Subplot problem in Octave 4.4.0?

classic Classic list List threaded Threaded
10 messages Options
Reply | Threaded
Open this post in threaded view
|

Subplot problem in Octave 4.4.0?

Tony Richardson
I've got a couple of scripts that create figures with several subplots.  There are a couple of gui slider controls that, when changed, trigger redrawing of some of the subplots.  Under 4.4.0 some of the subplots disappear when redrawn.  The scripts still work fine under 4.2.2.  I have attached one of the scripts.  (I have tried to come up with a simpler script that demonstrates the problem, but have not been able to do so.  The original script reads in sampled audio from a large file, but I changed the script to use simulated input data.)

Run the script, change around some of the gui slider controls and some of the subplots will disappear.  This doesn't happen under 4.2.2.  

This is under Windows 10 with the Qt graphics toolkit.  I don't have 4.4.0 running under Linux yet.  I'll try to test that later today.

I realize this may be too complicated an example to easily debug.  Maybe someone else is seeing similar behavior in a simpler script though?

Regards,
Tony Richardson




bugtest.m (15K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Subplot problem in Octave 4.4.0?

Pantxo
Tony Richardson wrote

> I've got a couple of scripts that create figures with several subplots.
> There are a couple of gui slider controls that, when changed, trigger
> redrawing of some of the subplots.  Under 4.4.0 some of the subplots
> disappear when redrawn.  The scripts still work fine under 4.2.2.  I have
> attached one of the scripts.  (I have tried to come up with a simpler
> script that demonstrates the problem, but have not been able to do so.
> The original script reads in sampled audio from a large file, but I
> changed the script to use simulated input data.)
>
> Run the script, change around some of the gui slider controls and some of
> the subplots will disappear.  This doesn't happen under 4.2.2.  
>
> This is under Windows 10 with the Qt graphics toolkit.  I don't have 4.4.0
> running under Linux yet.  I'll try to test that later today.
>
> I realize this may be too complicated an example to easily debug.  Maybe
> someone else is seeing similar behavior in a simpler script though?
>
> Regards,
> Tony Richardson
>
>
>
>
>
> bugtest.m (15K)
> <http://octave.1599824.n4.nabble.com/attachment/4687876/0/bugtest.m>

Hi,

On linux changing the slider values works but resizing the figure window
does make subplots disappear. Could you file a bug report please?

Thanks,

Pantxo



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-General-f1599825.html


Reply | Threaded
Open this post in threaded view
|

RE: Subplot problem in Octave 4.4.0?

Tony Richardson
I recently filed a bug report (53670) about subplots disappearing with window resizing.  That bug is present in previous versions of octave as well.
The problem I am having now is new in 4.4.0 (although it may be related to bug 53670).  

Tony

> -----Original Message-----
> From: Help-octave [mailto:help-octave-
> bounces+richardson=[hidden email]] On Behalf Of Pantxo
> Sent: Friday, May 4, 2018 3:25 AM
> To: [hidden email]
> Subject: Re: Subplot problem in Octave 4.4.0?
>
> Tony Richardson wrote
> > I've got a couple of scripts that create figures with several subplots.
> > There are a couple of gui slider controls that, when changed, trigger
> > redrawing of some of the subplots.  Under 4.4.0 some of the subplots
> > disappear when redrawn.  The scripts still work fine under 4.2.2.  I
> > have attached one of the scripts.  (I have tried to come up with a
> > simpler script that demonstrates the problem, but have not been able to
> do so.
> > The original script reads in sampled audio from a large file, but I
> > changed the script to use simulated input data.)
> >
> > Run the script, change around some of the gui slider controls and some
> > of the subplots will disappear.  This doesn't happen under 4.2.2.
> >
> > This is under Windows 10 with the Qt graphics toolkit.  I don't have
> > 4.4.0 running under Linux yet.  I'll try to test that later today.
> >
> > I realize this may be too complicated an example to easily debug.
> > Maybe someone else is seeing similar behavior in a simpler script though?
> >
> > Regards,
> > Tony Richardson
> >
> >
> >
> >
> >
> > bugtest.m (15K)
> >
> <http://octave.1599824.n4.nabble.com/attachment/4687876/0/bugtest.m
> > >
>
> Hi,
>
> On linux changing the slider values works but resizing the figure window does
> make subplots disappear. Could you file a bug report please?
>
> Thanks,
>
> Pantxo
>
>
>
> --
> Sent from: http://octave.1599824.n4.nabble.com/Octave-General-
> f1599825.html
>



Reply | Threaded
Open this post in threaded view
|

RE: Subplot problem in Octave 4.4.0?

Tony Richardson
In reply to this post by Pantxo
I just tested Octave 4.4.0 under Linux (Ubuntu/flatpak) and the same bug is present -
subplots disappear after adjusting the sliders (not necessarily after the first adjustment,
but it doesn't take long).  Did you test under 4.4.0?  The maximize bug has been
present for a while, but the disappearing subplots (without window resizing) is new to 4.4.0.

Tony

> -----Original Message-----
> From: Help-octave [mailto:help-octave-
> bounces+richardson=[hidden email]] On Behalf Of Pantxo
> Sent: Friday, May 4, 2018 3:25 AM
> To: [hidden email]
> Subject: Re: Subplot problem in Octave 4.4.0?
>
> Tony Richardson wrote
> > I've got a couple of scripts that create figures with several subplots.
> > There are a couple of gui slider controls that, when changed, trigger
> > redrawing of some of the subplots.  Under 4.4.0 some of the subplots
> > disappear when redrawn.  The scripts still work fine under 4.2.2.  I
> > have attached one of the scripts.  (I have tried to come up with a
> > simpler script that demonstrates the problem, but have not been able to
> do so.
> > The original script reads in sampled audio from a large file, but I
> > changed the script to use simulated input data.)
> >
> > Run the script, change around some of the gui slider controls and some
> > of the subplots will disappear.  This doesn't happen under 4.2.2.
> >
> > This is under Windows 10 with the Qt graphics toolkit.  I don't have
> > 4.4.0 running under Linux yet.  I'll try to test that later today.
> >
> > I realize this may be too complicated an example to easily debug.
> > Maybe someone else is seeing similar behavior in a simpler script though?
> >
> > Regards,
> > Tony Richardson
> >
> >
> >
> >
> >
> > bugtest.m (15K)
> >
> <http://octave.1599824.n4.nabble.com/attachment/4687876/0/bugtest.m
> > >
>
> Hi,
>
> On linux changing the slider values works but resizing the figure window does
> make subplots disappear. Could you file a bug report please?
>
> Thanks,
>
> Pantxo
>
>
>
> --
> Sent from: http://octave.1599824.n4.nabble.com/Octave-General-
> f1599825.html
>



Reply | Threaded
Open this post in threaded view
|

Re: Subplot problem in Octave 4.4.0?

Mike Miller-4
On Fri, May 04, 2018 at 15:01:12 +0000, Richardson, Anthony wrote:
> I just tested Octave 4.4.0 under Linux (Ubuntu/flatpak) and the same bug is present -
> subplots disappear after adjusting the sliders (not necessarily after the first adjustment,
> but it doesn't take long).  Did you test under 4.4.0?  The maximize bug has been
> present for a while, but the disappearing subplots (without window resizing) is new to 4.4.0.

If it helps narrow things down for you, I have found that it's the 'hup'
update that causes the subplots to disappear for me. For me the subplots
disappear only when the third and fourth sliders are adjusted.

If I comment out the lines that set the 'hup' state to true, so that
subplot(3,3,3) is never redrawn, then no subplots disappear at all.

Do you see the same?

--
mike



signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

RE: Subplot problem in Octave 4.4.0?

Tony Richardson
Yes, I see the same behavior.  That subplot is the only one above the
"control panel", so maybe there is some strange interaction between
the two (the control panel and the subplot).  I've got a few other scripts
that are behaving the same way under 4.4.0, I'll see if something similar
is going on with them.

Tony

> -----Original Message-----
> From: Mike Miller [mailto:[hidden email]]
> Sent: Friday, May 4, 2018 11:56 AM
> To: Richardson, Anthony <[hidden email]>
> Cc: [hidden email]
> Subject: Re: Subplot problem in Octave 4.4.0?
>
> On Fri, May 04, 2018 at 15:01:12 +0000, Richardson, Anthony wrote:
> > I just tested Octave 4.4.0 under Linux (Ubuntu/flatpak) and the same
> > bug is present - subplots disappear after adjusting the sliders (not
> > necessarily after the first adjustment, but it doesn't take long).
> > Did you test under 4.4.0?  The maximize bug has been present for a while,
> but the disappearing subplots (without window resizing) is new to 4.4.0.
>
> If it helps narrow things down for you, I have found that it's the 'hup'
> update that causes the subplots to disappear for me. For me the subplots
> disappear only when the third and fourth sliders are adjusted.
>
> If I comment out the lines that set the 'hup' state to true, so that
> subplot(3,3,3) is never redrawn, then no subplots disappear at all.
>
> Do you see the same?
>
> --
> mike

Reply | Threaded
Open this post in threaded view
|

RE: Subplot problem in Octave 4.4.0?

Tony Richardson
I've tried a few other scripts in which the control panel is in different locations.  It doesn't
appear to be due to interaction between a subplot and the control panel.
I've found that if I comment out any code that draws to subplot(3, 3, 3) then the other
subplots remain visible.  If I redraw subplot(3,3,3) the other subplots will eventually
disappear.

Tony

> -----Original Message-----
> From: Help-octave [mailto:help-octave-
> bounces+richardson=[hidden email]] On Behalf Of Richardson,
> Anthony
> Sent: Friday, May 4, 2018 1:18 PM
> To: Mike Miller <[hidden email]>
> Cc: [hidden email]
> Subject: RE: Subplot problem in Octave 4.4.0?
>
> Yes, I see the same behavior.  That subplot is the only one above the "control
> panel", so maybe there is some strange interaction between the two (the
> control panel and the subplot).  I've got a few other scripts that are behaving
> the same way under 4.4.0, I'll see if something similar is going on with them.
>
> Tony
>
> > -----Original Message-----
> > From: Mike Miller [mailto:[hidden email]]
> > Sent: Friday, May 4, 2018 11:56 AM
> > To: Richardson, Anthony <[hidden email]>
> > Cc: [hidden email]
> > Subject: Re: Subplot problem in Octave 4.4.0?
> >
> > On Fri, May 04, 2018 at 15:01:12 +0000, Richardson, Anthony wrote:
> > > I just tested Octave 4.4.0 under Linux (Ubuntu/flatpak) and the same
> > > bug is present - subplots disappear after adjusting the sliders (not
> > > necessarily after the first adjustment, but it doesn't take long).
> > > Did you test under 4.4.0?  The maximize bug has been present for a
> > > while,
> > but the disappearing subplots (without window resizing) is new to 4.4.0.
> >
> > If it helps narrow things down for you, I have found that it's the 'hup'
> > update that causes the subplots to disappear for me. For me the
> > subplots disappear only when the third and fourth sliders are adjusted.
> >
> > If I comment out the lines that set the 'hup' state to true, so that
> > subplot(3,3,3) is never redrawn, then no subplots disappear at all.
> >
> > Do you see the same?
> >
> > --
> > mike



Reply | Threaded
Open this post in threaded view
|

RE: Subplot problem in Octave 4.4.0?

Pantxo
Tony Richardson wrote

> I've tried a few other scripts in which the control panel is in different
> locations.  It doesn't
> appear to be due to interaction between a subplot and the control panel.
> I've found that if I comment out any code that draws to subplot(3, 3, 3)
> then the other
> subplots remain visible.  If I redraw subplot(3,3,3) the other subplots
> will eventually
> disappear.
>
> Tony
>
>> -----Original Message-----
>> From: Help-octave [mailto:help-octave-
>> bounces+richardson=

> evansville.edu@

> ] On Behalf Of Richardson,
>> Anthony
>> Sent: Friday, May 4, 2018 1:18 PM
>> To: Mike Miller &lt;

> mtmiller@

> &gt;
>> Cc:

> help-octave@

>> Subject: RE: Subplot problem in Octave 4.4.0?
>>
>> Yes, I see the same behavior.  That subplot is the only one above the
>> "control
>> panel", so maybe there is some strange interaction between the two (the
>> control panel and the subplot).  I've got a few other scripts that are
>> behaving
>> the same way under 4.4.0, I'll see if something similar is going on with
>> them.
>>
>> Tony
>>
>> > -----Original Message-----
>> > From: Mike Miller [mailto:

> mtmiller@

> ]
>> > Sent: Friday, May 4, 2018 11:56 AM
>> > To: Richardson, Anthony &lt;

> ar63@

> &gt;
>> > Cc:

> help-octave@

>> > Subject: Re: Subplot problem in Octave 4.4.0?
>> >
>> > On Fri, May 04, 2018 at 15:01:12 +0000, Richardson, Anthony wrote:
>> > > I just tested Octave 4.4.0 under Linux (Ubuntu/flatpak) and the same
>> > > bug is present - subplots disappear after adjusting the sliders (not
>> > > necessarily after the first adjustment, but it doesn't take long).
>> > > Did you test under 4.4.0?  The maximize bug has been present for a
>> > > while,
>> > but the disappearing subplots (without window resizing) is new to
>> 4.4.0.
>> >
>> > If it helps narrow things down for you, I have found that it's the
>> 'hup'
>> > update that causes the subplots to disappear for me. For me the
>> > subplots disappear only when the third and fourth sliders are adjusted.
>> >
>> > If I comment out the lines that set the 'hup' state to true, so that
>> > subplot(3,3,3) is never redrawn, then no subplots disappear at all.
>> >
>> > Do you see the same?
>> >
>> > --
>> > mike

Indeed with those additional information I was able to reproduce the bug
which is clearly different than the one related to resize. Here the position
of the axes is corrupted :

octave:3> get (findobj (gcf, "type", "axes"), "position")
ans =
{
  [1,1] =

     7.5645e+12   3.2287e+22  -1.5054e+13  -6.4573e+22

  [2,1] =

     7.5645e+12   3.2287e+22  -1.5054e+13  -6.4573e+22
 ...

As a workaround, I found that avoiding the call to high level "plot"
function (by changing the properties of the original line object) avoids the
issue for me. Something like

...
    subplot(3,3,3)
    persitent h333 = [];
    tt = t(1:round(L/15))*1000;
    hh = h(1:round(L/15));
    if (isempty (h333))
      plot(tt, hh);
      grid on;
      %xlim([0 tplotmax]);
      xlabel('time (ms)');
      ylabel('Amplitude');
      title('Impulse Response');
    else
      set (h333, "xdata", tt, "ydata", hh);
    endif
...

Anyway, Tony, can you file a bug report about this new bug so that the
discussion can take place there?

Pantxo



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-General-f1599825.html


Reply | Threaded
Open this post in threaded view
|

RE: Subplot problem in Octave 4.4.0?

Tony Richardson
> -----Original Message-----
> From: Help-octave [mailto:help-octave-
> bounces+richardson=[hidden email]] On Behalf Of Pantxo
> Sent: Friday, May 4, 2018 3:24 PM
> To: [hidden email]
> Subject: RE: Subplot problem in Octave 4.4.0?
>
> Tony Richardson wrote
> > I've tried a few other scripts in which the control panel is in
> > different locations.  It doesn't appear to be due to interaction
> > between a subplot and the control panel.
> > I've found that if I comment out any code that draws to subplot(3, 3,
> > 3) then the other subplots remain visible.  If I redraw subplot(3,3,3)
> > the other subplots will eventually disappear.
> >
> > Tony
> >
> >> -----Original Message-----
> >> From: Help-octave [mailto:help-octave-
> >> bounces+richardson=
>
> > evansville.edu@
>
> > ] On Behalf Of Richardson,
> >> Anthony
> >> Sent: Friday, May 4, 2018 1:18 PM
> >> To: Mike Miller &lt;
>
> > mtmiller@
>
> > &gt;
> >> Cc:
>
> > help-octave@
>
> >> Subject: RE: Subplot problem in Octave 4.4.0?
> >>
> >> Yes, I see the same behavior.  That subplot is the only one above the
> >> "control panel", so maybe there is some strange interaction between
> >> the two (the control panel and the subplot).  I've got a few other
> >> scripts that are behaving the same way under 4.4.0, I'll see if
> >> something similar is going on with them.
> >>
> >> Tony
> >>
> >> > -----Original Message-----
> >> > From: Mike Miller [mailto:
>
> > mtmiller@
>
> > ]
> >> > Sent: Friday, May 4, 2018 11:56 AM
> >> > To: Richardson, Anthony &lt;
>
> > ar63@
>
> > &gt;
> >> > Cc:
>
> > help-octave@
>
> >> > Subject: Re: Subplot problem in Octave 4.4.0?
> >> >
> >> > On Fri, May 04, 2018 at 15:01:12 +0000, Richardson, Anthony wrote:
> >> > > I just tested Octave 4.4.0 under Linux (Ubuntu/flatpak) and the
> >> > > same bug is present - subplots disappear after adjusting the
> >> > > sliders (not necessarily after the first adjustment, but it doesn't take
> long).
> >> > > Did you test under 4.4.0?  The maximize bug has been present for
> >> > > a while,
> >> > but the disappearing subplots (without window resizing) is new to
> >> 4.4.0.
> >> >
> >> > If it helps narrow things down for you, I have found that it's the
> >> 'hup'
> >> > update that causes the subplots to disappear for me. For me the
> >> > subplots disappear only when the third and fourth sliders are adjusted.
> >> >
> >> > If I comment out the lines that set the 'hup' state to true, so
> >> > that
> >> > subplot(3,3,3) is never redrawn, then no subplots disappear at all.
> >> >
> >> > Do you see the same?
> >> >
> >> > --
> >> > mike
>
> Indeed with those additional information I was able to reproduce the bug
> which is clearly different than the one related to resize. Here the position of
> the axes is corrupted :
>
> octave:3> get (findobj (gcf, "type", "axes"), "position") ans = {
>   [1,1] =
>
>      7.5645e+12   3.2287e+22  -1.5054e+13  -6.4573e+22
>
>   [2,1] =
>
>      7.5645e+12   3.2287e+22  -1.5054e+13  -6.4573e+22
>  ...
>
> As a workaround, I found that avoiding the call to high level "plot"
> function (by changing the properties of the original line object) avoids the
> issue for me. Something like
>
> ...
>     subplot(3,3,3)
>     persitent h333 = [];
>     tt = t(1:round(L/15))*1000;
>     hh = h(1:round(L/15));
>     if (isempty (h333))
>       plot(tt, hh);
>       grid on;
>       %xlim([0 tplotmax]);
>       xlabel('time (ms)');
>       ylabel('Amplitude');
>       title('Impulse Response');
>     else
>       set (h333, "xdata", tt, "ydata", hh);
>     endif
> ...
>
> Anyway, Tony, can you file a bug report about this new bug so that the
> discussion can take place there?
>
> Pantxo
>
>
>
> --
> Sent from: http://octave.1599824.n4.nabble.com/Octave-General-
> f1599825.html
>

I have filed a bug report.

Tony


Reply | Threaded
Open this post in threaded view
|

Re: Subplot problem in Octave 4.4.0?

bpabbott
Administrator
> On May 4, 2018, at 7:15 PM, Richardson, Anthony <[hidden email]> wrote:
>
>> -----Original Message-----
>> From: Help-octave [mailto:help-octave-
>> bounces+richardson=[hidden email]] On Behalf Of Pantxo
>> Sent: Friday, May 4, 2018 3:24 PM
>> To: [hidden email]
>> Subject: RE: Subplot problem in Octave 4.4.0?
>>
>> Tony Richardson wrote
>>> I've tried a few other scripts in which the control panel is in
>>> different locations.  It doesn't appear to be due to interaction
>>> between a subplot and the control panel.
>>> I've found that if I comment out any code that draws to subplot(3, 3,
>>> 3) then the other subplots remain visible.  If I redraw subplot(3,3,3)
>>> the other subplots will eventually disappear.
>>>
>>> Tony
>>>
>>>> -----Original Message-----
>>>> From: Help-octave [mailto:help-octave-
>>>> bounces+richardson=
>>
>>> evansville.edu@
>>
>>> ] On Behalf Of Richardson,
>>>> Anthony
>>>> Sent: Friday, May 4, 2018 1:18 PM
>>>> To: Mike Miller &lt;
>>
>>> mtmiller@
>>
>>> &gt;
>>>> Cc:
>>
>>> help-octave@
>>
>>>> Subject: RE: Subplot problem in Octave 4.4.0?
>>>>
>>>> Yes, I see the same behavior.  That subplot is the only one above the
>>>> "control panel", so maybe there is some strange interaction between
>>>> the two (the control panel and the subplot).  I've got a few other
>>>> scripts that are behaving the same way under 4.4.0, I'll see if
>>>> something similar is going on with them.
>>>>
>>>> Tony
>>>>
>>>>> -----Original Message-----
>>>>> From: Mike Miller [mailto:
>>
>>> mtmiller@
>>
>>> ]
>>>>> Sent: Friday, May 4, 2018 11:56 AM
>>>>> To: Richardson, Anthony &lt;
>>
>>> ar63@
>>
>>> &gt;
>>>>> Cc:
>>
>>> help-octave@
>>
>>>>> Subject: Re: Subplot problem in Octave 4.4.0?
>>>>>
>>>>> On Fri, May 04, 2018 at 15:01:12 +0000, Richardson, Anthony wrote:
>>>>>> I just tested Octave 4.4.0 under Linux (Ubuntu/flatpak) and the
>>>>>> same bug is present - subplots disappear after adjusting the
>>>>>> sliders (not necessarily after the first adjustment, but it doesn't take
>> long).
>>>>>> Did you test under 4.4.0?  The maximize bug has been present for
>>>>>> a while,
>>>>> but the disappearing subplots (without window resizing) is new to
>>>> 4.4.0.
>>>>>
>>>>> If it helps narrow things down for you, I have found that it's the
>>>> 'hup'
>>>>> update that causes the subplots to disappear for me. For me the
>>>>> subplots disappear only when the third and fourth sliders are adjusted.
>>>>>
>>>>> If I comment out the lines that set the 'hup' state to true, so
>>>>> that
>>>>> subplot(3,3,3) is never redrawn, then no subplots disappear at all.
>>>>>
>>>>> Do you see the same?
>>>>>
>>>>> --
>>>>> mike
>>
>> Indeed with those additional information I was able to reproduce the bug
>> which is clearly different than the one related to resize. Here the position of
>> the axes is corrupted :
>>
>> octave:3> get (findobj (gcf, "type", "axes"), "position") ans = {
>>  [1,1] =
>>
>>     7.5645e+12   3.2287e+22  -1.5054e+13  -6.4573e+22
>>
>>  [2,1] =
>>
>>     7.5645e+12   3.2287e+22  -1.5054e+13  -6.4573e+22
>> ...
>>
>> As a workaround, I found that avoiding the call to high level "plot"
>> function (by changing the properties of the original line object) avoids the
>> issue for me. Something like
>>
>> ...
>>    subplot(3,3,3)
>>    persitent h333 = [];
>>    tt = t(1:round(L/15))*1000;
>>    hh = h(1:round(L/15));
>>    if (isempty (h333))
>>      plot(tt, hh);
>>      grid on;
>>      %xlim([0 tplotmax]);
>>      xlabel('time (ms)');
>>      ylabel('Amplitude');
>>      title('Impulse Response');
>>    else
>>      set (h333, "xdata", tt, "ydata", hh);
>>    endif
>> ...
>>
>> Anyway, Tony, can you file a bug report about this new bug so that the
>> discussion can take place there?
>>
>> Pantxo
>>
>>
>>
>> --
>> Sent from: http://octave.1599824.n4.nabble.com/Octave-General-
>> f1599825.html
>>
>
> I have filed a bug report.
>
> Tony
>
>

https://savannah.gnu.org/bugs/?53840