay: this doesn't fix all the problems with apxs and out of
351 tree apr/apr-util, but it's a good start. There's still the
352 query cases; but I'm beginning to think that in these cases
353 the app should be querying ap{r,u}-config directly
354 gstein: agreed. apxs should deprecate the -q flag
355 pquerna: I vote for a httpd-config, and to deprecate the -q flag.
356 minfrin: +1 for httpd-config, and to deprecate -q.
359 TODO ISSUES REMAINING IN MOD_SSL:
361 * Do we need SSL_set_read_ahead()?
363 * the ssl_expr api is NOT THREAD SAFE. race conditions exist:
364 -in ssl_expr_comp() if SSLRequire is used in .htaccess
365 (ssl_expr_info is global)
366 -is ssl_expr_eval() if there is an error
367 (ssl_expr_error is global)
369 * SSLRequire directive (parsing of) leaks memory
371 * Diffie-Hellman-Parameters for temporary keys are hardcoded in
372 ssl_engine_dh.c, while the comment in ssl_engine_kernel.c says:
373 "it is suggested that keys be changed daily or every 500
374 transactions, and more often if possible."
376 * ssl_var_lookup could be rewritten to be MUCH faster
378 * CRL callback should be pluggable
380 * session cache store should be pluggable
382 * init functions should return status code rather than ssl_die()
384 * ssl_engine_pphrase.c needs to be reworked so it is generic enough
385 to also decrypt proxy keys
388 * mod_proxy: Ability to run SSL over proxy gateway connections,
389 encrypting (or reencrypting) at the proxy.
391 * mod_cache: Handle ESI tags.
393 * mod_cache: Resolve issue of how to cache page fragements (or perhaps
394 -if- we want to cache page fragements). Today, mod_cache/mod_mem_cache
395 will cache #include 'virtual' requests (but not #include 'file'
396 requests). This was accomplished by making CACHE_IN a
397 CONTENT_SET-1 filter to force it to run before the SUBREQ_CORE
398 filter. But now responses cannot be cached that include the
399 effects of having been run through CONTENT_SET filters
400 (mod_deflate, mod_expires, etc). We could rerun all the
401 CONTENT_SET filters on the cached response, but this will not
402 work in all cases. For example, mod_expires relies on installing
403 the EXPIRATION filter during fixups. Contents served out of
404 mod_cache (out of the quick_handler) bypass -all- the request
405 line server hooks (Ryan really hated this. It is great for
406 performance, but bad because of the complications listed above).
408 mod_cache/mod_mem_cache/mod_disk_cache:
410 * mod_mem_cache: Consider adding a RevalidateTimeout directive to
411 specify time at which local cached content is to be revalidated
412 (ie, underlying file stat'ed to see if it has changed).
414 * mod_cache: CacheEnable/CacheDisable should accept regular expressions.
415 jerenkrantz says: Too slow. Get regexs away from speedy caches by
416 default. Introduce a new CacheEnableRegex if you want.
418 * mod_mem_cache/mod_disk_cache: Need to be able to query cache
419 status (num of entries, cache object properties, etc.).
420 mod_status could be extended to query optional hooks defined
421 by modules for the purpose of reporting module status.
422 mod_cache (et. al.) could define optional hooks that are called
423 to collect status. Status should be queryable by
425 jerenkrantz says: Yawn. Who cares.
427 * MaxRequestsPerChild measures connections, not requests.
428 Until someone has a better way, we'll probably just rename it
429 "MaxConnectionsPerChild".
431 * Regex containers don't work in an intutive way
432 Status: No one has come up with an efficient way to fix this
433 behavior. Dean has suggested getting rid of regex containers
435 OtherBill suggests: We at least seem to agree on eliminating
436 the <Container ~ foo> forms, and using only
437 <ContainerMatch foo> semantics.
439 * orig_ct in the byterange/multipart handling may not be
440 needed. Apache 1.3 just never stashed "multipart" into
441 r->content_type. We should probably follow suit since the
442 byterange stuff doesn't want the rest of the code to see the
443 multipart content-type; the other code should still think it is
444 dealing with the <orig_ct> stuff.
445 Status: Greg volunteers to investigate (esp. since he was most
446 likely the one to break it :-)
448 EXPERIMENTAL MODULES:
450 Experimental modules should eventually be be promoted to fully supported
451 status or removed from the repository entirely (ie, the
452 'experiment' failed). This section tracks what needs to happen to
453 get the modules promoted to fully supported status.