The existing ebuild uses a really crazy hack for cross compiling which
may have worked a few versions ago but it doesn't now. The root issue is
that Mozilla mixes up the meaning of "host" and "target" so give in to
their stupid and setup the environment with their meaning.
The configure script claimed in a comment:
In Mozilla, we use the names $target, $host and $build incorrectly,
but are too far gone to back out now. See Bug 475488:
- When we say $target, we mean $host, that is, the system on which
Mozilla will be run.
- When we say $host, we mean $build, that is, the system on which
Mozilla is built.
- $target (in its correct usage) is for compilers who generate
code for a different platform than $host, so it would not be used
by Mozilla.
I'm inclined to smack someone with a stick.
Lots of things here, wget for example is not strictly related but
updating mit-krb5 required pulling in a multilib version of the openssl
library which in turn impacts how other dependencies work. The new libev
and libverto dependencies are the ones directly required by mit-krb5.
Packages updated:
app-crypt/mit-krb5
dev-libs/libev
dev-libs/libverto
dev-libs/openssl
net-misc/openssh
net-misc/wget
virtual/krb5
Fixes compatibility with a new portage patch that filters EROOT from the
ebuild environment. This newer version sets EROOT as needed in each
phase function so it is safe from filtering.
The PORTAGE_USERNAME and PORTAGE_GRPNAME variables are not exported by
default. The SDK uses them to get portage to run builds as the
developer's uid/gid instead of portage:portage. When running outside of
the SDK use the default portage user and group.
It appears turning off this flag does not actually prevent perl from
trying to build against gdbm. Enable the flag so dependencies are
consistent with perl's actual behavior.
The pkg-config wrapper re-implemented SYSROOT support because it didn't
work correctly in pkg-config before 0.24. That was released years ago so
it is time to update, custom code broke when paths were outside of /usr
so libraries installed to /lib64 broke things.
With this fixed we don't need to generate our own wrappers.
Upstream bug: https://bugs.gentoo.org/show_bug.cgi?id=517530
