update question

[Bernhard Reiter]

Hi Didrik,

On Thursday 28 September 2006 18:27, Didrik Pinte wrote:
> I've made little bugfix and one change to the code. Before commiting, i
> need your opinion on it.
>
> Here is the future change log :
> -----------------------------------------------------------------------
> 2006-09-28 Didrik Pinte <dpinte at itae.be>
>
>     * Thuban/UI/classgen.py: bugfix due to wx2.6 update
>
>     * Thuban/UI/view.py: removed non needed declaration
>
>     * Thuban/UI/viewport.py : extracted method from LabelShapeAt to
>     GetLabelPosForShape (see feature request #121 on wald)
>
>     * Thuban/UI/selection.py : typo correction
> -----------------------------------------------------------------------

my suggestion is to submit in small blocks that belong together.
So if you have a bunch of simple bugfixes this is fine.
Execept for the viewport.py change, I would directly commit
the smaller fixes.


> [1] Concerning the modification of the Thuban/UI/viewport.py file, i've
> just extracted some code of the LabelShapeAt method in order to be used
> by a more general method that will allow the user to label a complete
> layer by selecting a field of the associated table. See
> http://wald.intevation.org/tracker/index.php?func=detail&aid=121&group_id=6
>&atid=108 The new method GetLabelPosForShape is at the moment in the
> viewport.py file but IMHO, it could be moved to the Thuban/Model/map.py
> file. What do you think about this ? Is there an other place that could be
> more appropriate ?

Extracting methods, if you need the functionality from different places
in the code is the right idea. 
(I am currently commenting offline.)
Without having seen the method your are extracting I cannot comment
on the best place. From looking at LabelShapeAt I believe it works quite 
closly to windows coordinates which I would consider UI dependent.
So what is your idea in moving thi sot Model/map.py?

> [2] What do you think about adding a new tool allowing the user to label
> all the shapes of an existing layer (not only one shape like it is
> now) ? If the answer is yes, i can easily integrate the work i've just
> done into the trunk.

This is a feature I would like.
Automatical labelling can become quite an art easily, if you take cartographic
ideas into account, like try to position the labels nicely so that other 
features can be intentified well and so on.
Any progress is appreciated. ;)

Best,
Bernhard