syntax: leading and trailing periods / dots not tokenized as part of identifier

[kevinushey]

Given an R document with the code:

.hello.world.

The leading + trailing .s are tokenized as separate entities.

I believe the use of \b here forces VSCode to consider . as a word boundary, and so it fails to tokenize the whole identifier as a single token:

https://github.com/REditorSupport/vscode-R/blob/da579cc17a9485f7a11bf5e75bf6b9c347eca116/syntax/r.json#L165-L168

If so, I think those \bs can be safely removed?

[renkun-ken]

Yes, I think you are right. We should remove those \b. Do you find unnecessary \b in other matches?

[kevinushey]

I guess another example 42., which R parses as a single number but appears to be tokenized as 42 and .:

https://github.com/REditorSupport/vscode-R/blob/da579cc17a9485f7a11bf5e75bf6b9c347eca116/syntax/r.json#L123-L126

And also for references to packages containing a . in their name:

I'll poke around a bit more for other edge cases...

[gowerc]

Could one potential edge case be on S3 methods ? i.e.

as.data.frame.list()

though tbh I'm not exactly sure what you would want to tokenise here :s

[github-actions]

This issue is stale because it has been open for 365 days with no activity.

[eitsupi]

unstale

[ThomasSoeiro]

@eitsupi @Fred-Wu @grantmcdermott Can you please transfer this issue to https://github.com/REditorSupport/vscode-R-syntax/?

(Either I do not have the permission to do so or the documented process has changed)