[Qgis-developer] Call for applications: QGIS API Documentation Improvement

[Alessandro Pasotti]

2015-08-19 21:38 GMT+02:00 Tim Sutton <tim at qgis.org>:

> Hi Ale
>
> On 19 Aug 2015, at 09:41, Alessandro Pasotti <apasotti at gmail.com> wrote:
>
> 2015-08-18 23:19 GMT+02:00 Tim Sutton <tim at qgis.org>:
>
>> Hi all
>>
>> In the last few years we have been steadily improving the amount of
>> funding we are able to accumulate in the QGIS project. Our goal in
>> obtaining funding is always to 'make QGIS better'. Up until now we have
>> focussed funding on high profile aspects of the project: Funding regular
>> hackfests, paying for bug fixing work prior to releases, funding
>> infrastructure such as servers, domain name registrations etc.
>>
>> With improved funding levels we now have the opportunity to also start
>> addressing some of the many less obvious components of QGIS that badly need
>> attention, but often don't attract volunteers. In our July 2015 PSC meeting
>> it was agreed that we would start this initiative by funding one or more
>> developers to improve the python documentation in QGIS. If you are
>> interested in working on this, please be so kind as to visit the form [1]
>> that we have set up for this.
>>
>>
>> [1] http://goo.gl/forms/WRGSvWHkBb
>>
>> Regards
>>
>> Tim
>>
>>
>
> Hi Tim,
>
> Thank you for writing this proposal:  QGIS developer documentation really
> needs some more love :)
>
>
> Yes it does!
>
>
> While reading the form, I noticed that one of the goals is
>
> * port the cookbook content over to the API documentation
>
> Does this mean the end of the cookbook? I hope not!
>
>
>
> IMHO the purpose of the cookbook is not to be a full API reference but a
> collection of small tutorials (recipes), with some introductory material.
>
> I've spent quite some time lately trying to update the cookbook by
> proofreading and testing the examplesand adding new recipes and I think
> that the cookbook is an invaluable source of informations for PyQGIS
> programmers, the biggest problem is that it is not routinely updated when
> the underlying API changes, for instance there are no recipes for
> QgsMapSettings or QgsMapRendererCustomPainterJob and a few examples use
> deprecated methods.
>
> I liked a lot the idea we discussed in Denmark about adding tests directly
> in the rst sphynx code, this would be a nice improvement for the code
> examples and snippets.
>
> I think that the cookbook must continue to exist, it would be awesome if
> the process of writing code examples could be automated to transfer them to
> the cookbook, adding the snippets to travis would be a nice start but we
> all should definitely dedicate more resources for keeping the cookbook in
> sync with the source code changes.
>
>
>
>
> I also really like the cookbook, though I worry about the maintenance of
> it - if we can do the tests in ret to make sure that it remains current
> that would be one good motivation for keeping it current. An alternative
> approach would be to move the content of the cookbook into doxygen
> documentation (you can add arbitrary pages there - they don’t all need to
> be generated from header files.
>
>

> But let me rather step back and say that my intention in the proposal was
> more that we sponsor someone to really look into these things and come up
> with a well established standard - something like the django folks have
> done in your link below. We obviously don’t currently have funding for a
> complete revision of every single source file, but I think if the person we
> hire can establish a good working system, it will be easier to point others
> to when they submit patches, telling them ‘we need docs with that!’ - and
> pointing the to examples and instructions on how to document their code
> nicely.
>


Ok, agreed.

The Django folks are very picky about "Have documentation" and "Have tests"
before accepting any patch, and the documentation always have examples.

Doxygen also accepts RST and we could place python documentation into .sip
files, maybe a solution would be to include snippets and examples from
doxygen (.sip is preferred) into the cookbook, leaving the introductory
material in the cookbook itself interleaved with included examples and
snippets.

I don't know about QT but most of the (not API) documentation I've seen is
made with sphinx, I might be wrong (maybe is just that I know sphinx better
than doxygen) but I suspect that moving the whole cookbook into doxygen and
outside from the rest of the documentation (user manual, tutorial etc.) is
not a good idea.


-- 
Alessandro Pasotti
w3:   www.itopen.it
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20150820/f8c4b411/attachment.html>