Opinions needed: reporting lintian overrides
After a recent problem with a package with a fairly egregious error that
was overridden, hurting the ability of the sponsor to notice it, I added a
line of output to the default lintian output saying if any package
overrides error or warning tags.
As I sort of suspected at the time, someone else has now objected to
having this output by default on the grounds that the point of the
override was to shut lintian up and it's easier to verify that lintian
produces no output.
I'd really like to not have to make this decision myself. I'd like to get
opinions and see if a consensus emerges. I personally always run lintian
with -iI --show-overrides, so I'm clearly not the target audience for this
feature one way or the other. Here are the options:
* Show the N: line with a count of overrides per package by default and
provide an option to suppress this output if someone wants.
* Don't show the N: line by default and provide an option to turn it on.
Which should we do?
Separately, I'm working on adding lintian overrides to the lintian.d.o
pages to make it easier to see how people are using overrides across the
whole archive. It looks like people sometimes just add an override when
lintian makes a mistake rather than filing a bug, so this gives us a
fighting chance of finding those bugs and fixing them. It also uncovers
some fascinating overrides currently in the archive.
Oh, and there are 1759 unused overrides in the archive in 369 packages.
lintian -i will tell you about unused overrides. We do fix false-positive
Russ Allbery (firstname.lastname@example.org) <http://www.eyrie.org/~eagle/>