Should licensecheck list skipped files ? (was: Re: Bug#806424: scan-copyrights: Failed to notice copyrights in lisp files)
Dominique Dumont
dod at debian.org
Mon Nov 30 19:50:39 UTC 2015
On Friday 27 November 2015 10:52:47 Wookey wrote:
> This correctly catalogued the copyrights in the python file, but none of the
> list files
Indeed. The .lisp extension is missing from the regexp used by licensecheck to
decide whether to scan a file or not. That's easy to fix.
> Now if I use licensecheck manually with a changed -c regex:
> licensecheck -r -c=* --copyright *
> It notices the copyrights in the list files:
> s-xml-rpc/test/test-base64.lisp: UNKNOWN
> [Copyright: 2002, 2004 Sven Van Caekenberghe, Beta Nine BVBA]
>
> But it still fails to grok the licence.
I can tweak licensecheck to scan the LLGPL.
> Can scan-copyrights call licencecheck in such a way that it looks look
> in more (all?) files by default?
Scanning all files by default is a can of worms: some files are binary (png,
jpg...) and will lead to a lot of garbage issued by licensecheck. This can't
be the default. And I'm reluctant to change scan-copyright to use -c '.*'
because of the extra processing required to weed garbage out.
> Or perhaps this bugreport should be
> directed to licensecheck to make the default more comprehensive.
Yes. Making licensecheck more comprehensive will benefit more tools (and
people). I'm thinking of license-reconcile and some other packages like
ghostscript that have a fairly advanced processing of the output of
licensecheck.
I will reassign this bug to devscripts
> Really
> I want a tool like this to look in everything it can (not just code
> and docs: graphics and test files too). Missing things entirely is
> much worse than false positives I can check and weed out.
> If it can't
> reliably find nearly all the licence and copyright notices in the tree
> then it doesn't really help much as I still have to look in every damn
> file myself, by hand.
Good point.
May be licensecheck should list files that are not scanned (instead of
returning garbage) ?
Thoughts ?
All the best
More information about the devscripts-devel
mailing list