Re: .changes files as first-class objects
On Sun, 2010-01-17 at 14:48 -0800, Russ Allbery wrote:
> I unfortunately haven't had a chance to do a full review of this patch
> (trying to get to that soon), but the above patch indicates that we're
> adding laboratory data for changes files? I'm not sure that's a good
> idea.
>
> The main point of writing data to disk, now that we have Lintian::Collect
> that could hold it, is because it's very useful to use the lab to look at
> various archive-wide things. This doesn't apply to changes files since
> one is rarely going to run Lintian on a *.changes file with a persistant
> lab. They're not part of the archive under normal circumstances. There
> also isn't much interesting data there; anyone who does have a large
> collection of *.changes files that they want to review will just grep
> through them all.
Yes, currently changes data is stored in the laboratory.
Partly this is due to the fact that both Lintian::Collect::Changes and
the new changes-file check script make extensive use of the field()
method, which is inherited from the base Lintian::Collect class, where
it's implemented as reading fields/.
The frontend itself also assumes that collection and check scripts can
be executed from within $LINTIAN_LAB/$long_type/$pkg.
Adam
Reply to: