[Debian-science-sagemath] pari-sage or pari ?

[Ximin Luo]

Jeroen Demeyer:
>> There is a big issue that the *version* 2.8-2771-gb70b447-1 is not a release version of pari so Bill is probably not going to want to package that. But as far as I can tell, it is the actual upstream "2.8-2771-gb70b447-1" version i.e. 2.8 with 2771 extra git commits (if I am reading that correctly)
> 
> No, that version is actually a development version *before* 2.8. But recently, PARI-2.8.0 was released and Sage will use that version. See also https://trac.sagemath.org/ticket/21005
> 

OK, thanks for clarifying.

> But as usual, sooner or later, Sage will probably use a development version of PARI again. So you shouldn't package under the assumption that Sage will always use a released version of PARI.
> 

For the purposes of maintaining an OS distribution that satisfies everyone's needs, and not just that of Sage, in the vast majority of cases we do need to stick to release versions. We make exceptions, but these exceptions carry costs and we typically really don't like doing so.

The broader FOSS community understand this principle, and also typically use release-versions of their dependencies. So everyone will depend on version 1.4 of libfoo, instead of having project A depend on libfoo=1.4+425 and project B depend on libfoo=1.4+667 and then this is extra burden on volunteer's time to try to add both to Debian.

This is why it has taken so long to get Sage into Debian. We are constrained by everything else.

(BTW, could you fix your mail client to send the proper References: / In-Reply-To: headers, so threading works correctly in standard mail clients?)

X

-- 
GPG: ed25519/56034877E1F87C35
GPG: rsa4096/1318EFAC5FBBDBCE
https://github.com/infinity0/pubkeys.git