Submitted by: rasmus@madhaus.utcs.utoronto.ca


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76556 13f79535-47bb-0310-9956-ffa450edef68

than those without wildcards at the same quality value.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76555 13f79535-47bb-0310-9956-ffa450edef68

Corrected 1st parameter to siglongjmp() and sigsetjmp() calls to be declared
as sigjmp_buf; was incorrectly declared jmp_buf.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76554 13f79535-47bb-0310-9956-ffa450edef68

Editorial change at the header to connect_handler()


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76553 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76552 13f79535-47bb-0310-9956-ffa450edef68

Switch uid to the uid in the User config file directive, if run as root
from an inetd.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76551 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76550 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76549 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76548 13f79535-47bb-0310-9956-ffa450edef68

Fergot to put in the -Wall patch - chuck


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76547 13f79535-47bb-0310-9956-ffa450edef68

Add NoCache example to the other mod_proxy options in httpd.conf-dist


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76546 13f79535-47bb-0310-9956-ffa450edef68

Submitted by:	Howard Fear <howard_fear@pageplus.com>
I've finally tracked down the persistant core dumps we were experiencing
on Linux.  Attached is a patch for 1.0.5.  I'll do one for 1.1 later
tonight.

The problem is in mod_include.c.  The macro that reads the file
fcloses a pfopen'ed file.  This results in closing the file twice,
once during the fclose and again during the resource cleanup.
This apparently sends the linux libc.5.3.x fclose into a SIGSEGV,
its not smart enough to realize that the fd has already been closed.

This happens for every page served through mod_include.  This can be
a serious problem for sites that set up the include handler for
.html files.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76545 13f79535-47bb-0310-9956-ffa450edef68

Submitted by:	Chuck Murcko
Obtained from: Troy Morrison <spiffnet@zoom.com> (CONNECT patch)
This patch adds a CONNECT method SSL proxy, and a NoCache config file
option, which is a list of hosts or domains not to cache.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76544 13f79535-47bb-0310-9956-ffa450edef68

Submitted by:	Cliff Skolnick
Obtained from:  jk@tools.de
mod_auth.c already accepts any user id, if there are
no "requires" directives found for a directory or
location.

Since "requires" directives can occur inside a
<Limit> directive it should also accept any
user id if there are "requires" directives, but
done matches the actual request method.

The same problem exists in mod_digest.c


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76543 13f79535-47bb-0310-9956-ffa450edef68

Submitted by:	Cliff Skolnick
Obtained from:  jk@tools.de
Sending a SIGTERM (as user root using /bin/sh as
shell) doesn't terminate apache under Solaris 2.5
(but other SVR4 systems are probably affected, too).

After sending the SIGTERM, /usr/proc/bin/psig on the
pid of the apache master process shows
    TERM        blocked,caught  0

As far is I understand it, the problem is caused by
jumping out of a signal handler using setjmp/longjmp
instead of sigsetjmp/siglongjmp.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76542 13f79535-47bb-0310-9956-ffa450edef68

Submitted by:	Cliff Skolnick
Obtained from:  jk@tools.de
When the server listens to multiple ports (via
the httpd.conf 'Listen <port>' directive), a
SIGHUP sent to the main server doesn't call
reclaim_child_processes.  The status of the killed
child processes is collected much later, after the
server is fully restarted.  Also an error log
entry for the SIGHUP is missing in logs/error_log.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76541 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76540 13f79535-47bb-0310-9956-ffa450edef68

explanations as to the functionality of each module.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76539 13f79535-47bb-0310-9956-ffa450edef68

This patches several things:

1) brings back complete URL's in Location: headers
2) makes the default "base" into the URI for the mapfile
   rather than ""
3) pointinrect() now can take any two opposing corners in
   any order
4) closes potential for an infinite loop in mod_imap.c
5) if a mapfile isn't found, returns NOT_FOUND instead of SERVER_ERROR.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76538 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76537 13f79535-47bb-0310-9956-ffa450edef68

Make the message about server version strings get noticed


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76536 13f79535-47bb-0310-9956-ffa450edef68

The HTTP specification says that it is legal to merge duplicate
headers into one.  Some browsers that support Cookies don't like
merged headers and prefer that each Set-Cookie header is sent
separately.  Lets humour those browsers for now.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76535 13f79535-47bb-0310-9956-ffa450edef68

to the SERVER_VERSION define.  Most of them don't meet the HTTP/1.0 spec.
Added a helpful message explaining what to do to meet the spec.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76534 13f79535-47bb-0310-9956-ffa450edef68

Submitted by: Rob Hartill


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76533 13f79535-47bb-0310-9956-ffa450edef68

requests that aren't protected with digest authentication. This was
apparently causing some problems with certain setups.

Submitted by: jk@tools.de


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76532 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76531 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76530 13f79535-47bb-0310-9956-ffa450edef68

I have the following patches to cleanup some -Wall warnings.
At least one real bug in the version of mod_auth_pg95.c that
I had.

/export/pub/apache/dist/contrib/modules/mod_auth_pg95.c is also updated - chuck


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76529 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76528 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76527 13f79535-47bb-0310-9956-ffa450edef68

This change allows setting SO_LINGER socket option for SVR4s with
broken socket libs using -DNEED_LINGER


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76526 13f79535-47bb-0310-9956-ffa450edef68

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76525 13f79535-47bb-0310-9956-ffa450edef68

since rprintf() appeared.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76524 13f79535-47bb-0310-9956-ffa450edef68

1) fixed IP address formation in host2addr()
2) fixed SIGALRM on big cache cleanup
3) fixed temp files #tmp not removed
4) changed PF_INET to AF_INET in socket() calls


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76523 13f79535-47bb-0310-9956-ffa450edef68

Changed to turn off default setting for ScriptAlias (closing a potential
security hole)


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76522 13f79535-47bb-0310-9956-ffa450edef68

These files will now be distributed *separately* from Apache.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76521 13f79535-47bb-0310-9956-ffa450edef68

obsoleted.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76520 13f79535-47bb-0310-9956-ffa450edef68

Added printenv as a simple Perl CGI script
Added back test-cgi so the perms could be set appropriately (444, so that
  the adminitrator needs to enable it specifically, preventing it from
  appearing out of the blue to an unsuspecting admin)


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76519 13f79535-47bb-0310-9956-ffa450edef68

Removed because they have not been touched or supported by the Apache
group, and represent possible security holes we wish to not be associated
with.

As per Message-ID <Pine.SGI.3.91.960423195138.579v-1000000@fully.organic.com>
"Re: 1.1b2" from Brian Behlendorf, Tue 23 Apr 1996 20:06:28 -700


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76518 13f79535-47bb-0310-9956-ffa450edef68

Obtained from:  "ken@spry.com"?

Removed hole through which carriage returns could be passed to shell scripts.


git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/1.3.x@76517 13f79535-47bb-0310-9956-ffa450edef68