Re: units updater for figures is broken when using fltk

[Ben Abbott]

On Feb 12, 2012, at 4:16 PM, Michael Goffioul wrote:

> On Sun, Feb 12, 2012 at 9:06 PM, Ben Abbott <address@hidden> wrote:
>>>>>> Do you think this is an appropriate approach?
>>>>>> 
>>>>>> Or, might a 3rd boolean input be added get_boundingbox to allow the 
>>>>>> origin to be selected as LL or UL ?
>>>> 
>>>> Possibly, but I don't see the usefulness at the moment. Any windowing
>>>> toolkit I know always uses the UL corner as origin. Only Matlab
>>>> doesn't use that reference, god only knows why... ;)
>>> 
>>> ah-ha ... I see why I've been confused by your comments.
>>> 
>>> You are correct, Matlab does use the LL as its reference for the figure's 
>>> position property, but Octave's fltk toolkit is using the UL.
>>> 
>>> I ran the code below with 3.6.1-rc1and the default branch. Both give the 
>>> same result.
>>> 
>>>       close all
>>>       graphics_toolkit gnuplot
>>>       figure (1, "position", [50 50 400 300])
>>>       graphics_toolkit fltk
>>>       figure (2, "position", [50 50 400 300])
>>> 
>>> Figure 1 is in the LL corner of the screen, and figure 2 is in the UL 
>>> corner.
>>> 
>>> Octave's backends are not behaving the same for me on MacOS X or Ubuntu. 
>>> Looks like a bug.
> 
> That's my suspicion from the beginning (when I said FLTK didn't honor
> the Y-axis flipping). The code in octave (like
> get_boundingbox/set_boundingbox) is correct, but FLTK backend
> interpret things incorrectly.
> 
>> The attached chageset results in both figures being place in the LL corner 
>> of my display.
>> 
>> If there are no comments, I'll push this later today.
> 
> This is the correct way to handle things, and as you can see, it's way 
> simpler.
> 
> Michael.

I've pushed the changeset.

        http://hg.savannah.gnu.org/hgweb/octave/rev/adb352685ded

Sorry it took so long for me to catch on.

Ben