Using set -o pipefail we can detect when gzip -d and friends fail when
decompressing archives. This is better than a partially extracted
archive leading to failures to patch or build later.
ok phessler@ landry@
shared libraries don't record a dependency on libc so normal pkg_add
mechanisms don't result in everything using libc getting updated
following this change)
With perl 5.40, there's a new warning when calling an "import" method
that doesn't exist. Previously the error was ignored, although still there.
Reported and reminded by rsedowski@
mention the DIAGNOSTICS section of the bsd.port.mk(5) manpage when the
packing-list doesn't match installed pkgs vs what's in the portstree.
from espie@
this was used earlier, but lost when changing to PATCH_DEBUG by default.
Without this, if a diff appears to be reversed, patch asks whether to
assume that you used -R, with a default 'y'. If there's no tty, as is the
case with a bulk build, the default is accepted, resulting in problems
(potentially nasty in the case of a security fix which was applied via
a patch and later committed upstream) if 'cvs rm' was missed.
Sync option order between PATCH_ARGS and PATCH_DIST_ARGS while there.
Grepping ports logs doesn't show any current ports where this will cause
unexpected changes in behaviour.
ok tb
To be reconsidered when we move to lang/gcc/11 or newer, which
supports -mbranch-protection. The fix isn't complete since some ports
may use BTI-incompatible fortran code through shared libraries.
ok tb@ phessler@ pascal@ sthen@
$state->{default_prop} contains the already built object, so passing it
to a constructor will not end up in a new host with the desired properties
from espie
The goal is to avoid editing modules.inc by removing the MODGO_VERSION
variable from that file by the go ports maintainer when using it like:
make modgo-gen-modules > modules.inc
OK: paco@ sthen@
rsadowski