Le Mon, Jan 18, 2010 at 12:05:17PM +0100, Andreas Tille a écrit : > > > If it fits wells the blends script that creates the web sentinels, > > perhaps the ???long??? format (package name / keyword / value) will help us > > to keep > > the system most simple. Also, that is the closest to an RDF tuple??? > > I don't think that it is a good idea to aggregate all the bibliographic > information into a text field. This makes no sense if you want to > attract a more general usage. IMHO we should go with > > package text, > title text, > authors text > published-in text, > year int, > url text, > doi text
Hi Andreas, This is more or less what I meant by “long” format (in the sense of R’s reshape() function). In order to fit this information in a SQL table, I was more thinking about: package title text package authors text package published-in text package year text etc. > We should decide whether to allow more than one bibliographic > information dataset per package (and how to distinguish these). In a previous discussion (which I can not find for the moment), there was no objection to do the contrary: support only one reference. The web sentinel has currently the same limitation anyway. In the case a package contains multiple programs to be cited with multiple articles, I was considering using a field with a name like ‘Reference-List-URL’, that contains a link to an upstream web page that contains all the instructions and references. Such pages are usually available in these cases. Have a nice day, -- Charles Plessy Debian Med packaging team, http://www.debian.org/devel/debian-med Tsurumi, Kanagawa, Japan -- To UNSUBSCRIBE, email to debian-qa-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
