[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#751378: ITP: openlibm -- standalone implementation of C mathematical functions



Le jeudi 12 juin 2014 à 11:41 +0000, Thorsten Glaser a écrit :
> Sébastien Villemot wrote:
> 
> >The OpenLibm code derives from the FreeBSD msun implementation, which in turn
> […]
> >OpenLibm builds on Linux, and with little effort, should build on FreeBSD as
> >well. It builds with both, GCC and clang. Although largely tested on x86, it
> >also includes experimental support for ARM.
> 
> Uhm. Use NetBSD’s libm instead, and you’ll have support for roughly
> all Debian architectures. And the FreeBSD changes can be piled on top
> if/when/where needed.

Thanks for the suggestion. I however doubt that julia upstream is going
to switch to NetBSD's libm since they have invested in their own libm.
Also, NetBSD's libm is not packaged in Debian AFAIK. So the most
practical alternative to openlibm at the Debian level is rather glibc's
libm.

> I do not like an addition of such an unportable package.

Note that julia is currently only available on amd64 and i386. Hopefully
it will soon work on armhf. And julia is currently the only rdep of
openlibm. So introducing openlibm as a separate package will imply no
regression in arch coverage of existing packages.

-- 
 .''`.    Sébastien Villemot
: :' :    Debian Developer
`. `'     http://www.dynare.org/sebastien
  `-      GPG Key: 4096R/381A7594

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: