Re: fast scatter plots - advice sought

[Jaroslav Hajek]

On Mon, Jan 11, 2010 at 9:32 PM, David Bateman <address@hidden> wrote:
> Michael D Godfrey wrote:
>>
>> "matlab compatibility" should not be carried too far here.  How many users
>> would be
>> affected by changes in the children hierarchy?  Absolute matlab
>> compatibility would require
>> the the handle values returned by Octave to match those in matlab.  I do
>> know of one program which
>> depends on this!  But, I think too much has been made of the compatibility
>> issue in this case already.
>> Matching functionality and achieving best performance should be the main
>> point.  If I understand
>> the current proposed code, it does this very well (subject to testing and
>> verification for both
>> gnuplot and fltk backends).
>
> Agreed but don't go looking for incompatibility unless there is a damn good
> reason.. If there are several ways of doing something picking the matlab
> compatible manner is safest..
>

Perhaps, but performance of plotting *is* a damn good reason. A
scatter plot taking a minute is almost as good as if it didn't work at
all. Just count the people on the ML worried about plotting
performance vs. those that are worried about compatibility of
primitive representations. OK, the fltk backend will improve
performance in general, but this was one case where the lousy
performance was not a backend issue.

> Line objects don't support colormaps and the example I gave won't work with
> Jaroslav's code.. This seems a good reason "to keep" matlab compatibility to
> me (ie using patch)
>

OK, I copied the old patch constructors back in there. Hence, your
example now works. I take it that lines don't support colormaps in
Matlab either?

> Though note that the mess in __go_draw_axes__ to handle the colormaps will
> probably make Jaroslav's change much slower using patch objects, though
> still faster for a larger number of points..
>

Actually, it doesn't seem to be significantly slower for the main
cases I care about (plotting lots of points in the same color).

However, the "filled" option does not work now:
lt-octave:1> scatter (rand (1, 20), rand (1, 20), "filled")
error: invalid color specification: flat
error: invalid value for color property "markerfacecolor" (value = flat)
error: called from:
error:   /home/highegg/devel/octave/scripts/plot/private/__scatter__.m
at line 154, column 11
error:   /home/highegg/devel/octave/scripts/plot/scatter.m at line 70, column 7

and the result was the same even prior to all these patches. Can
anyone give me a clue how to correct this?

regards

-- 
RNDr. Jaroslav Hajek, PhD
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz