This reverts commit 5574df3549.
I also can't reproduce the problem anymore; discussion: #41312.
Fixes#53569, fixes#53948. (Vulnerabilities in old curl.)
Otherwise docutils propagates python2.
Since we install umpv (a python wrapper script around mpv) python is
pulled in as a runtime dependency of mpv.
Youtube-dl already uses python3 so we can avoid a full python version
in the closure.
There are definetively a few Linux-only and Unix-only Perl packages, but `meta.platform` looks like it set at random, presumable by copy-pasting from code blocks nearby.
Many packages marked as Unix-only are in fact cross-platform and works well on Windows (and presumable on every platform where Perl runs)
So let's reset those per-package platform limitations, until we will have a reliable source of information for `meta.platform`.
Also removed empty `meta.maintainers` and `meta.license = "unknown"`
This also covers "androidStudioPackages.preview".
I believe that the name is confusing (is it from the beta or dev/canary
channel?) and with this error message it should be obvious how to update
ones configuration.
Two reasons for this:
- more fine-grained space/functionality tradeoff
- preparation for the sage 8.6 update, which finally doesn't need a
downgraded gap anymore but does break when unexpected (non-standard)
packages are installed. Details: https://trac.sagemath.org/ticket/26983
The proper way to deal with gap packages would be to create a package
set, package each one individually and have something like gap
equivalent of `python.withPackages`. I am not willing to do that
however.
There are some starts of a `make install`, but nothing complete yet.
Upstream now ships a `libgap` as a replacement of the custom one used by
sagemath.
As a data package `buildPackages.cacert` suppose to be equal to `hostPackages.cacert`.
But `hostPackages.cacert` causes needless build of `hostPackages.python` and other packages, which takes time and requires them to support cross-compilation.
This fixes a bizarre regression in the NextPNR GUI that causes the
background of the OpenGL floorplanning window to become 'transparent' on
certain platforms, greatly limiting its utility. However, QT 5.11 seems
to work just fine here.
Signed-off-by: Austin Seipp <aseipp@pobox.com>
