public inbox for libc-alpha@sourceware.org
 help / color / mirror / Atom feed
From: Joseph Myers <josmyers@redhat.com>
To: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Cc: Carlos O'Donell <carlos@redhat.com>,
	libc-alpha@sourceware.org,  adhemerval.zanella@linaro.org,
	schwab@suse.de, dilfridge@gentoo.org,
	 michael.hudson@canonical.com, simon.chopin@canonical.com,
	 aurelien@aurel32.net, fweimer@redhat.com
Subject: Re: Planning for glibc releases --- What should go into glibc 2.40?
Date: Wed, 14 Feb 2024 17:26:41 +0000 (UTC)	[thread overview]
Message-ID: <13512626-576-2cf6-41ca-68d353cb72fa@redhat.com> (raw)
In-Reply-To: <p9u08r3n12oq.fsf@coriandre.loria.fr>

On Wed, 14 Feb 2024, Paul Zimmermann wrote:

>        Hi Carlos,
> 
> > https://sourceware.org/glibc/wiki/Release/2.40#Planning
> >
> > Please add items to the plan :-)
> 
> I don't think I have write access to the wiki, but I'd like to see the 
> new C23 math functions (acospi, asinpi, atanpi, atan2pi, cospi, sinpi, 
> tanpi, exp10m1, exp2m1, log10p1, log2p1, rsqrt) available in GNU libc.

I think this wiki list is for things that already have patches - not for 
aspirations that some feature would be nice to have.

I hope to work on these functions at some point, but it depends on time 
available between other projects (and my time available will be very 
limited from April through July in general).  As previously discussed, the 
functions would start with type-generic templates where possible (thereby 
making the functionality available for all formats, including legacy ones 
such as IBM long double, without needing individual implementations for 
each format - we don't want functions for only a subset of formats - and 
dealing with all the documentation, test inputs, ABI baseline etc. updates 
at that point), and then faster and more accurate implementations can be 
incrementally added for particular (function, format) pairs later (with 
benchmarks added for any function being replaced to show that the new 
implementation is indeed faster).

Note that your list omits the logp1 alias, and the compoundn rootn pown 
powr functions.  Also, there are functions we have from TS 18661-1 whose 
prototypes changed in C23: we need to change the prototypes of fromfp 
functions (see bug 28327), with backwards ABI compatibility similar to 
that for totalorder.

On the C23 front, there is at least one feature where a patch has already 
been posted but needs review and updating (to put things in the latest 
NEWS section, at least): the new optional time bases (posted by Yonggang 
Luo last June).  There are also several C23 features where we don't yet 
have patches, but that would be simpler to add than the new libm 
functions.

-- 
Joseph S. Myers
josmyers@redhat.com


  reply	other threads:[~2024-02-14 17:27 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-14 13:29 Carlos O'Donell
2024-02-14 15:32 ` Paul Zimmermann
2024-02-14 17:26   ` Joseph Myers [this message]
2024-02-14 16:22 ` Andreas K. Huettel

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=13512626-576-2cf6-41ca-68d353cb72fa@redhat.com \
    --to=josmyers@redhat.com \
    --cc=Paul.Zimmermann@inria.fr \
    --cc=adhemerval.zanella@linaro.org \
    --cc=aurelien@aurel32.net \
    --cc=carlos@redhat.com \
    --cc=dilfridge@gentoo.org \
    --cc=fweimer@redhat.com \
    --cc=libc-alpha@sourceware.org \
    --cc=michael.hudson@canonical.com \
    --cc=schwab@suse.de \
    --cc=simon.chopin@canonical.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).