httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rainer Jung <rainer.j...@kippdata.de>
Subject Questions on open ports between trunk and 2.4.x
Date Tue, 31 Jan 2012 15:07:18 GMT
I backported some trunk changes to 2.4.x (mostly cosmetic ones) and a 
few changes also in the other direction to keep code changes small in 
parts were there are still no functional changes.

A few backports are open, because I'm not sure, whether we want to 
backport. I left out the obvious changes (serf, apreq and similar), and 
also the changes related to the recent core filter improvements, but the 
below ones are not obvious to me. Comments on them welcome:

1) mod_reqtimeout
=================

r1202255 | pquerna | 2011-11-15 16:49:19 +0100 (Tue, 15 Nov 2011) | 1 line
disable mod_reqtimeout if not configured

It looks like mod_reqtimeout is now inactive by default in trunk, but 
ses some builtin default values in 2.4.x. Is that what we want? I'd 
expected that if we have it automatically use default values when loaded 
in 2.4.x, we would want the same behaviour in trunk.


2) log tags
===========

r1209743 | sf | 2011-12-02 23:26:54 +0100 (Fri, 02 Dec 2011) | 3 lines
Add APLOGNO() macro for unique tags for every log message.
Add some scripts to make adding these tags easier.

This has only been backported partially. The directory 
docs/log-message-tags is mising in 2.4.x as well as the update-log-tags 
and update-log-msg-tags targets in Makefile.


3) Recent install optimization
==============================

r1237447 | sf | 2012-01-29 21:11:55 +0100 (Sun, 29 Jan 2012) | 11 lines

Install multiple files per install.sh invocation

Libtool and BSD install support installing several files in one run.
Add support to install.sh and instdso.sh, too.
Replace for-loops in the Makefiles.

This reduces 'make install' time by approx. 50% because we save
lots of calls to the huge libtool shell script.

The AIX-specific magic in instdso.sh could use some testing.


Backport?


4) mod_proxy_html
=================

r1200556 | jorton | 2011-11-10 22:21:03 +0100 (Thu, 10 Nov 2011) | 3 lines

* modules/filters/mod_proxy_html.c: Fix various code style issues.  Use
   apr_is* ctype macro wrappers.


I guess we want that also in 2.4.x.


5) mod_authnz_ldap
==================

r1231257 | covener | 2012-01-13 20:18:03 +0100 (Fri, 13 Jan 2012) | 3 lines

whitespace only: shift a block refactored in r1231255 over 8 spaces.


r1231255 | covener | 2012-01-13 20:16:50 +0100 (Fri, 13 Jan 2012) | 5 lines

   *) mod_authnz_ldap: Don't try a potentially expensive nested groups
      search before exhausting all AuthLDAPGroupAttribute checks on the
      current group. PR52464


6) mod_cache
============

Applied to 2.4.x but not in trunk:

r1208384 | minfrin | 2011-11-30 12:21:43 +0100 (Wed, 30 Nov 2011) | 4 lines

mod_cache: Revert 
http://svn.apache.org/viewvc?view=revision&revision=1070179
as per the following thread:
http://www.gossamer-threads.com/lists/apache/dev/395830?do=post_view_threaded#395830

------------------------------------------------------------------------
r1208824 | minfrin | 2011-11-30 23:00:43 +0100 (Wed, 30 Nov 2011) | 4 lines

Backport:
mod_cache: Apply the API change that allows future mod_cache providers to
invalidate cache entries, which will fix PR15868.


Should both be ported forwards?


7) .htaccess handling
=====================

------------------------------------------------------------------------
r1229021 | niq | 2012-01-09 05:01:06 +0100 (Mon, 09 Jan 2012) | 5 lines

Core configuration: add AllowOverride option to treat syntax
errors in .htaccess as non-fatal.
PR 52439


8) ErrorLog directory checking
==============================

r1225199 | sf | 2011-12-28 15:54:49 +0100 (Wed, 28 Dec 2011) | 6 lines

Check during configtest that the directories for error logs exist

Testing under Windows is welcome

PR: 29941


9) Event MPM
============

I think everything apart from Paul's new queue has been backported, but 
please double check this list:

------------------------------------------------------------------------
r1203858 | trawick | 2011-11-18 22:39:33 +0100 (Fri, 18 Nov 2011) | 4 lines

follow up to r1202257 -- perform normal wakeup processing when
APR_EINTR is seen from apr_pollset_poll(), with expectation that
it was triggered by apr_pollset_wakeup()

------------------------------------------------------------------------
r1203404 | trawick | 2011-11-17 23:48:35 +0100 (Thu, 17 Nov 2011) | 10 lines

event no longer requires APR_POLLSET_THREADSAFE, or any
other non-standard feature (all but OS/2 have APR_POLLSET_WAKEABLE)

config-foo: don't require thread-safe pollset in order to
             build event

event.c: don't stress APR_ENOTIMPL in apr_pollset_create errors;
          just give the standard "check system or user limits"
          message

------------------------------------------------------------------------
r1202395 | trawick | 2011-11-15 20:38:31 +0100 (Tue, 15 Nov 2011) | 2 lines

spellcheck r1202258

------------------------------------------------------------------------
r1202259 | pquerna | 2011-11-15 16:52:59 +0100 (Tue, 15 Nov 2011) | 2 lines

Use apr_pollset_wakeup to ensure that the listener thread will process 
most enqueue'd events quickly

------------------------------------------------------------------------
r1202258 | pquerna | 2011-11-15 16:52:00 +0100 (Tue, 15 Nov 2011) | 2 lines

Because the pollset is now only mutated from the event thread, we no 
longer need the APR_POLLSET_THREADSAFE flag for the event_pollset

------------------------------------------------------------------------
r1202257 | pquerna | 2011-11-15 16:51:03 +0100 (Tue, 15 Nov 2011) | 2 lines

Create a new lock free circular queue, and use it in the EventMPM to 
remove the timeout mutex that was wrapping both timeout queue operations 
and pollset operations.

------------------------------------------------------------------------
r1202256 | pquerna | 2011-11-15 16:50:09 +0100 (Tue, 15 Nov 2011) | 3 lines

Instead of disabling the listening sockets from the pollset when under 
load, just stop calling the accept call, but leave the sockets in the 
pollset.


Regards,

Rainer

Mime
View raw message