[Distutils] PEP 440 ad PEP 508 requirement formats

[Robert Collins]

Oh that reminds me, I noticed the other day that PEP 426 references
James' draft markers PEP that actually ended up in 508. Whats the
process for fixing that? Just do it in hg?

On 17 January 2016 at 04:46, Nick Coghlan <ncoghlan at gmail.com> wrote:
> On 16 January 2016 at 00:12, Paul Moore <p.f.moore at gmail.com> wrote:
>> Pip refers to PEP 440 when defining the format of a requirement (see
>> https://pip.pypa.io/en/stable/reference/pip_install/#requirement-specifiers).
>> But PEP 508 *also* defines a requirement - the title implies that it's
>> for dependency specification, but the content of the PEP says "the
>> language defined is a compact line based format which is already in
>> widespread use in pip requirements files".
>>
>> So what's the relationship between PEP 440 and PEP 508? Which one is
>> the definitive definition of what is a "requirement"? The "packaging"
>> library implements specifiers which conform (according to the docs) to
>> PEP 440.
>
> As Donald already noted, 508 is the higher level specification, that
> includes 440 by reference to cover the "version specifier" section of
> a full dependency specifier
>
> The key parts that 508 adds above and beyond 440 are:
>
> - package names (the current rules were previously only given in the 426 draft)
> - extras
> - environment markers
>
> The main benefit of keeping the two levels separate is that there's a
> *lot* of arcana in PEP 440 around version ordering and how that
> relates to version comparison operators that you don't really need to
> think about when working at the level of processing a list of
> dependency specifiers.
>
> Cheers,
> Nick.
>
> --
> Nick Coghlan   |   ncoghlan at gmail.com   |   Brisbane, Australia
> _______________________________________________
> Distutils-SIG maillist  -  Distutils-SIG at python.org
> https://mail.python.org/mailman/listinfo/distutils-sig



-- 
Robert Collins <rbtcollins at hpe.com>
Distinguished Technologist
HP Converged Cloud