Just in case the Makefile is incompatible, remove the fluff when we rebuild

consolidate hints/instructions about using APR (and maybe APR-util) with httpd

* APR-UTIL trunk is dead. Point to the latest stable branch.

A stab at libapreq build integration, still disabled by default.

disable libapreq stuff until it works generically

* buildconf: Pass through APR source dir to apreq's buildconf.

hook up srclib/libapreq to the build system

Change file touching order for util_expr to be in sync with Makefile dependency. The existing order lead to regeneration.

Replace ap_expr with a parser derived from mod_ssl's parser. Make mod_ssl use the new parser. Rework ap_expr's public interface and provide hooks for modules to add variables and functions. The Netware and Windows build files still need to be adjusted

prevent an error message if an old autoconf without AC_PROG_CC_C99 is used but print a meaningful warning instead

Add support for an "epoch" in the RPM spec file. This epoch is a discrete number used to determine if one RPM is newer than another, which in turn ensures that v1.2.10 is always interpreted as newer than v1.2.9.

buildconf: Fix error "[: -lt: unexpected operator"

* With APR 2.x we no longer have APR-UTIL. So don't care about it.

* buildconf, configure.in: Don't attempt to configure the no-longer-bundled pcre.

Permit --with-apr=/path/to/apr alone, where the default path will be handled by replacing all /apr path segments with /apr-util (or /aprutil) to handle nearly any developer's path scheme. One rare exception would be /asf/srcs/aprutil/trunk/apr-util or similar where both /aprutil and /apr-util path segments exist. So for example --with-apr=../apr-1.3 would find either ../apr-util-1.3 or ../aprutil-1.3 before looking for srclib/apr-util or ../apr-util. These refer to source trees. httpd ./configure deserves a similar feature to use --with-apr= for an installed apr-util lib tree.

update license header text

Update the last year of copyright.

Solaris build proposal; don't fail on missing .h files within a VPATH build, and don't test trees with -d (simply -f the expected files) in case a tree such as srclib/apr is actually a symlink rather than a true directory in ./buildconf.

Update remaining 2004 copyright notices.

Test for APR and APR-Util one directory bellow httpd too. I like not having to put them inside srclib.

Fix the RPM spec file so that an RPM build now works. An RPM build now requires system installations of APR and APR-util.

removed call to aclocal for pcre (we're now able to build pcre-5.0 without it)

call aclocal before autoconf for srclib/pcre (required for PCRE-5.0)

* buildconf: Remove the autom4te cache from srclib/pcre too, before and after running autoconf.

* buildconf: Point at http:// for both, and the "cd srclib" is not necessary.

Add some magic to buildconf so it will tell you to get both apr and apr-util the first time you run it.

'svn co' would be the important part of the command.

Tell users how to get APR from subversion instead of CVS.

use -z7 in the example cvs checkouts for APR/APR-Util. Makes the checkout go much faster on slow dialup modems...

fix name of The Apache Software Foundation

fix copyright date

begin relicensing httpd-2.1 to Apache License, Version 2.0

* buildconf: Clean autoconf cache after running autoconf too.

update license to 2004.

* buildconf: Ensure that make never regenerates the mod_ssl expression parser files from the lex/yacc sources.

fix oops in prior commit; we don't copy libtool.m4, so no need to remove it

steal some code from APR intended to allow you to properly switch between libtool 1.3 and 1.4+... newer libtool doesn't have ltconfig, so having a stray ltconfig around will confuse some of our config logic This also fixes cp failure on FreeBSD since ltconfig and ltmain.sh are installed with 0400 perms, and when buildconf is run a second time the cp of those files would fail.

Add code to buildconf that produces an httpd.spec file from httpd.spec.in, using build/get-version.sh from APR. PR: Obtained from: Submitted by: Reviewed by:

Removing rpm stuff from buildconf until it becomes portable PR: Obtained from: Submitted by: Reviewed by:

Added an RPM build script PR: Obtained from: Joe Orton <jorton@redhat.com> Submitted by: Reviewed by:

even users of libtool 1.3.x want out-of-tree apr builds to work

finished that boring job: update license to 2003. Happy New Year! ;-))

Make buildconf work with out-of-tree apr and apr-util by adding --with-apr and --with-apr-util flags. Submitted by: Thom May <thom@planetarytramp.net> Reviewed by: Justin Erenkrantz (with minor stylistic nit)

PR:

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.

Ensure that we toss the autom4te.cache directory before running autoconf.

Add support for ${AUTOHEADER} and ${AUTOCONF} in buildconf (apr and apr-util already support this convention.)

Update our copyright for this year.

APR-util has renamed buildconf.sh to buildconf to match APR and httpd-2.0.

Note the change to the CVS server in the buildconf script.

give the poor users (i.e. us) a clue that something might be wrong when buildconf fails in subdirectories

Abort apache buildconf if apr buildconf or apr-util buildconf fails. Otherwise, the error message from apr or apr-util (e.g., invalid libtool version) is lost in the stream of output.

Untangled the buildconf script and eliminated the need for build's aclocal.m4, generated_lists, build.mk, build2.mk, and a host of other libtool muck that is now under srclib/apr/build. More to follow.

We don't have config.guess until later in the build cycle, so change the test to use uname. with this change a clean CVS copy will build using ./buildconf;./configure;make exactly like it should. Thanks Sascha.

Add Convert-to-BSD-Makefiles-On-The-Fly system Note 1: Only needed for BSD/OS. Note 2: Invent better heuristic (timestamp checks for build/*.mk). Note 3: Conversion is currently one-way Note 4: Don't package on BSD/OS.

Always delete generated_lists. This should cure problems with makes which don't handle the .PHONY target correctly.

Less verbose solution.

Tru64 make uses the wrong directory for includes.

Tru64 make gets confused by null argument!

Allow arguments to buildconf. Makes cvsclean, etc. a bit easier.

Just make 'buildconf' run on BSD (there is a space after the hash-bang) and added bogus INSTALL file plus some blurp in readme.mpm to point out that you really need libtool and autoconf (which need m4). On some platform's I found that I really HAD to use 'gmake' as the normal make would not grok it. So I am wondering if adding something like # Work out a correct 'make'. The issue is that we should use # 'gmake' which is called 'make' on some platforms, and 'gmake' # on some others. # GUESS=${MAKE:-make} if `(${GUESS} -v | grep "GNU Make") > /dev/null 2>&1 `; then # Looks like gmake. else GUESS=gmake if `${GUESS} -v | grep "GNU Make" > /dev/null 2>&1`; then echo Warning: Using 'gmake' instead of 'make'. else echo No 'gmake' found. Please set the MAKE environment echo variable first. exit 1 fi fi MAKE=${GUESS} export MAKE to 'buildconf' would be appropriate to detect this ?? Dw. PR: Obtained from: Submitted by: Reviewed by:

These changes are committed together, because they depend on each other. - shared modules can be built in the tree - added support for --with-layout, uses APACI's config.layout - working 'make install' - working 'make depend' - working Pthreads checks - buildconf replaced

Fix IFS handling. This has the nice side effect that we do not need tr anymore.

--include-deps only seems to take advantage of pregenerated dependencies, and makes the build process refuse to generate them, so this removes it. It should be used when making a distribution, though.

We don't need the capability to build an acconfig.h anymore; AC_DEFINE provides everything we need for equivalent functionality. This change deletes the intrastructure that built acconfig.h.in before.

Don't actually run ./configure in buildconf, it wastes lots of time if we want to use ./configure's flags, among other things.

We actually don't have to remove config.cache, because AC_CONFIG_SUBDIRS redirects APR's config.cache to point to Apache's.

Put back the removing of config.cache. Also add a -f flag to the rm to eliminate a warning.

Make Apache's configure script run APR's.

We need to run autoheader to compile APR.

Build APR, too.

The second part of the big autoconf patch, which actually adds autoconf building to the tree.