[Product-Developers] Releasing plugins to pypi AND plone.org

[Luca Fabbri]

On Mon, Apr 4, 2011 at 11:29 AM, Dylan Jay <djay at pretaweb.com> wrote:
> On 04/04/2011, at 6:42 PM, Andreas Jung wrote:
>
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Luca Fabbri wrote:
>>
>>>> Maybe plone.org should use the pypi API to scrape packages with a Plone
>>>> classifier and instead of requiring double uploads.
>>>>
>>>
>>> This can be good, but for me something must be done. Maybe also remove
>>> completely the Products section on plone.org can be good, and make
>>> specific links for search on pypi for new Plone products.
>>
>> My 2 cents:
>>
>> in general the products section is a good thing and should stay but
>
> +1
>
>> it should reuse as much as possible from PyPI. Blueprint:
>> a package of registration on plone.org/products hold only
>> basic metadata like the PyPI URL, url to bugtracker etc.
>> All other information need to generate listings on plone.org
>> are afaik available through the PyPI XML-RPC or from the simple
>> version of the PyPI index (/simple).
>
> so the workflow would be like below?
>
> 1. User registers and uploads their package to pypi
> 2. Login in to plone.org
> 3. Add "mypackage" to PSS
> 4. Give the pypi url
> 5. PSS parses pypi and pulls in all releases and metadata
> 6. User submits for publication
> 7. after time user makes new release to pypi
> 8. nightly batch job pulls in the new metadata? (or is PSS hooked up to feed
> from pypi somehow?)
>
> There's a bit of coding in there + there would perhaps need some work to
> transition the existing projects.
>
> Would this also mean you can no longer use plone.org as a source for package
> downloads?


I think that plone.org is also used as backup, in the case (not so
uncommon in the past) that pypi is not responding.
Didn't it?

>
>>
>> Releasing in two different repositories is error-prone and not necessary
>> and has more cons than pros.
>
> I agtee. Still it's a big change.
>




-- 
-- luca