Re: BibTeX file from debian/upstream data (Was: New Debian Science metapackages)
On Fri, May 04, 2012 at 07:24:13PM +0900, Charles Plessy wrote:
> I think that the key information that we need to give to our users is how to
> cite a work that we packaged, rather than where to read about it.
I'm not sure whether we could mark some information as "key" information
or not. From my perspective its a valuable resource to read a
publication about a program before I install / use it - so the
publication links are from my perspective pretty relevant. Finally we
started with the structured collection of publication data because
maintainers tended to put it into the long description and thus
considered this as something important.
> In most of
> the case it will be the same, but for some programs it will differ.
>From my perspective the way you are *refering* to a programm in a
publication is perfectly different than citing a publication. Please
have a look at the examples Michael (mpqc, cp2k) or yesterday I checked
for publications about Jmol (but failed) and detected a clear statement
of the authors at their homepage how they want people who use Jmol
how they should cite it. After having read this I think I understood
what Michael was approaching and I consider this type of information as
orthogonal to any publication data because it rather is a definition
of a "string" you should use while publication is a link to some further
> What is
> important is that we follow the upstream developers instructions, because the
> they know which of the citations count or the URL hit count will affect their
Helping the authors is the other reason (in addition to informing the
user) why I'm engaged in this effort. I think we are doing pretty well
here compared to the status 6 monthes ago. But I see no reason to mix
up two different things and even more I can not see in how far it would
decrease their hit count if I follow the approach I mentioned in my
mail. Both will be displayed - I just will need to work down the work
items in a sequence.
> Typical journals limit the number of references per article, and the first to
> be deleted are the ones for the programs that are easy to find with Google. If
> the authors have established their work as a URL rather than as a peer-reviewed
> manuscript, it also can be fine with some other publishers, and I think that
> BibTeX files and citation managers are able to handle this situation through
> fields such as "URL".
The "Cite-As" links in Michaels examples are no BibTeX relevant data.