REVU: Automated Package Checks

[Emmet Hikory]

James Westby wrote:
> On Thu, 2009-01-22 at 18:44 -0600, Nathan Handler wrote:
>> On Thu, Jan 22, 2009 at 6:31 PM, Loïc Martin  wrote:
>>> What happens when lintian (or another automated check) throws an error,
>>> but that error is not justified? I've seen the case for all cdemu
>>> related packages (for example
>> I actually had thought about this issue a bit. This lintian error that
>> you mention is not the only instance of a package that is not lintian
>> clean that is technically correct. The issue is, I personally can't
>> think of any reliable way to check for these cases. Having a MOTU
>> check the error is certainly possible, but I would prefer to keep this
>> as automated as possible. This is one reason that I sent an email to
>> the mailing list; I am hoping that someone can come up with an
>> efficient way to handle these exceptions.
> 
> For lintian there are lintian overrides.

    While adding more automated checks to REVU would be nice, the
existence of things such as lintian overrides, means that some of the
more substantive tests can be masked.  While it's nice when a given
package has all the trivial bits right, these aren't typically as
important as having the packaging right.  Aside from false positives,
many lintian warnings or errors may not make sense initially, and I'd
hate to find packages that passed the robot tests from excessive
overrides that needed to be detangled to be reviewed.


-- 
Emmet HIKORY