Map violations to error codes (similar to pep8)

[campbellr]

Map violations to error codes (similar to pep8)

[keleshev]

Let me be devil's advocate for a second. Why would you ignore a rule?

[campbellr]

You might want to ignore certain rules for various reasons, either for purely subjective reasons (you just don't agree with it) or legacy projects where you just haven't convinced everyone to get entirely on the pep257 bandwagon.

Aside from the 'ignoring' use-case, the error codes just make the the project more friendly for external tools, so they don't have to depend on matching the human-readable version of the error, which may change.

[sigmavirus24]

I think this is also more desirable in the context of Flake8 and the extension written for it using pep257. In general, having error codes with shorter descriptions are more in line with the other tools we already support and use. I didn't mention it because it doesn't particularly bother me since we're using pep257 as an extension. Were we introducing it as a core feature, I may have been a bit more concerned. That said, I agree that ignoring a rule is generally a bad idea, but rules are meant to be broken. :-P

[keleshev]

@sigmavirus24 do you have any suggestions for error code naming?

[sigmavirus24]

Well easiest for Flake8 is to use a D (for doc-string)[1] followed by 3 or more numbers as you see fit (and to store them somewhere on the error object for us to use for ignoring). Generally similar errors are grouped in sort of classes like pep8 so you have 99 errors to report for each like idea, e.g., D1xx are all similar but different errors relating to essentially the same idea. D2xx works the same.

I don't recall the exact error strings as they are in pep257 now, but one of this week's nights I might be able to provide a suggestion for grouping the errors together. I have no good suggestions for shortening the messages though because I find them to be as terse as I can imagine them (while working on flake8-docstrings I thought about re-mapping the errors but couldn't think of better messages frankly).

[1]: Mainly because I used D in flake8-docstrings which integrates pep257 into flake8.

[keleshev]

D123 sounds good. I guess it will be a net-positive for Python community to allow to silence the errors, since otherwise many people wouldn't follow PEP 257 at all.

[JensRantil]

D123 sounds good. I guess it will be a net-positive for Python community to allow to silence the errors, since otherwise many people wouldn't follow PEP 257 at all.

+1 I've had the opportunity to work with pep8 and like its concept of error codes.

[keleshev]

@JensRantil how did you use them?

[JensRantil]

@JensRantil how did you use them?

I used them to skip a certain error that was incorrectly reported. I then filed a bug upstream.

[Nurdok]

I'm working on this in my fork. Expect a pull request later today.

[Nurdok]

I think this issue can be closed now.