Re: Status bibref gatherer (Was: Tasks pages (close to) fixed; Bibref does not seem to be updated automatically)

[Charles Plessy <plessy@debian.org>]

Le Fri, Mar 09, 2012 at 08:51:45AM +0100, Andreas Tille a écrit :
> 
> Ups, seems there is some misscommunication?  I simply assumed it would
> have been the agreement that you would provide the upstream files that
> way and I volunteered to parse them for UDD.

Hi Andreas,

I agreed to provide flat files in a way that they can be parsed by anybody,
because I agree that such a repository has some value, especially since it
includes the copyright files as well.

But for feeding the UDD with the bibliographic information, I feel that your
effort is redundant with my work.  Most UDD tables are fed by downloading the
data from other sites, and while your approach is more direct in apparence, I
do not think that it is necessary.  Moreover, it uses the files collected by
umegaya, and therefore produces the same output, with the same limitations.

Also, your importer is inducing changes in the syntax of the files, that I
would prefer to be discussed first.  The debian/upstream files are a good
medium to feed the UDD, but they can have other uses.  The more we take ad-hoc
decisions with only the UDD in mind, the more we close doors for other uses.

For the references in particular, I think that we need to think twice before
introducing arrays.  As I explained earlier, I would prefer that the fields
remain flat.  With the array syntax, packages for which there is only one
reference, and they are the majority, would need to use an array as well:

  Reference:
   - Title: foo
     Name: bar

instead of

  Reference:
    Title: foo
    Name: bar

or

  Reference-Title: foo
  Reference-Name: bar

With the array syntax, I expect a lot of confusion and errors.  This could be
solved as simply as keeping the original syntax for the main reference, and
using an array for all the other ones.  I think that this example underlines
that we need to discuss first before making choices.  Otherwise, it becomes
your personal project.

Regarding the use of hashes to represent complex field names, this is not a
feature of YAML.  If this poses problem, then I would rather stop using hashes
altogether.  Or at least we should discuss it.  Otherwise, I will gladly accept
your changes for bootstaping the system, and spend more time on other
developments (like using Debian Installer to prepare amazon clould images).  If
you want that way, do not hesitate to tell it; I will not be upstet.

Have a nice week-end,

-- 
Charles Plessy
Debian Med packaging team,
http://www.debian.org/devel/debian-med
Tsurumi, Kanagawa, Japan