[llvm-commits] r153205 is still breaking CMake builds and win32 builds, I've reverted it, and the three follow-up patches
dmalyshev at accesssoftek.com
Thu Mar 22 09:00:36 PDT 2012
Thank you for your offer of help me.
Andrew Kaylor also wrote me about these problems. The first looks as I missed null pointer checks for JMM reference, I'm working with it now.
Unfortunately, these tests not executed when I run regress tests. I'll look why this is happening.
It is a strange errors, because These tests should not run on Windows, ExecutionEngine/MCJIT/lit.local.cfg contains the following filter:
if root.host_os in ['Win32', 'Cygwin', 'MingW']:
config.unsupported = True
I'll solve this problem, after will fixed the previous errors.
From: Chandler Carruth [mailto:chandlerc at google.com]
Sent: Thursday, March 22, 2012 9:52 AM
To: Danil Malyshev; Commit Messages and Patches for LLVM
Subject: Re: r153205 is still breaking CMake builds and win32 builds, I've reverted it, and the three follow-up patches
First off, sorry for the blind revert, and sorry for the direct email, but the mailing list is down and we really need to get the bots back up and caught up. The windows bots are half a day behind already.
It's also not clear what all will be required to fix the CMake tests, and so I couldn't just fix it directly. I tried a few things, and wasn't confident they were the correct solutions. I'm happy to work with you to figure out the correct fix, feel free to ping me whenever with questions.
I've reverted r153205 and the subsequent entangled patches r153207, r153208, r153221. The revert is r153241.
Linux CMake failures:
Let me know if I can help in specific ways. I can also take a stab at fixing this later and re-instating the patch.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the llvm-commits