| Summary: | dev-lang/swig-3.0.10: import wrapper for dev-util/lldb guesses wrong package name | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Michał Górny <mgorny> |
| Component: | Current packages | Assignee: | Tim Harder <radhermit> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | ave, dschridde+gentoobugs, holger, llvm, radhermit, rr.underwood94, scheme |
| Priority: | Normal | Keywords: | REGRESSION |
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| URL: | https://github.com/swig/swig/issues/769 | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
I will need a while to test the older version but it seems that it is the linked regression in 3.0.9. AFAICS upstream is not keen on fixing it before 3.1, so I'll probably end up forcing the older version in lldb ebuild. Yep, confirmed 3.0.8 works fine. I suggest to change the dependency string to python? ( || ( <dev-lang/swig-3.0.9 >dev-lang/swig-3.0.10 ) ) for lldb 5.0.0 and above, swig is compatible again since 3.0.11, in accordance with https://reviews.llvm.org/D33409. lldb-4.0.1 would need this patch to be backported though. The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=4cea56b7d3193645fe884ff7e7241ecc5867a1e0 commit 4cea56b7d3193645fe884ff7e7241ecc5867a1e0 Author: Michał Górny <mgorny@gentoo.org> AuthorDate: 2018-03-28 10:09:23 +0000 Commit: Michał Górny <mgorny@gentoo.org> CommitDate: 2018-03-28 10:10:58 +0000 dev-util/lldb: Allow dev-python/swig >= 3.0.11 in 5+ Closes: https://bugs.gentoo.org/598708 dev-util/lldb/lldb-5.0.1.ebuild | 4 +--- dev-util/lldb/lldb-6.0.0.ebuild | 4 +--- dev-util/lldb/lldb-6.0.9999.ebuild | 4 +--- dev-util/lldb/lldb-9999.ebuild | 4 +--- 4 files changed, 4 insertions(+), 12 deletions(-) |
swig generates some crappy code in /usr/lib64/python2.7/site-packages/lldb/__init__.py: from sys import version_info as _swig_python_version_info if _swig_python_version_info >= (2, 7, 0): def swig_import_helper(): import importlib pkg = __name__.rpartition('.')[0] mname = '.'.join((pkg, '_lldb')).lstrip('.') try: return importlib.import_module(mname) #... Now, since __name__ is just 'lldb', pkg is null and it tries to import top-level '_lldb' (instead of 'lldb._lldb'). I guess the hackery they're doing is supposed to be handle importing from named .py files but doesn't cover the special case for __init__.py.