Closed ghost closed 2 years ago
I can apply your patch. But if you would do it as a GitHub pull request your name would be in the history.
The patch is only a proposal and does not represent a complete solution. Are such defines desired in the benchmarks? How should the defines be handled? Were should they be documented? How should they be integrated into the build system?
I can help with this, but this was not my initial intention here. I just wanted to point these issues out.
Ok, I understand. You proposal then would need a define at compilation, otherwise it would not compile, right? Maybe a default define that can be overridden. But this still is very much a workaround :-(
My comment on pull request was not just for this patch, but in general to have the history who contributed what. Looks like this gets lost when I apply your patch with git apply.
The function bitcount_btbl_bitcnt contains an issue concerning endianness.
A potential fix using a define "BIGENDIAN" can be found in the patch: 0006-fix-issue-with-endianness-and-missing-semicolon-for-.txt