[Bug 238887] devel/py-llfuse: fix build with GCC-based architectures
bugzilla-noreply at freebsd.org
bugzilla-noreply at freebsd.org
Sun Jun 30 04:09:21 UTC 2019
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=238887
Kubilay Kocak <koobs at FreeBSD.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |python at FreeBSD.org
Status|New |Open
Assignee|ports-bugs at FreeBSD.org |pkubaj at FreeBSD.org
Keywords| |needs-patch, needs-qa
--- Comment #2 from Kubilay Kocak <koobs at FreeBSD.org> ---
Reporter is committer, assign accordingly
@Piotr -Werror should be patched out here too (it shouldn't be used/enabled in
packaged/distributed/released code.
setup.py does the following:
DEVELOPER_MODE = os.path.exists(os.path.join(basedir, 'MANIFEST.in'))
if DEVELOPER_MODE:
print('found MANIFEST.in, running in developer mode')
<snip>
compile_args.append('-Werror')
compile_args.append('-Wfatal-errors')
The conditional for DEVELOPER_MODE is too widely-scoped, as source
distributions (sdists) are released to end-users, and the presence of a
MANIFEST.in shouldn't mean DEVELOPER_MODE.
A slightly better form might be to check for the presence of a git checkout,
though this is also problematic, as users can depend on packages via their git
URL's.
The check should probably be an explicit opt-in via the presence of an
environment variable DEVELOPER_MODE=yes or similar.
Either way, I'd patch the conditional to be False, along with the patch
provided fixing the warning.
If you can, please report this '-Werror should be opt-in, not based on
MANIFEST.in' issue upstream, with a PR if you can manage it
Note also, both the build (warning=error) fix, and the -Werror removal are:
Approved by: portmgr (blanket(s): build fix, ports compliance)
--
You are receiving this mail because:
You are on the CC list for the bug.
More information about the freebsd-python
mailing list