[pkg-bioc] Conversion of BioC Views to Debtags

[David Vernazobres]

Sorry for the late answers, [but I got a corrupted /usr on my laptop,
the first day of my holiday... and yet I am back.]

Yep, I start looking at it (some time ago) but did not come with any
strategy.

On Wed, May 30, 2007 at 04:27:26PM +0200, Steffen Moeller wrote :
> Hi all,
> 
> we have not yet addressed the tagging of our packages, something imminently 
> important as the number of packages has grown so high.
> 
> I have just packaged the biocViews
> http://bioconductor.org/packages/2.0/bioc/html/biocViews.html
> package and am daydreaming about how to merge that with the Debtags out there. 
> There are two hurdles:
> 
> * preparation of new Debtags
> 
>   We should come up with our own sensible tagging and send kind emails. Daniel 
> might meet some Debtaggers at the DebConf in Edinburgh and discuss things 
> further.
We may get some inspiration from the debian-med tagging ;)
http://svn.debian.org/wsvn/debian-med/trunk/community/debtags/tags?op=file&rev=0&sc=0
for microarray something like the following sound good :  
* field::biology:genomics
* use::experiment:molecular-biology
* works-with-format::biology:affy


> * semantics 
> 
>   There are some default tags since our packages are all R packages - easy ;o) 
> For the others I mostly  see them under field::biology with the hierarch 
> kept. But what should we do with a description of a microarray chip? I do not 
> real idea yet. use::converting maybe. And made-of::data:R.

Well field::biology for bioc. For Cran & Omegahat, it's a mix between
math, physics, chemistry, biology....
But the global tag made-of::data:R sound good for me.

 
> * Assignments to packages
> 
> The file format is fairly straight forward. Tags are identified by their full 
> path, consequently there is no term with multiple parents. A single line per 
> package:
> arb-doc: field::biology, made-of::data:html, role::documentation
> I think we can find ways to generate such entries dynamically and I'll come 
> with something. An interference with the build process should not be 
> required, just optional, so I would say.
> 

thanks for your nice work,
I am still checking the global CVS commit email.

david