Sunday, June 14, 2015

Takayuki Shimizukawa asked me to test whether Eric Larson’s pull request is still okay regarding #1061.

First challenge for me was to activate Eric’s pull request. I tried to configure Erics fork as a remote to my working copy, but didn’t really advance… Here is how I finally did it:

$ mv sphinx sphinx-main
$ git clone https://github.com/Eric89GXL/sphinx.git
Cloning into 'sphinx'...
remote: Counting objects: 40969, done.
remote: Total 40969 (delta 0), reused 0 (delta 0), pack-reused 40969
Receiving objects: 100% (40969/40969), 12.39 MiB | 1.12 MiB/s, done.
Resolving deltas: 100% (31193/31193), done.
Checking connectivity... done.
$ git checkout fix-no-members
Branch fix-no-members set up to track remote branch fix-no-members from origin.
Switched to a new branch 'fix-no-members'
$

The problem I needed to fix with #1061 was that Sphinx was documenting also imported members of a module. Especially in modules which have the following line:

from django.db.models import Q

So I re-built the Lino docs using Eric’s version. Seems to work. For example lino.modlib.vat.models does not mention any imported things from Django.

But there was another issue which accidentally got committed together with this problem.

So I also wrote a test case to see whether Sphinx tells me the source file when an error orrurs when executing the content of a atelier.sphinxconf.insert_input. The following renders as a paragraph containing “1 2 3 4 5 6 7 8 9”:

1 2 3 4 5 6 7 8 9 ()

My manual test consists in adding a syntax error into above code snippet, trying to build my blog and verifying that the file name and line number of the syntax error are being reported correctly.

Everything seems to work.

Oops, I noticed only later that I have to switch to the fix-no-members branch above. But fortunately the result was the same.

TODO: add a test case for Sphinx which tests the above automatically.