On Wed, Apr 20, 2022 at 3:16 PM Paul Smith <psm...@gnu.org> wrote:
>
> On Wed, 2022-04-20 at 10:30 -0500, Noah Goldstein wrote:
> > Is this project still accepting patches from new contributors? If so,
> > what is the review process?
> >
> > I have my FSF agreement signed.
>
> We do accept changes from anyone.  However, we do not accept all
> changes.  Bug fixes are always welcome.  GNU make is a critical
> infrastructure project with a LOT of portability issues, dependency
> issues, etc. so not all ideas for improvements/new features will be
> accepted.

Hi Paul,

The change I am looking at are 2-fold.


1) Replace unnecessary calls to strcmp with memcmp (i.e cases
where string length has already been compared)

2) Remove the `isupper()` branch in STRINGI_HASH. This is just
objectively slower.

The motivation is I'm seeing pretty high CPU usage in hash_find_slot
and `strcmp` when making the GLIBC benchmark suite.

>
> If you are planning on working on a larger change that may have
> backward-compatibility, etc. implications it's best to discuss your
> idea first, on the bug-m...@gnu.org mailing list.
>
> If you have a patch for a fix, you can email it (as an attachment
> please) to the bug-m...@gnu.org mailing list.
>
> You can also file a bug (even if the change is not technically a bug
> fix) then attach the patch to it on Savannah:
> https://savannah.gnu.org/projects/make/
>
> Cheers!

Reply via email to