Add class and line number to warning about unknown section

[amueller]

Fixes #99

[amueller]

It looks like right now it raises multiple warnings for each occurrence (for sklearn) but the test only shows one warning... hm...

[jnothman]

I've not checked this in the case of the current patch, but one thing to be careful with is that the object that a docstring applies to is not necessarily in the same file as where the docstring is defined. Apart from directly setting __doc__ to some variable from elsewhere (which doesn't happen in scikit-learn, but does in Pandas and, I think, Numpy), this happens often through inheritance.

Btw, @amueller, you may be interested in jnothman@1d21eb1 where I get line offsets within a docstring for parameter entries. I just haven't quite pulled this together into something useful yet :|

[amueller]

Hmm yeah good point. But I'm a fan of iterative solutions ;) This one is helpful in many cases, but not all. Inheritance might be a reason why I get multiple warnings in sklearn. I'll check out your commit. Last time I checked the munging made it really hard to get corresponding lines. I was thinking that displaying the offending line might be easier. Sent from phone. Please excuse spelling and brevity. On Jul 19, 2017 20:03, "Joel Nothman" <[email protected]> wrote: I've not checked this in the case of the current patch, but one thing to be careful with is that the object that a docstring applies to is not necessarily in the same file as where the docstring is defined. Apart from directly setting __doc__ to some variable from elsewhere (which doesn't happen in scikit-learn, but does in Pandas and, I think, Numpy), this happens often through inheritance. Btw, @amueller <https://github.com/amueller>, you may be interested in jnothman/numpydoc@1d21eb1 <jnothman@1d21eb1> where I get line offsets within a docstring for parameter entries. I just haven't quite pulled this together into something useful yet :| — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#101 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAbcFttvjqutC5RZRwyavWk2mG0rdhamks5sPpk9gaJpZM4OdLXw> .

[stefanv]

I ran into this exact error recently: it said "error in None from None", or something like that. So, I'm all for iterative improvement.

[jnothman]

Don't get me wrong, I'm all for this change.

…

On 20 July 2017 at 15:48, Stefan van der Walt ***@***.***> wrote: I ran into this exact error recently: it said "error in None from None", or something like that. So, I'm all for iterative improvement. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#101 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz66K5fX1Gu2JYU9Eiq_KoRECOYxoFks5sPuo2gaJpZM4OdLXw> .

[rgommers]

+1 helpful indeed. Doesn't work for everything but does for plain Python code at least. Just tried it on SciPy and got a couple of each of these:

/Users/rgommers/Library/Python/3.5/lib/python/site-packages/numpydoc/docscrape.py:360: 
UserWarning: Unknown section Exampes in the docstring of None in None.
  warn(msg)
/Users/rgommers/Library/Python/3.5/lib/python/site-packages/numpydoc/docscrape.py:360: 
UserWarning: Unknown section Parameers in the docstring of <function periodogram at 0x10ac9e840> in /Users/rgommers/Code/scipy/scipy/signal/spectral.py.
  warn(msg)

[rgommers]

@amueller not sure if you wanted to check out that commit of @jnothman for this PR, but I'd be happy to merge this now.

[rgommers]

@jnothman and @amueller looks like you are both up to speed with the numpydoc internals and have an interest in the package. Do you want commit rights?

[amueller]

@rgommers Maybe I should remove the "of None in None"? But I guess that's details.
Happy to have this merged now, or I can detect "None".
And I'd take commit rights, but I can't promise I'll review anything. I'm a couple hundred reviews behind on sklearn ;)

[stefanv]

Yes, please spread the commit rights to numpydoc! Thanks for taking initiative, Ralph.

[rgommers]

Maybe I should remove the "of None in None"? But I guess that's details.

Yeah that doesn't bother me, removing it won't locating the offending line easier.

And I'd take commit rights, but I can't promise I'll review anything. I'm a couple hundred reviews behind on sklearn ;)

Excellent, team made (https://github.com/orgs/numpy/teams/numpydoc-devs) and invite sent. And yeah, totally understand about no commitments. I suspect we're all in the same boat here:)

[jnothman]

Merge?

[amueller]

I think consensus was merge and we iterate?

[amueller]

@jnothman you should have merge rights, right? (I can't but maybe there is a "can't merge own PRs setting?")

[jnothman]

Why should I?

…

On 11 September 2017 at 01:28, Andreas Mueller ***@***.***> wrote: @jnothman <https://github.com/jnothman> you should have merge rights, right? (I can't but maybe there is a "can't merge own PRs setting?") — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#101 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAEz6_FQESv_eww238iy5BThp2LMxPExks5shAAxgaJpZM4OdLXw> .

[amueller]

Oh I thought @rgommers gave us rights, but maybe not?

[rgommers]

Oh I thought @rgommers gave us rights, but maybe not?

I asked both of you, and you said yes. @jnothman didn't respond so I didn't give him the rights. Very happy to do so now though. Do you want them @jnothman?

[jnothman]

I'm happy to have rights. Thanks

[rgommers]

Awesome! Invite sent.

[amueller]

thanks :)