openteachermaintainers team mailing list archive

[Marten de Vries <marten-de-vries@xxxxxxxxxxx>]

Hello Cas, *

Cas Widdershoven schreef op ma 04-04-2011 om 19:22 [+0200]: 
> Hello OpenTeacher maintainers and others,
> 
> one of the blueprints still open on Launchpad is
> https://blueprints.launchpad.net/openteacher/+spec/morethan2answers.
> While talking on IRC, Marten mentioned it to me, and asked if we should
> close it. When I saw it, I wanted to see how it was implemented (in the
> otwd file format).
I personally believe this blueprint and the design of the .otwd format
are two different things. This also implies I believe we can just close
that blueprint as 'Implemented' if there are no other complaints and
still continue the discussion on the second point.

> I saw that the results are only saved in the specific
> tests the result was accomplished, and I thought that isn't the correct
> way of doing it. I already discussed this with Marten, who thinks this
> ís the right way to do it, because he didn't want to save information
> twice (once at the specific test and once at the word itself).
> In conclusion:
> I think the results of a word must at least be saved at the word itself,
> and also at the test (easier code, easier for a human to interpret the
> file), but Marten thinks it has to be saved at the tests, and we should
> get the 'complete' results coding (looping through all the tests), which
> leaves us without saving things twice, but it needs some extra coding to
> get the overall results.
Nice summary. I indeed believe that double data (which has to be updated
in two places) is an important possible source of bugs, and that finding
the results by some code isn't really a problem (dictionary lookups are
one of the most optimized parts of Python, so speed isn't the problem).

> What do you think about it?
> 
Question for Milan, I guess.

-- 
Marten de Vries