metacpan.org is a search engine, not canonical source

[Axel Beckert]

Hi,

[Jonas and me had a longer discussion #debian-perl to check if we talk
about the same thing. Basically we disagree on the preference between
indirection vs per-author URLs. Jonas: Feel free to post your view of
that discussion, too. :-]

Jonas Smedegaard wrote:
> What you write matches the advice that I wrote about in first sentence 
> of my initial post:
> 
> > We advice at <https://pkg-perl.alioth.debian.org/policy.html> to use
> > metacpan.org URL in watch file and - if not provided - as homepage.
> 
> That is not the issue I raise here.

Good to hear. It was initially not clear to me, if you want to change
them away from metacpan.org, too. (I consider metacpan.org as current
default for debian/copyright, too, despite it's not written down
anywhere yet.)

I'm btw. fine with adding a recommendation for what should go into
debian/copyright.

> You do not comment on second sentence of my initial post:
> 
> > We do not, however, advice about URL to provide as source in copyright
> > file.

I think I did. Twice:

> > I have no strong opinion about which URL should go into the 
> > debian/copyright file, but I prefer URLs which don't change too often 
> > to lower the need for packaging changes, too.
> > 
> > So I prefer to stick with the 
> > https://metacpan.org/release/$DISTRIBUTION scheme, at least for 
> > Homepage header and watch files, but would also recommend it for 
> > debian/copyright.

> To summarize: I believe metacpan should not be used as canonical source, 
> because it is not source, only a (quite nice) search interface for 
> sources at www.cpan.org.

And I disagree for the reasons quoted above, but wouldn't try to veto
as my opinion for this is not as strong as with watch file and
Homepage header.

I only think that the package maintenance effort that would be needed
for distributions with multiple authors/uploaders and hence regularily
changing upstream source URL is not worth getting rid of the one
indirection (if we consider any CPAN mirror including metacpan.org's
as canonical).

Citing mst for an example of such a distribution:

22:15:33 <mst:#debian-perl> XTaran: if a different person releases the
next version the module will move author directories
22:15:43 <mst:#debian-perl> people assuming that wouldn't be the case
were the bane of my life for years
22:16:19 <mst:#debian-perl> since DBIx::Class was the first major cpan
project to have a pool of half a dozen or so people who shipped dists

(Another result of that discussion is btw. that I'll make a pull
request to patch MetaCPAN's about page. :-)

		Regards, Axel
-- 
 ,''`.  |  Axel Beckert <abe at debian.org>, http://people.debian.org/~abe/
: :' :  |  Debian Developer, ftp.ch.debian.org Admin
`. `'   |  1024D: F067 EA27 26B9 C3FC 1486  202E C09E 1D89 9593 0EDE
  `-    |  4096R: 2517 B724 C5F6 CA99 5329  6E61 2FF9 CD59 6126 16B5