Re: goals for 3.1

[Shai Ayal]

On Dec 20, 2007 9:15 PM, John W. Eaton <address@hidden> wrote:
> On 20-Dec-2007, Shai Ayal wrote:
>
> | However today the gnuplot backend does
> | a lot more then drawing -- The first thing that comes to mind is
> | determining tick marks,
>
> I think tic mark locations should be computed in the database code
> whenever data is added to the axes.  This is needed so that things
> like
>
>   x = -10:0.1:10; plot (x, sin(x)); get (gca, 'xtick')
>
> can work correctly (i.e., the 'xtick' property is set correctly before
> drawnow is called).  We recently made a similar change for the
> {x,y,z}lim properties.  We just need to do something similar for tic
> locations and labels.  Then the backend will just display these tic
> marks and labels rather than compute them.

I am inclined to agree, but It might be a little more complicated --
usually you would want to use information on the pixel-size of the
axis and the pixel-size of the labels to compute the correct number of
labels (i.e. in a small plot window, you wouldn't want the labels to
overwrite one another). The pixel-size of the axes can be easily
computed form the figure.position & axes.position properties, but the
label size is harder to get at -- My idea is to create a "dummy" text
object and get it's size from it's properties, and then use these
metrics to compute the pixel size of labels. This is quite ugly though
...

> | but I'm sure that other things will pop-up.
>
> Another thing is the legend, which is maybe more complicated than tic
> marks.  For compatibility, it should be a separate axes object and
> added to the list of children of the figure in which it appears.
>
> As long as the gnuplot backend is working and more capable than the
> new backend, I think we can continue to use the gnuplot backend as the
> default, with the new backend available as an option.

I agree completely. I think it will take more then 3.1 until we have
something close to gnuplot in terms of sheer number of output formats
and quality.

> jwe
>