ast.parse with type_comments=True should allow extra text after # type: ignore

[msullivan]

BPO	36878
Nosy	@gvanrossum, @ilevkivskyi, @miss-islington, @msullivan
PRs	bpo-36878: Allow extra text after # type: ignore comments #13238 bpo-36878: Track extra text added to 'type: ignore' in the AST #13479 bpo-36878: Only allow text after # type: ignore if first character ASCII #13504

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2019-05-22.21:08:33.760>
created_at = <Date 2019-05-10.20:40:16.492>
labels = ['interpreter-core', 'type-feature', '3.8']
title = 'ast.parse with type_comments=True should allow extra text after # type: ignore'
updated_at = <Date 2019-05-22.21:28:13.436>
user = 'https://github.com/msullivan'

bugs.python.org fields:

activity = <Date 2019-05-22.21:28:13.436>
actor = 'msullivan'
assignee = 'none'
closed = True
closed_date = <Date 2019-05-22.21:08:33.760>
closer = 'gvanrossum'
components = ['Interpreter Core']
creation = <Date 2019-05-10.20:40:16.492>
creator = 'msullivan'
dependencies = []
files = []
hgrepos = []
issue_num = 36878
keywords = ['patch']
message_count = 9.0
messages = ['342130', '342144', '342213', '343130', '343198', '343205', '343236', '343237', '343240']
nosy_count = 4.0
nosy_names = ['gvanrossum', 'levkivskyi', 'miss-islington', 'msullivan']
pr_nums = ['13238', '13479', '13504']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue36878'
versions = ['Python 3.8']

[msullivan]

Per discussion during the typing summit at PyCon, it would be a good idea to allow extra information to be included in # type: ignore comments, in order to allow behavior such as suppressing individual errors (for example, with syntax like # type: ignore[E1000], to suppress error 1000).

My proposal, then, is to generalize the definition of type: ignore comments to be # type: ignore followed by a non-alphanumeric character. Then # type: ignore[E1000]and# type: ignore E1000would be valid type ignore comments while# type: ignoreE1000` would not be.

Now that ast.parse can parse type_comments, this needs to make it into 3.8, basically, if we want to do this and be able to use the ast type_comment feature.

Ideally, the text of the type ignore would be actually included in the produced AST. As a bare minimum first step, we need to recognize type ignores with extra information and report them (and, critically, not detect them as regular type comments and produce errors when they appear in unexpected places).

I'll put up a PR to do the second part shortly.

[ilevkivskyi]

I like the idea of separate smaller PRs, the first one looks good.

[ilevkivskyi]

New changeset d8320ec by Ivan Levkivskyi (Michael J. Sullivan) in branch 'master':
bpo-36878: Allow extra text after # type: ignore comments (GH-13238)
d8320ec

[gvanrossum]

How many more PRs do you plan? Or is this it?

[ilevkivskyi]

New changeset 933e150 by Ivan Levkivskyi (Michael J. Sullivan) in branch 'master':
bpo-36878: Track extra text added to 'type: ignore' in the AST (GH-13479)
933e150

[msullivan]

I think there will be one more PR to disallow non-ASCII characters immediately after a # type: ignore, but otherwise I think this is done

[miss-islington]

New changeset d8a82e2 by Miss Islington (bot) (Michael J. Sullivan) in branch 'master':
bpo-36878: Only allow text after # type: ignore if first character ASCII (GH-13504)
d8a82e2

[gvanrossum]

Thanks!

[msullivan]

I think this is done!