Fixed docs (n2 diagram)

[kanekosh]

Purpose

Fixed the documentation to accommodate a new format of n2 diagram in Aerodynamic Optimization Walkthrough.

Type of change

• Documentation update

Testing

Local build succeeded.

[ewu63]

Looks good to me. I don't think there's a way to do a PR build with GitHub Pages unfortunately. Was this the only N2 diagram in the docs?

[kanekosh]

Yes this was the only one.

[coveralls]

Coverage remained the same at 95.877% when pulling 247068e on kanekosh:fix_docs into eb9d2b4 on mdolab:master.

[ewu63]

I think it would be good to get @johnjasa to double check this, as I'm not really familiar with how sphinx works with Travis/GitHub Pages for building/deploying docs.

[johnjasa]

Thanks for this fix, Shugo. I like how you're using the embed-n2 directive from OM now.

I tried building the docs locally in the openaerostruct/docs folder by using make html, but they failed with this message:

  File "/mnt/c/Users/jjasa/git/OpenMDAO/openmdao/docs/_exts/embed_n2.py", line 56, in run
    raise IOError('File does not exist({0})'.format(np))
OSError: File does not exist(/mnt/c/Users/jjasa/git/OpenAeroStruct/openaerostruct/docs/openaerostruct/docs/aero_walkthrough/generate_n2.py)

Basically, the path was wrong to the generate_n2.py file with duplicate openaerostruct/docs parts there. Does the local build work for you? This might be something from before and not from your changes. I'll keep looking at it a bit.

[johnjasa]

Don't try tackling it - I think I've got it. Can fix it locally in just a bit and I'll push to a branch.

[kanekosh]

Thanks John for checking this! The local build actually fails for me as well. When we build locally in openaerostruct/docs, the path origin is openaerostruct/docs. However with Travis, it seems the origin is OpenAeroStruct and Travis test fails in that case.
But I guess I'm missing something as your embed-code works but embed-n2 does not.

[johnjasa]

I see, that's right on the different origin directories. Do you think the paths I changed in aero_walkthrough.rst will work for Travis, then? They work locally, but it seems there might be different behavior.

Does us including the docs path here help that at all? Maybe I'm misinterpreting what that does, it's been a while.

[johnjasa]

I guess what it comes down to is that it'd be nice if the docs can be made both on Travis and locally, in my opinion. Do you think that's worth sorting out, or should we say Travis-only is okay?

[kanekosh]

I think it will fail, at least for embed-code. Actually my initial commit was similar and it failed.
But since the paths seem to behave differently between embed-code and embed-n2 when built locally, I don't know what will happen with Travis.
I agree, we should be able to check if Travis works by doing make html locally, otherwise the same thing like this will happen and bother reviewers again.

[johnjasa]

Oof, I should've scrolled up and checked that. Thanks for clarifying. I've pushed a fix to the .travis.yml to change directories, hopefully it works.

[johnjasa]

Neat, the build looked okay on Travis. All good in my book!

[johnjasa]

Wait, Coveralls looks wacky still: https://coveralls.io/jobs/63931317
I tried to fix it but I'm not certain if that'll do it. I see it was reported as 0% earlier, too.
@kanekosh, do you have any ideas for that?

[kanekosh]

make html works on my local too, thanks!
I think Coveralls will work fine this time. When I first passed Travis build (but not locally) in 94c9498, where Travis was working on OpenAeroStruct dir, there was no problem.

[johnjasa]

Okay, great! If Coveralls is still weird, I'll leave it up to you to handle. :)

[ewu63]

Is this PR ready to be merged? Happy to do so as long as you guys ironed out the issues @johnjasa @kanekosh

[kanekosh]

It's ready!