Bug#833384: wordnet: multiple packages for the same database

[Andreas Tille]

Hi Ritesh,

On Wed, Aug 03, 2016 at 11:07:12PM +0530, Ritesh Raj Sarraf wrote:
> Source: wordnet
> Severity: normal
> 
> Is there a good reason for having 3 separate packages for the same
> wordnet lexical database?
> 
> Right now, we have:
> 
> wordnet
> goldendict-wordnet
> dict-wn

This is fully intended.  The packages are building from the same source
and are input for different backends requiring a different database
format.  So in this case the bug report is void.

> dico-module-wordnet

I have no idea about this but the maintainer is in CC.  May be the bug
report could be turned into:  Please create the dico wordnet database
from same source as wordnet?

> >From a dictionary server point of view, having just one pacakge would
> have been nicer. But I can see the argument that may come the other way
> around too.
> 
> Tools like GoldenDict are nicer that they are able to scan the installed
> dictionaries in dict format. This way, it helps the user who could use
> the client/server dict interface, as well as the very pretty and useful
> goldendict interface.
> 
> Is there some way to have it common ? I can see some information loss of
> the results from goldendict-wordnet vs dict-wn use case.

I admit I have no idea about the differences and how the binary
databases could be merged.  Feel free to provide a patch - I simply
added patches of other users before to *create* exactly these binary
representations of the wordnet database.
 
> The other concerning thing is the multiple copies with different
> versions/wordcounts.
> 
> dico-module-wordnet is at version 2.2-9. I'm not sure what the db
> version is at.

The other versions are at version 3.0 (as the package name should imply).
 
> On the rest of the wordnet packages in Debian, you are the maintainer
> for all of them. All the packages are at the same version, but their
> word counts are slightly different.
>
> For example, as per goldendict dictionaries tab:
> goldendict-wordnet      => Total Words: 148730
> dict-wn                 => Total Words: 147311

How did you git these numbers?  All I can say is that they are created
from the original database files using grind.  Please see the source
package if in doubt.
 
> Also, the results of dict-wn vs goldendict-wordnet are slightly
> different.

Could you please give an example for these differences?

Kind regards

     Andreas.

PS: I need to admit that my interest in the wordnet package is extremly
low.  It is team maintained in the Debian Science team and I'd like to
invite everybody who is interested to work on the package.  I'd happily
help in case of packaging questions but I feel not able to spent larger
amounts of time into this package.

-- 
http://fam-tille.de