Re: Ping Re: Implement C23 log2p1

[Paul Zimmermann <Paul.Zimmermann@inria.fr>]

       Dear Joseph,

I have already given my opinion on this patch (after sending several hours
to exercise it):

https://sourceware.org/pipermail/libc-alpha/2024-April/156007.html

I now let other people comment.

Paul Zimmermann

> Date: Mon, 29 Apr 2024 13:34:08 +0000 (UTC)
> From: Joseph Myers <josmyers@redhat.com>
> 
> I'd like to ping this patch 
> <https://sourceware.org/pipermail/libc-alpha/2024-April/155960.html>, and, 
> in particular:
> 
> (a) the actual content of the patch, which no-one has commented on 
> (instead preferring to suggest other things to add to it);
> 
> (b) the questions of what goes in such patches, where I argue that we 
> should not expect libm-test-ulps updates for more than one architecture, 
> or additional test inputs based on searches for worst-case inputs, to be 
> included in such a patch: that it should be sufficient to include 
> libm-test-ulps updates for a single architecture, and tests that seem 
> appropriate based on the mathematical properties of the function being 
> tested, and further additions there can reasonably be made in followup 
> patches.  (Note that even if we moved to a single libm-test-ulps shared 
> between architectures and listing expected ulps based on floating-point 
> format rather than type, it still wouldn't be possible to update all ulps 
> on a single system: no architecture supports both ldbl-96 and 
> ldbl-128ibm.)
> 
> -- 
> Joseph S. Myers
> josmyers@redhat.com
>