documentation coverage report

[Vincent Delecroix]

Hello,

Because of #17841 we can get global statistics of documentation
coverage... and right now

Global score: 94.4% (35344 of 37421)

401 files with wrong documentation
1514 functions with no doc
563 functions with no test
310 doctest are potentially wrong

Vincent

[kcrisman]

401 files with wrong documentation

I couldn't get a good sense of what "wrong" meant from that patch.  What does that specifically refer to?

Thanks,

- kcrisman 

[Nils Bruin]

On Monday, February 23, 2015 at 10:07:13 AM UTC-8, kcrisman wrote:

I couldn't get a good sense of what "wrong" meant from that patch.  What does that specifically refer to?

 

I wondered about that too. Apparently, it's "sage --coverage" jargon. A doctest is "possibly wrong" if the name of the doctested function doesn't occur in the doctest.
I was having a faint hope of being impressed with our automatic documentation testing. If we could automatically distinguish "correct" documentation from "wrong" documentation, we could just solve our documentation woes by letting the tester run over all possible docstrings and pick out the correct documentation. Naturally, the concept turned out a little more mundane than that.

[kcrisman]

There is even a tag (I forget the exact wording) for saying that this is an indirect doctest, which hopefully this script picks out.

[Simon King]

Hi Karl-Dieter,

On 2015-02-23, kcrisman <kcri...@gmail.com> wrote:
> There is even a tag (I forget the exact wording) for saying that this is an
> indirect doctest, which hopefully this script picks out.

sage: I think the tag it is # indirect doctest

Cheers,
Simon