All CLA requirements met.

Isn't it enough to just have ' and " as an auto pair character? The only difference I'm aware of is that undo will only remove the quote, instead of the letter before, but IMO that's better anyway.

The original auto-close setting did not auto-close r', etc. Presumably, it's coded to only trigger after a whitespace (or non-word boundary), but I didn't look at the code.

Oddly, this doesn't explain why the above configuration auto-closes fr', etc. If the open pattern must be preceded by whitespace, then that shouldn't work. Perhaps it's coded differently if the pattern starts with an alphanumeric character?

Yes, there's a condition to not autoclose ' after a word.
@alexandrudima Do you want to reconsider that, or is the PR ok?

Sorry, this slipped. @alexandrudima Waiting for your opinion.

The following added rules do nothing:

		{ "open": "r\"", "close": "\"", "notIn": ["string", "comment"] },
		{ "open": "R\"", "close": "\"", "notIn": ["string", "comment"] },
		{ "open": "u\"", "close": "\"", "notIn": ["string", "comment"] },
		{ "open": "U\"", "close": "\"", "notIn": ["string", "comment"] },
		{ "open": "f\"", "close": "\"", "notIn": ["string", "comment"] },
		{ "open": "F\"", "close": "\"", "notIn": ["string", "comment"] },
		{ "open": "b\"", "close": "\"", "notIn": ["string", "comment"] },
		{ "open": "B\"", "close": "\"", "notIn": ["string", "comment"] },
		{ "open": "r'", "close": "'", "notIn": ["string", "comment"] },
		{ "open": "R'", "close": "'", "notIn": ["string", "comment"] },
		{ "open": "u'", "close": "'", "notIn": ["string", "comment"] },
		{ "open": "U'", "close": "'", "notIn": ["string", "comment"] },
		{ "open": "f'", "close": "'", "notIn": ["string", "comment"] },
		{ "open": "F'", "close": "'", "notIn": ["string", "comment"] },
		{ "open": "b'", "close": "'", "notIn": ["string", "comment"] },
		{ "open": "B'", "close": "'", "notIn": ["string", "comment"] }

Out of the added rules, this is the only one that might do something:

		{ "open": "'", "close": "'", "notIn": ["string", "comment"] },

Closing Pairs are implemented to be working on single characters. Basically, the code evaluates if someone typed the character in the open part of a closing pair and does not consider any characters before the auto closing open character.

@alexandrudima I tested this configuration myself when I submitted this PR. Without my changes, when you type r', it does not auto-close to r''. With my changes, it properly auto-closes raw, unicode, f-strings, and bytes.

Yes, that's what I'm saying too. The only line that makes this work from this PR is:

{ "open": "'", "close": "'", "notIn": ["string", "comment"] },

This defines that ' - ' forms an auto-closing pair.

The other 16 lines serve no purpose.

@alexandrudima
{ "open": "'", "close": "'", "notIn": ["string", "comment"] }, is already part of the autoClosingPair.
It doesn't work after r' due to this.
Adding { "open": "r'", "close": "'", "notIn": ["string", "comment"] }, fixes the problem

@Syeberman @aeschli I agree the change makes this feature work. I was quite stumped at first, since I am pretty sure how auto-closing characters should work and how they're implemented.

As I thought, the execution does not enter the auto closing open character code (since that is implemented to only work on single characters), but it enters BracketElectricCharacterSupport._onElectricAutoClose and these newly added auto-closing pairs somehow make their way to _complexAutoClosePairs.

@aeschli Since this is something you've authored in ebc3074 , I cannot be of further assistance.

Perhaps now would be a good time to document somewhere this behaviour!

Talked to @alexandrudima again and we are fine with the PR. Not autoclosing ' and " after a letter is the expected behaviour most languages, so the PR makes sense to add additional rules fo the cases where this makes sense.

Thanks @Syeberman!

I still have the issue on Version 1.21.1 (1.21.1) and running OSX Sierra.

@diazgilberto Please file a new issue with a code sample that shows the problem. Also try to run without extensions to see whether an external extension causes this.