outch. there are some possible NULL pointer references. Have you ever tried
AuthDigestProvider dbm? This results in a great kaboom. The patch makes
apache throw an error, if someone tries a provider, that doesn't support
the particular auth scheme.
when asking the providers for authentication, the main loop should
not only break, if access is granted. It should also break, if
access was *denied* by one provider. To be safe, it has to break
also, if an error occured. So the patch turns the condition around
and continues only, if the user was not found.
I find it also weird, that if auth was denied (by password
usually), the AuthBasicAuthoritative behaviour can override that
by "passing to lower modules". The patch changes that behaviour,
too.
Justin notes:
I'm kind of on the fence about that. I was originally thinking
optimistically, but yeah, it might make sense to do it
pessimistically. If there's any error, bug out.
André Malo [Wed, 4 Dec 2002 20:06:52 +0000 (20:06 +0000)]
keep german version in sync
(Note: upgrading.xml.de is not up-to-date in both branches)
(Note2: in Head also the original should be updated to 2.1 anyway)
Move the check of the Expect request header field after the hook
for ap_post_read_request, since that is the only opportunity for
modules to handle Expect extensions.
André Malo [Mon, 2 Dec 2002 21:58:35 +0000 (21:58 +0000)]
- removed obs_* files
- keep sitemap in sync
- Note: mod_auth_digest is currently exactly the same as in 2.0
(no changes were made due to the aaa-rewrite)
- modified the example in mod_info (was a bad one ;-)
- forward port (i.e copy) of rotatelogs.html, so we shouldn't forget the
recent changes
- fixed xslt/css for module index page
-> update transformation
Martin Kraemer [Mon, 2 Dec 2002 13:36:49 +0000 (13:36 +0000)]
If this worked on Mac OS-X, then OS-X should be fixed ;-)
Otherwise, it looks like Apache-2.x has not been compiled on any unixoid platform
anywhere since 2002-Nov-29
Final corrections; much of these .dsp definitions were redundant.
When you add a module to the Apache.dsw workspace, and make the new
module dependent upon libapr, libaprutil, libhttpd (and in the case
of mod_auth_ldap, make it dependent upon util_ldap) then everything
just falls into place.
the following CHANGES entries do not occur in the APACHE_2_0_BRANCH
CHANGES file. If you already committed them, please reflect in both
CHANGES files. If you would like to commit them, please do so, and
then reflect the patches under Apache 2.0.44.
Per Greg's request, add a version string component to the ap_provider.h
functions. This allows modules to register different versions of the
same provider.
Wilfredo Sanchez [Fri, 29 Nov 2002 07:23:57 +0000 (07:23 +0000)]
If apr and apr-util are not in-tree, we need to be able to find the
include directory for each in order to generate the server/exports.c
and server/export_vars.h files.
configure.in:
- Provide APR_INCLUDEDIR, APU_INCLUDEDIR.
server/Makefile.in:
- Use $APR_INCLUDEDIR $APU_INCLUDEDIR vars.
- Add export_files target to generate a list of headers with symbols
to export, use that list to generate exports.c and export_vars.h.
Wilfredo Sanchez [Fri, 29 Nov 2002 07:21:07 +0000 (07:21 +0000)]
If apr and apr-util are not in-tree, we need to be able to find the
include directory for each in order to generate the server/exports.c
and server/export_vars.h files.
configure.in:
- Provide APR_INCLUDEDIR, APU_INCLUDEDIR.
server/Makefile.in:
- Use $APR_INCLUDEDIR $APU_INCLUDEDIR vars.
- Add export_files target to generate a list of headers with symbols
to export, use that list to generate exports.c and export_vars.h.
Wilfredo Sanchez [Thu, 28 Nov 2002 23:31:37 +0000 (23:31 +0000)]
Changes to reduce dependency on in-tree apr and apr-util.
Change AC_CONFIG_AUX_DIR in configure to build from srclib/apr/build.
Have buildconf copy the needed files (config.guess, config.sub,
PrintPath) from srclib/apr/build to build.
Change binbuild.sh to stop using srclib/apr.
This gets to to where ./configure --with-apr --with-apr-util actually
works without an in-tree apr and apr-util, though we still need them
in-tree to run buildconf.
André Malo [Wed, 27 Nov 2002 06:02:33 +0000 (06:02 +0000)]
nobody seems to have objections...
* XHTML 1._0_. delivering xhtml 1.1 as text/html is bogus.
(see <http://www.w3.org/TR/xhtml-media-types/#summary>)
* replaced the also bogus <dl> stuff by normal <p>s
and added some CSS instead.
(embedding CSS compatible to xml *and* html causes the very
ugly hack, you can see in top.html, because in HTML <style> contains
CDATA and in XHTML PCDATA, see also
<http://lists.w3.org/Archives/Public/www-html/2002Apr/0053.html>)
* fixed some encoding issues.
* removed the weird empty lines at the beginning of every document
* added CONTENT_LANGUAGE parameter for pt-br
Set the appropriate mime response headers for negotated type-map
Body: responses.
wrowe's review;
My earlier concern doesn't apply. Type maps have always required the
author to be explicit in all header fields to be negotiated; if the
headers aren't given in the type map file the resulting redirect
(for non-body maps) is broken anyways. As to my other concern, the
charset should only exist if the Content-Type was given.
Wilfredo Sanchez [Mon, 25 Nov 2002 04:17:11 +0000 (04:17 +0000)]
Don't use GNU tar.
Problem here is that GNU tar creates tar archives which are not POSIX-compliant and cannot be unpacked using a POSIX tar program.
Typical oopsie is a developer uses binbuild to pack up an httpd release on a system which does not include gtar in the standard distro, but it's there on the developer's computer. User downloads it and find that it won't unpack with standard tar, even though it claims to be a tar archive. Blah.
Win32 builds populate include from modules/foo/*.h and os/win32/os.h.
Eliminate the noise. However, I'm itching to simply move those 'standard'
module headers back down to httpd-2.0/includes/ ... any objections?
Keep track of the changes. My guess (although we will see in practice)
is that 'new stuff' will stay here in HEAD, while problems will probably
linger in APACHE_2_0. History will be relative to the tree one is viewing.
Graham Leggett [Fri, 22 Nov 2002 12:51:02 +0000 (12:51 +0000)]
Replace APU_HAS_LDAPSSL_CLIENT_INIT with APU_HAS_LDAP_NETSCAPE_SSL
as set by apr-util in util_ldap.c. This should allow mod_ldap
to work with the Netscape/Mozilla LDAP library.
PR:
Obtained from:
Submitted by:
Reviewed by:
Graham Leggett [Fri, 22 Nov 2002 12:35:44 +0000 (12:35 +0000)]
Reorder the definitions for mod_ldap and mod_auth_ldap within
config.m4 to make sure the parent mod_ldap is defined first.
This ensures that mod_ldap comes before mod_auth_ldap in the
httpd.conf file, which is necessary for mod_auth_ldap to load.
PR:
Obtained from:
Submitted by:
Reviewed by:
Jeff Trawick [Fri, 22 Nov 2002 12:23:10 +0000 (12:23 +0000)]
Fix critical bug in new --enable-v4-mapped configure option
implementation which broke IPv4 listening sockets on
systems with IPV6_V6ONLY socket option.
That option should only be done on IPv6 listening sockets.
Submitted by: hiroyuki hanai <hanai@imgsrc.co.jp>]
Reviewed by: Jeff Trawick
Branch tag APACHE_2_0_BRANCH now contains Apache 2.0 development.
Persist cvs HEAD as Apache 2.1.
After discussion at AC, a number of individuals including Sander, Will,
Justin and Rich will begin reverting the appropriate changes from the
APACHE_2_0_BRANCH, while continuing their documentation and development
of the Authorization reorganization on this branch.
Jeff Trawick [Thu, 21 Nov 2002 23:53:21 +0000 (23:53 +0000)]
mod_autoindex: AddDescription directives for directories
now work as in Apache 1.3, where no trailing '/' is
specified on the directory name. Previously, the trailing
'/' *had* to be specified, which was incompatible with
Apache 1.3.
Paul J. Reder [Thu, 21 Nov 2002 21:52:47 +0000 (21:52 +0000)]
Fix for PR 14556. The expiry calculations in mod_cache were
trying to perform "now + ((date - lastmod) * factor)" where
date == lastmod resulting in "now + 0". The code now follows
the else path (using the default expiration) if date is
equal to lastmod. [rx@armstrike.com (Sergey), Paul J. Reder]