[Catalog-sig] PEP 345 Update

[Tarek Ziadé]

On Mon, Aug 23, 2010 at 5:28 AM, P.J. Eby <pje at telecommunity.com> wrote:
> At 04:09 AM 8/23/2010 +0200, Tarek Ziadé wrote:
>>
>> The fields descriptions are quite clear, Obsoletes is useful for
>> reorganizing
>> softwares into different releases names, whereas Conflicts marks a release
>> to be incompatible with another one,
>
> If that's the case, then it should suffice to explain in the PEP that the
> intent of this field is for an author/owner to describe reorganization of
> their own software, rather than for one package to claim that it's a
> replacement for another.

We can improve the Obsoletes-Dist description, sure. Notice that
it will be misused if we don't add Conflict-Dist. That's basically
why I wanted to add this field, as suggested by someone on IRC (sorry
I forgot who)

>> the PEP is about the metadata, not the softwares that will implement it.
>
> Which is why I've found the previous package metadata PEPs to be pretty
> useless: they described fields in the abstract without much concrete
> semantics.  And thus, they were not worth writing software to parse, most of
> the time.
>
> To put it another way, without suggested semantics, people will put whatever
> they feel like in the fields, because they likewise have no idea of how the
> information will be used, or what the consequences of entering that
> information will be.

The biggest problem imho, was that the dependencies where at the module
level like Perl, and the semantics were completely artificial in Python.

> In short: if it's not going to be used, why have it?  And if it *is* going
> to be used, why leave the semantics undefined?

Sure, more explanation is always better

Tarek

-- 
Tarek Ziadé | http://ziade.org