add post install script target

add go

add partial support for flavors.

more fixes

remove dead code

fix the white space

add erlang handler

add meson

update cross toolchain and os reported during builds.

Remove LIBDIR from build env variable

This is leaking LIBDIR information (which is defined in base share/mk/*) to
the build environement with a value which is not really suitable for ports
(/usr/lib) which may have an impact on some build system forcing the maintainer
to override the value.

remove openssl extension in favor of ssl.mk

add ssl module

support sed for substitutions.

add samba config

skip the python extension as it freaks out with USE_PYTHON and USES=python:3 at the same time.

revert back to versioning.

temporarily go back to new version

revert to original

change dependencies again.

fixup the path if the new syntax is used for installing LIB depends that are missing.

refactor var handling, use sinclude

osrel and arch combo broken

remove cd mount support

migrate all-depends-list to a script

attempt to add group blacklist

add ypldap to list

export several OS related vars

remove useless var

make wayland a valid category.

fix a syntax bug

replace COPYTREE_SHARE and COPYTREE_BIN with faster versions.

include default versions always

attempt to teach bsd.mport.mk to require /usr/mports prefix on dependencies.

go to NO_MANCOMPRESS

add kde4 handler from freebsd for the lumina port

remove old ghostscript hack

support alternate configure commands. (for use with waf / python)

try deleting these crazy soft depends

drop freebsd compatibility with destdir. this gets in the way of our fake system

revert the last change and fix wrksrc instead

disable chroots for now. we are not actively using this feature

add tk extension to list

update all ext list

add jpeg extension

remove fpc from ports

fix bug in patch fetching where it never thinks the patch was downloaded correctly

switch to @dir

fixup

refactor fetching

add retry back

get rid of soft make args on makesum target

add keywords section

rename HASH_FILE -> DISTINFO_FILE as it is more specific. We went to HASH_FILE when we dropped MD5

remove the complex ignore logic and just die. This seems to work with postgres at least

add tex extension

support silent patches. Add LIBS to make environment. Also, support USES_POST compatiblity with FreeBSD ports system.

disable debug before ssp handling.

add openldap extension

use a var for the _TARGETS list

dont worry about deprecation

cleanup more names and minimize diff

load commands earlier.

drop soft makeflags. refactor os version check

remove testing comment

add gettext tools/runtime to extensions list

add support for cpe if defined to package

revert the PKG_IGNORE_DEPENDS change to fix the grep error we see.  Print a nice message when package create fails. Also, remove the package cookie so we can re-run the target without issue

drop ul compat

refactor x dependencies

always load x extension on xorg cat

switch to tu and tl from :U and :L modifiers

make USE_MOTIF also work.

add desthack.

change backup site

move compatibility hack above loading of extensions

add a fake-man-compress target that pre-compresses all our man pages if they're not already compressed.

This allows us to put man pages in the plist now like FreeBSD does.

remove legacy check license

add display extension

check the license

start work on importing FreeBSD license framework with our own mport license system. They have a few features that we do not from the experience of planning after we and openbsd did this.

add CONFIG_SITE to configure environment

add mate category

rework patch fetching

man stage compatibility with FreeBSD ports

remove mbone category. most of these ports don't work well on newer midnightbsd releases anyway and it's stale.

all in

add ssp and kmod modules

explicitly set make command.  Due to order of loading commands.mk, we need to force this when using gmake or fmake.

As we're not always using USE_GMAKE anymore, just use one path for run-build target. USES=gmake was not working properly.

upgrade desktop file format

fixup some properties.

add gmake extension that's compatible with USES= and EXTENSIONS.

aha! catch the new config options so we run make config when we want to

MAKE JOBS SAFE is default now.

Switch to new localbase file.

fix typo in comment.

add fuse depends handling

add more uses crap

include the extension using a separate mechanism until we get this sorted

add pkgconfig to the list

Update extensions for various ports.

This will break things before it gets better.

add charsetfix from freebsd.

make some changes for cmake.  introduce *_REL directories for plists and cases we need relative paths to DOCS, DATA, ... directories.

fix dos2unix substitution.
cleanup options.mk

Turn on random master sites for magus.

improve dos2unix

Refactor bsd.mport.mk to include man pages stuff rather than embedding it in this file.

fix broken package depends list line. this took awhile to track down.

os version detection hack

Force clean on a depends install.  This loses something in the cached case, but it's necessary.  It should only happen when a depends is actually missing.

use limited clean depends so we don't have to go through useless cleaning that is slow

format fixes, add CPPFLAGS< LDFLAGS CPP to make env, kill some legacy stuff like INSTALLS_SHLIB

change the behavior of CONFIGURE_TARGET. It will try to use it with --build now if it's a recent autotools setup

add INSTALL_LIB shortcut

This is part of our problem with package names.. suffix was not included.

Default to fedora 8 libs

MD5_FILE -> HASH_FILE.. this is the path to the distinfo file for a port.

Remove md5 hash algorithm from distinfo files.  Only use sha256 and rmd160 now.  This speeds up "verification" slightly for distinfo files  and md5 isn't exactly secure  anyway.

Don't include xz dependancy when os version is greater than 4003 when we introduced it in the base.

Doh!  It was .endif!

Forgot the \

Enable private dist files for magus.

Add glew depends for USE_GL

Reorder licenses.  Put the "special" cases on the last row.

Add several additional GNU style licenses to the list.

add lua category.

Remove icc hack.

change default os version we lie about to 7.0 for current.

fix the patch depends

subversion license.

fix typo

What did intel call it?  Multiplexity?  Yeah...

Execute multiple make jobs in parallel within a single port.

This uses make -jX where X is the number of CPUs on ports that
have this feature enabled to make use of those fancy multicore CPUs.

By default ports are built as they always have been.

Use DISABLE_MAKE_JOBS to turn this off globally (make.conf)
FORCE_MAKE_JOBS turns it on for everything in make.conf except those ports
with explicit MAKE_JOBS_UNSAFE set.

Finally port maintainers will use MAKE_JOBS_SAFE to mark a port as "OK" with
this feature.

Ports can use make or gmake with this option.

This is similar to the feature introduced in FreeBSD ports last year; in fact it's based on their approach.

Since this is used at compile time, it should not have a bearing on our faking or other features.  It's local to the wrksrc only!

add WWWDIR and ETCDIR shortcuts.

WWWDIR is the location to store the packages web files.

ETCDIR is the diretory to store the packages etc files.

Also document them in the pod file

Pull in the old package tools right before fake.mk, this will give the
various extensions time to twiddle things before we setup the depends
targets.

change default openldap to 2.4

Add a third hash algorithm to the fray.  New distinfo files will contain and check rmd160 in addition to md5/sha256.

Tested on ports with and without rmd160 present.

Also remove check for ancient ports.  This was a very old method of describing ports that we've never used.  While it might be vaguely useful when basing ports on those from openbsd, etc. it doesn't makes sense for most cases.

move variables needed by package stuff to the right place.

include old package tools before the comps.  This fixes various breakages.

move fake into a component

remove some mport.query debugging that shouldn't have gotten into cvs.

add kld as valid category.  This is for kernel modules.

Display the various stage messages in bold.

Various fixes.
Add ocaml.
Move KDE4 to 4.1.3.
Make sure that PREFIX is defined before we load extensions.
s/PKGSUBNAME/PKGBASE/g

Move all the documentation at the top to docs/bsd.mport.mk.pod.

Fix bug with PERL_CONFIGURE and PERL_MODBUILD.

Add support for bsd.mport.options.mk.  Make sure you have
bsd.mport.options.mk in /usr/share/mk.  (it is in source for 0.2, will be in
current soon).

Have USE_GECKO autoload bsd.gecko.mk

use PKGBASE is the home install warning.

Update zip to 3.0.

 * Add infozip license type and place the license in Licenses.
 * Set the license in zip port.
 * Use new sourceforge mirror

Document USE_XFCE.

Catch a case we missed earlier.

Suck in bsd.qt.mk for QT4 ports.

Redhat's got a special license for this based on the GPLv2.  The FSF approved of the license, but it is not the GPL.  Create a special license type for this.

Add @rmtry and @stopdaemon to plist generation

These should probably be handled like @dirrmtry eventually in the pkg_install tools. (or mport)

add lgpl3 and opera license types.

All ports using lgpl v3 should use lgpl3.  lgpl v2 and v2.1 should continue to use lgpl

Remove openldap 2.2 reference.  We've never had a port with that version.

bump ldap version

Add mport.inc directory.  This directory contains logical units that
together form the bsd.mport.* makefiles.  metadata.mk and commands.mk have
been created so far.

In the furture there will be a ext directory for things like the xorg or
perl extensions.

Add tk and tcl generic categories. Remove zope and a few others we don't need.

Put install-desktop-entires in the right spot (with the rest of the
install-) targets.
Fix some FreeBSD stupidity.

INDEXING is dead.  Long live PACKAGE_BUILDING.
Cleanup bsd.java.mk, don't shotgun variables, fix logic so that system
doesn't check for installed jdk when PACKAGE_BUILDING is set.

protect against a dependant port being null, resulting in the entire category being depended on.

Document what i did to bison!

Since we only support xorg, why the hell bother with testing for it!

Fix preserve owner on tar extract

Add some fixes for cases where we're using -02 and GCC 4

accidenently lost the PKGBASE definition.

clean up the package depends stuff for the mport tools side.

We need gccmakedep with imake usage too.

Add the MIT license.

Removed the skip for packlists from bsd.perl.mk.  It didn't work anyways.
Make sure that man3's for perl modules don't end up in gen-plist.

makeplist will now run make fake without the fake checking.  This way if you
have a bad plist, you can use makeplist to get a good one.

Make sure that DOCSDIR and MANPREFIX are also corrected when trueprefix is
set.

Add run-configure and run-build targets.  These targets are run by
do-configured and do-build respectfully (the do-targets are thin wrappers).
This allows you to override do-build, do something, and the then run the
normal build process without having to reproduce it in your port makefile.

Broke the fake process out of the monolithic fake-install target into
several smaller targets.  This allows the port maintainer to override the
do-fake target if needed.

These changes are protential unstable.  I believe there to be no
regressions, but until a run magus run can be done, be wary.

add is_interactive to the yaml description.

add a few things to make magus's life easier.

add x11-drivers category.

Fix check-license for empty LICENSE.
Add agg (agragate) license type.

Add sgi license.

Use the new shell version of chkfake.

Don't automatically add infodir or mandir to CONFIGURE_ARGS if
CONFIGURE_ARGS already has the relevent variable.  This fixes the bug with
lang/gcc42.

Add MidnightBSD ftp server to backup list.

Add LICENSE to individual port readmes (HTML)

Fix the MAKE_FLAGS bug properly this time.  At port.post.mk include time
any instance of -f is removed from MAKE_FLAGS and -f is appended to the end.
Care is taken to not break things like '-fno-strict-aliasing'.

Introduce new USE_GCPIO option.  This must be set to use ${CPIO} in a makefile.  It sets an extract dependancy on the new archivers/gcpio port.  This is needed to fix the linux ports as they rely on GNU style -- features.

Cleaned up bugs in new prefix handling.  This should fix problems with
gnustep ports.

Removed reference to unsed variable USE_GETOPT_LONG.
Fixed fake bug in portdocs.
Added unknown license type.
Changed how check-fake works.  It no longer depends on fake, but now make
fake will run the check if MPORT_MAINTAINER_MODE is defined.

Added support for the LICENSE variable.  The current list of valid licenses
can be found with the license-list target.  This list is incomplete and will
probably change.
Renamed check_fake to check-fake to be more consistant with similar targets.

Added check_fake target.

Put MAKE_ARGS at the begining of FAKE_MAKEARGS, so that things can be
overridden for fake.

Various cleanups.  bsd.perl.mk should now work with post/pre properly. Fixed
an edge case in cached-install where the PKGFILE didn't exist but the
package cookie did.

Moved all the perl stuff into a seperate bsd.perl.mk.
Fixed a bug where add-plist-docs wasn't fake-aware.
A debugging line in fix-fake-symlinks crept into cvs, removed.

Make FAKE_SETUP a public variable.  It should be considered read-only, but
it is useful if you need to execute make from a *-install target.

Removed more cruft.  Much remains, and this still has years of FreeBSD
backwards compatibility which we do not need.
makeplist should work better when NO_MTREE is in effect.

fix-fake-symlinks failed when there where no symlinks in the fake dir:
fixed.

Fix bugs with USE_LINUX.  It was set to a run depend, but it should be a
build depend.
Fix .ORDER for parallel builds.

Move makeplist to the end so that all the vars are defined.
Update MTREE_LINUX_FILE to the new location.
USE_LINUX no longer implies NO_MTREE, as we are going to have an mtree file
for each linux-base.

Fixed and cleaned up ldconfig configuration file magic.
Added refake target.

Add KMODDIR to the variables faked.

Make sure that PREFIX is still twiddled with for {pre,post}-install if the
trueprefix FAKE_OPT is set.
Don't delete the package cookie on make deinstall.

Added new options for FAKE_OPTS: trueprefix.  This is for ports that let you
override both DESTDIR and PREFIX.  Most ports don't do this, but if
something is trying to install to $DESTDIR$PREFIX, and you get silly long
paths, this is the option you need.

Fixed RC scripts.  Minor bug fix where MAKEARGS wasn't propogated to
FAKE_MAKEARGS.

Fixes for MANCOMPRESSED.  This needs to get refactored more.

Added makeplist target.  This is currently undocumented, and missing
several features, notably symlink support.

FAKE_TARGET was hard wired to install, it should be ${INSTALL_TARGET}.
Fixed.

Various bug fixes.
imake now works properly.
linux ports now see /compat/linux in their fake dirs.
you can now move FAKE_INSTALLDIR upwards from $WRKDIR, and make clean will
still delete it.

make install now creates the package that make package built.  This is the
first alpha version.

reworked fake system. Somewhat similar to Open's now.  Most ports should
build correctly into their fake dirs now, though we'll have to check all the
ports...

make install is not yet up to speed with the changes, so users are still
advised to use bsd.port.mk.

Fixed a nasty bug that broke pkg_delete.  Our tmp prefix wasn't recognized
as one of the "standard" prefixes, so the packing list was setup to delete
the prefix.  Only problem is that the prefix for the package itself is the
default prefix, which is something like /usr/local or /usr/X11R6.

The very begining of bsd.mport.mk.  Right now there is very little
difference between this and bsd.port.mk.  There is a new target,
tmpdir-install, that instals the port in a temporary directly.  make package
depends on this now, but given that the package target hasn't been updated
to work with the tmp dir, make package is broken.  Also, TMP_INSTALLDIR
needs to be moved up to the top, with the rest of the vars.

The port makefile in general needs to be refactored.  It is close to 6000k
lines long, and real bitch to work with.