[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Review of all the Debian Med debtags, and questions.

On Sun, Jan 11, 2009 at 02:48:40PM +0900, Charles Plessy wrote:

> Greetings, Debtags developpers, and happy new year !
> I would be really interested in reading your answer to the mail I posted last September:
> http://lists.alioth.debian.org/pipermail/debtags-devel/2008-September/001841.html
> Executive summary: please create works-with::sequence and delete
> field::biology:{bioinformatics|molecular|structural} and
> field::medecine:imaging.

You did well in pinging.

On Thu, Sep 18, 2008 at 12:04:00AM +0900, Charles Plessy wrote:

> I reviewed all the Debtags of Debian Med (except latest mgltools); I
> hope it gave me a better understanding of the system.
> http://debtags.alioth.debian.org/todo.html?maint=debian-med-packaging%40lists.alioth.debian.org
> First of all, I realised that Debtags is not a tree structure:
> Field::Biology and Field::Biology:Molecular are two different tags. Also
> I realised that subdivisions like this can be expressed by combinations
> of simpler tags. For this reason, I would like to ask the removal of:
> field::biology:bioinformatics
> field::biology:molecular
> field::biology:structural
> field::medicine:imaging
> Many of our programs work on biological sequences. I know we discussed
> this before, but I really think that works-with:sequence would make
> sense. In combination with field::biology, it would fully replace
> field::biology:bioinformatics. I volunteer to do the transition by hand
> if necessary. We currently have 84 field::biology:bioinformatics
> packages.

You're definitely going in the right direction.  The only thing I'm
worried about is that "works-with::sequence" seems to be an unclear name
if we take it outside of the biology context.  Would
"works-with::biological-sequence" be acceptable as well?

Alternatively, we can stick to "works-with::sequence" if we can think of
other sequences, non-biological but still somehow with similar behaviour
as far as software is concerned, that software may work with.  I can't
think of anything like that, so I'd go with "biological-sequence" at the
moment, and we're always in time to rename it later.

> works-with::trees
>   We would use this for our phylogeny packages, and many other tools not
>   related to biology could use it. works-with::graphs could be an
>   alternative, but potentially confusing.

works-with::graphs would probably make more sense considering we have
loads of graph theory and visualisation software around.  We may fix the
confusion by describing the tag as "Trees and Graphs".

> use::simulatinng
>   At least two of our packages would use it (adun.app, epigrass), and
>   others like flightgear could definitely use it too. I think it would
>   easily gain critical mass.

Totally.  I'd say this is uncontroversial, so I've added it.

> special::unmaintained
>   We unfortunately package some programs that are Upstream-dead, as many
>   other Debian packages are. Sadly, this tag could become very popular.

Yes.  I am planning a maint::* facet, in cooperation with Debian-QA, to
convey this sort of information (also, rc-buggy, fringe, obsolete-deps,

> works-with::temperature
>   We would have three candidate packages, but criticall mass would
>   probably attained with sensors and weather packages.

I see the need, but I'm looking for a better name: sensors and weather
packages wouldn't just measure temperature, but also pressure, light,
wind and whatnot.  So maybe something along the line of
"works-with::physical-measurements".  "works-with::measurements", even?
That may include benchmark tools as well, and instinctively I'd say "why

> The following ones are just ideas not really reflecting our needs:

I'll give quick comments, so that going through the list won't delay the
answer to the more important questions above.

> made-of::data:examples, or role::example

Is it worth to separate examples from documentation?

> role::translation

I'm undecided whether this should be considered a special case of
role::app-data, and then be more clearly defined elsewhere (like,
culture::*) or if it's worth putting it as a first-class role.

The main goal for role::* is to have the smallest possible number of
tags that can apply to the whole archive: every package should be able
to get at least one role::* tag.

> role::library (for languages like Perl, the current dichotomy is not relevant)

Indeed, that's a can of worms.  It's an essential distinction for C, C++
and so on, but it really doesn't make sense for perl, python and ruby.

role::library is tempting, but I still need the distinction shared/devel
because, for example, you often want to hide shared libraries from a
package manager interface, but you often don't want to hide development

> use::calculating

Definitely needed.  "calculating" or "computing" ?

> role::policy
>   We actually have one such package, mipe (Minimal Information for PCR
>   experiments), and soon our Group policy will be
>   part of a package as well.

This should probably be role::documentation, plus something else.

> uitoolkit::xulrunner (in our case: biofox)

This makes a lot of sense: how many packages do we have?  iceweasel,
biofox, then?

> Lastly, I have the impression that there are some strong redundancies:
> role::devel-lib and devel::library (actually, maybe the whole devel::
> facet could be formulated with appropriate combination of other
> Debtags). I have systematically used the first and ignored or even
> deleted the second. (I can repair this if you disagree).

That is redundant indeed, but still both tags make sense in their
facets.  My gut feeling is to just live with it, and maybe even write
some code to add one of the tags if the other one is present.

> role::program always have a scope. Definition of scopes are such as they
> can only apply to programs. If this facet is not broadened, maybe you
> can consider to use role::{application|utility|suite} instead of the
> scope facet.

You might have scope applied to role::metapackage, though.  But more
importantly, scope::* is such a can of worms that I'd like to keep it
separated from role::*, which is quite straigtforward.

> Executive summary: please create works-with::sequence and delete
> field::biology:{bioinformatics|molecular|structural} and
> field::medecine:imaging.

Will do as soon as we have an agreement about the
"works-with::(biological-)sequence" name of the tag.



GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini <enrico@debian.org>

Attachment: signature.asc
Description: Digital signature

Reply to: