On 06/11/12 14:05, Edward Pilatowicz wrote:
On Mon, Jun 11, 2012 at 09:13:20AM -0700, Shawn Walker wrote:
On 06/10/12 22:59, Edward Pilatowicz wrote:
...
i don't think that's realistic. there are multiple situations that i
can think of (wildcard imports, relative imports, no exception specified
when catching exceptions) that we generally want to avoid, but there
will always be special case there will be exceptions. but having these
errors enabled in pylint will force us to consider which case really
should be special, and otherwise fix the issue. (banning the temporary
disabling of pylint messages would be like saying you can't use LINT
directives in C code. not very realistic.)
Be that as it may; I despise littering our code with pylint
directives :-( I hope we can find a better alternative.
i agree the code should not be littered. but define littering. imo, if
you've got a 1000+ line file and 2 or 3 pylint directives for a given
error, then i wouldn't call that littering, i'd call that having a rule
with occasional exceptions. on the other had, if the same file has 10
or 20 pylint directives for a single error, then i'd say we should
re-evaluate the benifit of catching that error (either in the gate, in
that file, in that object, etc).
It's the 2 or 3 in every file that I would find distasteful. In short,
if we end up having to disable the same things in lots of places, I'd
rather just disable it everywhere.
-Shawn
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
