Re: problem with "set x11 {size WW,HH} {position XX,YY}"

[Ben A. <bpa...@ma...>]

On Dec 29, 2008, at 9:00 PM, Daniel J Sebald wrote:

> Ben Abbott wrote:
>> On Dec 29, 2008, at 1:34 PM, Ethan Merritt wrote:
>>> On Monday 29 December 2008 10:04:32 Ben Abbott wrote:
>>>
>>>> On Dec 29, 2008, at 12:25 PM, Ethan A Merritt wrote:
>>>>
>>>>
>>>>> On Monday 29 December 2008, Ben Abbott wrote:
>>>>>
>>>>>
>>>>>>>>>> Are you suggesting that Octave's plot stream contain only one
>>>>>>>>>> "set
>>>>>>>>>> terminal ..."? ... if so by what means is it possible to  
>>>>>>>>>> clear
>>>>>>>>>> the
>>>>>>>>>> canvas and start from scratch?
>>>>>>>>>
>>>>>>>>> That happens every time you issue a "plot" command.  You don't
>>>>>>>>> have
>>>>>>>>> to do anything special.  Although there is a "clear" command  
>>>>>>>>> if
>>>>>>>>> for
>>>>>>>>> some reason you want to blank the window but not draw  
>>>>>>>>> anything.
>>>>>>>>
>>>>>>>> I'm not sure I understand the context correctly, but ... what  
>>>>>>>> of
>>>>>>>> multiplot? ... there can be multiple plot commands that do not
>>>>>>>> delete
>>>>>>>> anything correct? ... this is the way Octave draws everything.
>>>>>>>
>>>>>>> It probably should not do that.  Do you know why it does?
>>>>>>
>>>>>> Octave & Matlab each place multiple plots in each figure  
>>>>>> window. By
>>>>>> multiple plots, I mean multiple axes in one window. Some axes  
>>>>>> might
>>>>>> be
>>>>>> 2D of various types and others 3D.
>>>>>
>>>>> It seems to me that this would be better done by opening separate
>>>>> windows
>>>>> for each plot, perhaps tiled to be adjacent.  That would let  
>>>>> you  edit
>>>>> and interact with each component plot separately, which is not
>>>>> possible
>>>>> for the separate components of a multiplot.
>>>>>
>>>>> Easy for me to say - I am neither programming nor using Octave.
>>>>
>>>> I agree. However, that would break compatibility with Matlab ...   
>>>> which
>>>> is actually what I'm attempting to improve.
>>>
>>> Well, I'm totally lost.  I don't understand why it would make any
>>> difference to the user, other than the gain in functionality.
>>>
>>>
>>>> Am I correct in inferring that rotation of plots does work but   
>>>> zooming
>>>> does not (which is what I observe for Octave's gnuplot windows on  
>>>> Mac
>>>> OSX).
>>>
>>> No, this is not correct.  At least, not if you have correctly   
>>> described
>>> how Octave uses multiplot. You cannot interact with the individual
>>> component plots of a multiplot.  Not zooming, not rotation,
>>> not mouse tracking, no toggling ruler/grid/logscale, none of that.
>>> Doing things inside multiplot severely limits the features that
>>> gnuplot would normally give you automatically.
>>>
>>>
>>>>> you still have not explained why you (or Octave) would
>>>>> want to close and re-open the terminal plot window before starting
>>>>> a new plot.  What do you gain by this?
>>>>
>>>> Opps, I apparently wasn't clear in my description.
>>>>
>>>> Octave may have many plot windows open simultaneously.  Each  
>>>> gnuplot
>>>> window contains one Octave figure whose objects may be modified  
>>>> via a
>>>> command line interface at the users discretion. The user may  
>>>> close a
>>>> figure's window either via the mouse or via the command line.
>>>>
>>>> I've added support for the position and size options of the x11
>>>> terminal to my local code for octave. I notice that the x11  
>>>> terminal
>>>> only respects these properties for the first instance of "set  
>>>> termina
>>>> x11 ..." in the plot stream. This is different than how the aqua  
>>>> and
>>>> wxt terminals behave. Meaning that each time I change the terminals
>>>> size, both the aqua and wxt windows change their size
>>>
>>>
>>> I think you are either doing something wrong, or failing to explain
>>> the difference between what you are seeing and what you are  
>>> expecting.
>>> The following sequence of commands will open three separate x11 plot
>>> windows with different sizes.  Each window can be managed  
>>> separately.
>>> This behaviour may not be truly identical to that for wxt, but so  
>>> far
>>> as I know the differences are subtle rather than dramatic.
>>> Do you have a counter-example?
>>>
>>>  set term x11 1 title "Plot window #1"
>>>  plot foo
>>>  set term x11 2 title "Plot window #2"
>>>  splot baz
>>>  set term x11 3 title "3rd plot window"
>>>  plot blech
>>>
>>>  # Close window 2, leave 1 and 3 unchanged:
>>>  set term x11 2 close
>>>
>>>  # Replace the contents of window 1, leaving window 3 unchanged
>>>  set term x11 1
>>>  plot new-stuff
>>>
>>>
>>>
>>>> Would it be possible for the x11 terminal to behave in the same   
>>>> manner?
>>>
>>> Please describe "same manner" more explicitly.'
>> Ok, try this
>> 	set term x11 1 size 500,400 position 100,100
>> 	plot sin(x)
>> 	set term x11 1 size 500,200 position 300,100
>> 	close 1
>> 	plot sin(x)
>> After the second plot command the window remains the same size and  
>> in  the same position as the first.
>> If I use the mouse to close the window prior to the second "plot   
>> sin(x)" then I get the effect I desire.
>
> Ben,
>
> I haven't followed this thread closely, but keep in mind that Octave  
> is currently programmed to create a new instance (process, version,  
> whatever) of gnuplot for every plot it creates.  John did things  
> that way because gnuplot currently only has the mouse active in one  
> window due to the fact that plot data is not retained but for the  
> most recent plot.  John figured gnuplot is small, so launch however  
> many versions of it that are required.
>
> So, I think that when you typed "close 1", gnuplot was completely  
> exited and a new plot command starts gnuplot from fresh.  However,  
> when you used the mouse to close the window, the X11 window was  
> closed but gnuplot_x11 is still active so retains the information  
> about the window size before it was externally closed (not by  
> Octave, but by the mouse).
>
> Some work was done on retaining plotting data for all plots but I  
> think it is only in experimental stage right now.  Such a feature  
> would enable Octave to not require opening a new version of gnuplot  
> for every new Octave plot.  Also, my guess is that gnuplot would be  
> able to handle multiplot more robustly with data retention.  There  
> is a bit of work on both sides of the exchange to do that integration.
>
> Dan

ah-ha

I'd not been aware that a new instance of gnuplot was created for each  
plot (although it is quite obvious now that you mention it).

Thanks
Ben