Hello Niels ! On 08/04/11 23:04, Niels Thykier wrote: >> Cool, will have a look (hopefully Friday). From my initial review I >> might split and merge some patches. The merge of my branch will wait >> until Tuesday, so I will do the changes required for the merge. If you >> could provide some test cases for these tags before then, that would be >> great. > > > Alright, I had a look and have merged the patches into 4 with some minor > code-style changes. Before I commit the patches, I would like some test > cases to go with them.
Attached is a first draft that covers most of the checks (excepted
two, I'll handle that later on). I'm just sending you additional patches
with respect to what I sent already.
> I see you have tested some of your tags[1]; I am curious about your
> findings, you say that 30% of the javalibs you have tests triggers the
> tags. Have you any ideas if any of those are false-positives?
I'm afraid that in this specific case, chances of false positives are
unlikely. If a java package depends on a lib...-java, it means that some
of its code depend on one of the jars in this lib...-java. So it must
have an appropriate classpath entry. It is a warning as of now, but I'm
pretty sure they are all java policy violations, we may have to raise it
into an error. The only possibility I see for false positives is if a
package depends on the data contained in a lib...-java package, but not
on the code. I don't think it applies to a significant fraction of these
30%...
Cheers,
Vincent
--
Vincent Fourmond, Debian Developer
http://vince-debian.blogspot.com/
Au royaume des aveugles, il y a des borgnes à ne pas dépasser.
-- Soeur Marie-Thérèse des Batignoles (Maëster)
Vincent, listening to The Hardest Button To Button (The White Stripes)
java-lintian-checks-2.tar.gz
Description: application/gzip
