On Thu, Jul 05, 2012 at 06:47:35PM +0200, Olivier Grisel wrote:
> None of this is doctested. And I don't want to put pollute the code
> with boilerplate to make that testable. It's up to the teacher to
> check that those exercises still work prior to using them in an
> interactive session.

I afraid that disagree with you here. Tutorials in the documentation are
not only for a one time use. They are exposed on Internet and when they
get obsolete we get frustrated users. This was actually one of the
motivations to move the tutorials in the main scikit-learn documentation
and version the tutorials with the main codebase.

When I recently tried to teach using the scipy-lecture note, I found that
quite a few set of interactive examples didn't work: the authors of the
corresponding chapters had probably not typed them in the tutorial
exactly as they had done in their interactive session. It took me quite a
while to fix all this, and I made sure that everything was doctested.

If the model is that the teacher maintains them and makes sure that they
work before a class, then it seems to me that the tutorials should not be
merged in the main documentation.

I do agree that the notebook adds value to such a workflow, and this is
why I would like a way to provide notebook in addition to what we have.

Gaël

PS: Using the literalinclude directive with the start-after/end-before is
actually a fairly easy way to provide a skeleton in sphinx.

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to