Re: figure positioning problem

[Konstantinos Poulios]

On Fri, Jun 8, 2012 at 8:01 AM, Konstantinos Poulios
<address@hidden> wrote:
> On Mon, Nov 21, 2011 at 2:30 PM, Ben Abbott <address@hidden> wrote:
>> On Nov 21, 2011, at 2:13 AM, John W. Eaton wrote:
>>
>>> On 21-Nov-2011, "John W. Eaton" <address@hidden> wrote:
>>>
>>> | I checked it in.
>>>
>>> Then I backed it out, because I noticed that it caused worse problems
>>> with some other figures, making them blank and with completely wrong
>>> sizes.
>>>
>>> jwe
>>
>> John,
>>
>> I've been looking at problem with printing using FLTK. For example, the 
>> images produced below are different.
>>
>>        close all
>>        plot (rand (2)); print -r200 test1.png
>>        drawnow
>>        print -r200 test2.png
>>
>> Also, on MacOS using fltk-1.1.10-aqua, the plot canvas is not the proper 
>> size. It looks to be off by about half the status bar (menu bar?) height.
>>
>> In any event, can you provide an example of the problem you are seeing? That 
>> way I can keep an eye on it while I poke around looking for a solution to 
>> the problems above.
>>
>> Ben
>
> Hi Ben, the attached patch seems to solve the issue with your last
> test. Actually the situation with figure positioning is quite a mess
> because the information has to go in both directions:
>
> - the toolkit touches the figure position property through
> set_boundingbox and this can be triggered by different X events during
> Fl:check() (e.g. when the user resizes the window).
> - octave touches the figure position through set_position directly
> (e.g. on user request through set(gcf, ....)).
>
> The attached patch fixes a kind of race condition, where on my system
> most times the toolkit managed to reset the figure positioning through
> some X event before octave managed to propagate its position request
> to the toolkit and the Xserver. So, the value requested by octave was
> reverted to the last value X was aware of.
>
> Please test the attached patch so that we can push it.
>
> Kostas
>
> P.S.: one point changed by this patch is:
> fp.set_boundingbox (pos, /* do_notify_toolkit */ true);
> to
> fp.set_boundingbox (pos, /* do_notify_toolkit */ false);
> it doesn't make any sense to me that the toolkit does a change and at
> the same time requests to be notified about this change.


oh, actually it should be
fp.set_boundingbox (pos, true, /* do_notify_toolkit */ false);
I missed the second argument but never mind it still doesn't work anyway.

Either the window will not be resizable, or Ben's test will give
negative results. I have to look again into this.

Kostas