[Catalog-sig] PEP 381 (Was: PyPI down?)

[Tarek Ziadé]

On Mon, Jan 18, 2010 at 10:29 PM, Jannis Leidel <jannis at leidel.info> wrote:
> Am 18.01.2010 um 21:47 schrieb Martin v. Löwis:
>
>>> FWIW, I don't see the PEP to be completed. The actual mirror
>>> protocol, how to handle multiple (unsynchronized) indexes and the API
>>> design are clearly undecided -- and not discussed.
>>
>> The actual mirror protocol *is* decided, even though it's not explicitly
>> spelled out in the PEP. It is based entirely on existing API; no new API
>> on the PyPI side is planned. Basically, you do a lot of HTTP GETs.
>
> If it's decided why isn't it public in the PEP? If you want developers to contribute you need stop deciding in private.

Come on, that's not what happened. IIRC Jim Fulton and Philip Eby
worked with Martin to make easy_install / zc.buildout calls on PyPI
efficient. Then I started the PEP later, but because I wanted to set
up an "official ring" of mirrors.

I am the one to blame because I didn't update that part of the PEP yet
consequently. But the PEP doesn't really address the protocol to be
used to browse PyPI. It's just informative. This is a de-facto
standard for years now (you use it everytime you install something
using pip or easy_install), and it was discussed in the Mailing Lists
back then when it was created. It didn't land in a PEP back then, like
other stuff don't.

So nothing was decided in private.

Now the push stuff could be great to have maybe, but I agree with
Martin that we should first setup the mirrors then learn from there.
(For the story, I've proposed a push stuff at first when we started to
discuss this PEP, but I eventually agreed that this was not the most
important thing to have our first version of a mirror ring).

Regards,
Tarek