Idea: issue an error if a '# type: ignore' comment is unused

[gvanrossum]

Sometimes we put in # type: ignore comments because of some mypy or typeshed bug (or a missing feature). It would be nice if mypy told us when we can take those out (e.g. because the bug is fixed).

Maybe this should be a command-line flag, I can imagine it might be annoying occasionally too.

[ddfisher]

This is a good idea! I think it should definitely be behind a flag, though -- it'd be annoying to cause errors in people's code whenever we fix a bug. Also, they may need the code to work across multiple mypy versions for some reason, which this would disallow. The fast parser transition is a concrete example: because it can change the line errors are assigned to, you'll need some extraneous type ignores during the transition period.

[gvanrossum]

It was @rwbarton's suggestion -- IIRC he said GHC has this.

--Guido (mobile)
On Apr 7, 2016 5:10 PM, "David Fisher" notifications@github.com wrote:

This is a good idea! I think it should definitely be behind a flag, though
-- it'd be annoying to cause errors in people's code whenever we fix a bug.
Also, they may need the code to work across multiple mypy versions for some
reason, which this would disallow. The fast parser transition is a concrete
example: because it can change the line errors are assigned to, you'll need
some extraneous type ignores during the transition period.

—
You are receiving this because you authored the thread.
Reply to this email directly or view it on GitHub
#1345 (comment)

[ddfisher]

I think it's become clear that this is a good idea to have available behind a flag. I think I'd discourage running with this on in CI (because sometimes you need unused # type: ignores for forward compatibility), but it would be great to have for periodic sweeps.

[gvanrossum]

Yes!

[rwbarton]

Even a simple version of this behind a flag will be useful, and I sort of promised this to the Zulip people, so I'll take a shot at this in the near future (traveling again this weekend but will still have time for mypy work).

[wittekm]

Potentially related request: could the same thing be accomplished for superfluous cast(some_obj, SomeType) when some_obj is already guaranteed to be SomeType?

(Sample use case: in our codebase, we were doing a lot of IntEnum casting due to a mypy bug that was just fixed).

[ddfisher]

Sounds like a good idea to me -- could even be the same flag, IMO.

On Fri, Jun 10, 2016 at 5:29 PM, Max Wittek notifications@github.com
wrote:

Potentially related request: could the same thing be accomplished for
superfluous cast(some_obj, SomeType) when some_obj is already guaranteed to
be SomeType?

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
#1345 (comment), or mute
the thread
https://github.com/notifications/unsubscribe/AAPAkUcfG-aeShbdOV8d9JgY6ODZ11pvks5qKgFqgaJpZM4IClTg
.

[rwbarton]

could the same thing be accomplished for superfluous cast(some_obj, SomeType) when some_obj is already guaranteed to be SomeType?

Yes! In fact, that is #958. Mypy itself has many redundant casts from before mypy understood isinstance checks; I removed about 60 of them a while ago but I'm sure there are a lot more.