ghostscript5 package as (a) we should be encouraging people to use the
packages collection where possible, and (b) it messes up the package
depencies if we don't, to the extent that pkg_delete will have problems.
Specify the working directory name for pre-requisite packages the same
way that Hubert did (using ${WRKDIR:T}). This works both with and
without OBJMACHINE set.
Change the way that paper size is selected. All packages that use
a paper size definition should use a PAPER_SIZE definition, set in
either the environment or /etc/mk.conf. This can be set to "A4" or
"a4", but defaults to US letter ("letter") if not set. Re-define
the do-build target accordingly.
called NMH_MTA (or a variable of the same name in /etc/mk.conf), use
the value of that variable when configuring nmh.
Does not solve the MTA wars, but makes things more configurable.
I modified the test in the Makefile which checks for a previous
installation of real motif, since it wouldn't work if HAVE_MOTIF was
set in /etc/mk.conf.
- Document NO_IGNORE, __ARCH_OK
- Make __ARCH_OK overridable, so it can be set to YES e.g. for fetching
all distfiles.
- Move ".ifndef(__ARCH_OK): block, so some targets like clean work
in any case.
- Add LOCALBASE and X11BASE to PATH in MAKE_ENV, CONFIGURE_ENV and
SCRIPT_ENV; this fixes PR 4573
We need some dependency here, and it's either on ${LOCALBASE}/bin/gs
to pick up the ghostscript 5 package's gs binary, or simply gs, to
pick up any gs binary in the path. The latter was chosen for the
moment.
/usr/X11R6/include/X11/Xaw3d and not /usr/X11R6/include/X11/X11/Xaw3d.
Also clean up the PLIST so that it doesn't contain the static library,
whcih isn't built or installed.
Make the linked man pages (zipcloak.1, zipnote.1 and zipsplit.1) into
symbolic links, as, with MANZ set, the original zip.1 manual page
isn't compressed until after the post-install: target is made, which
means that a link is attempted to a non-existent file.
directory - see the post-install target), so use a MAN1 definition to
make sure that any compression takes place, using the usual package
mechanism. Fixes a problem pointed out by Charles Hannum.
bsd.port.mk, but this time use a definition before the inclusion,
which stops <bsd.own.mk> defining its own install target when
using the package system.
This makes the package system take notice of the definitions in
/etc/mk.conf