[issue31354] Fixing a bug related to LTO only build

[Charalampos Stratakis]

Charalampos Stratakis <cstratak at redhat.com> added the comment:

Hi Ned,

I recently pushed a fix on the master and 3.7 for this exact issue: https://bugs.python.org/issue35351 but it builds on top of https://github.com/python/cpython/pull/10922 which is not yet in 3.6.

Thus 3.7 is fine for the rc cutoff.

However there is another side effect of that change with the linker flags also leaking: https://bugs.python.org/issue35257 but it proved to be more complicated than what I initially thought to fix it.

PR here: https://github.com/python/cpython/pull/10900

The issue is that I figured it wouldn't make it to the 3.6 branch as it would require extensive review due to the complexity of the build system and the requirement for extensive review to catch any possible regressions, thus it would be late for the rc cutoff.

It would take some days to figure out everything I believe. Would it be possible to push that after rc1? Or maybe not fixing that would be better, deadline wise? I consider this an important issue but some more feedback is always welcome.

Side note on your comment: The compiler used (CC) also seems to leak into distutils instead of the system default CC, on linux with 'python3 setup.py build'.

----------

_______________________________________
Python tracker <report at bugs.python.org>
<https://bugs.python.org/issue31354>
_______________________________________