Re: Help with Octave standalone

[Ian McCallion]

On Sun, 15 Mar 2020 at 14:53, "Markus Mützel" <address@hidden> wrote:
>
> Am 15. März 2020 um 15:28 Uhr schrieb "Ian McCallion":
> > On Sun, 15 Mar 2020 at 12:36, "Markus Mützel" <address@hidden> wrote:
> > >
> > > Am 15. März 2020 um 13:00 Uhr schrieb "Ian McCallion":
> > > > On Fri, 13 Mar 2020 at 16:37, Ian McCallion <address@hidden> wrote:
> > > >
> > > > I am having great difficulty with this and I think there is a bug in
> > > > Octave 5.2 although I'd be happy to be proved wrong.
> > > >
> > > > The script below demonstrates the problem:
> > > > ----------------------------
> > > >  graphics_toolkit('qt');
> > > > myfont={'fontsize',25,'fontweight','bold'};
> > > > mygraphs={'-dpng','-S800,800'};
> > > > mywidegraphs={'-dpng','-S1800,800'};
> > > > set(0, 'defaulttextfontsize', 15)
> > > > figure(1, 'visible', 'off')
> > > >         xrange = 1:10;
> > > > ssd=xrange.+rand(1,10);
> > > > plot(xrange, [ssd; mean(ssd)*ones(1,10)], 'Linewidth',2);
> > > > legend('Actual','Average over touch')
> > > > axis([0, xrange(end)+2, min(ssd)-5, 5+max(10,max(ssd))])
> > > > xlabel('Row number', myfont{:})
> > > > ylabel('Milliseconds', myfont{:})
> > > > title('Standard Deviation (Touch only)', myfont{:})
> > > > grid ('on')
> > > > print (1, [graphics_toolkit, '1.png'], mywidegraphs{:})
> > > > ------------------------
> > > > It runs fine when launching Octave from a commandline using the command
> > > > - c:\octave\octave-5.2.0\octave.vbs
> > > > - C:\octave\octave-5.2.0\octave.vbs --no-gui
> > > > - C:\octave\octave-5.2.0\mingw64\bin\octave.bat --gui
> > > > It fails when when launching Octave from a commandline using the command
> > > > - C:\octave\octave-5.2.0\mingw64\bin\octave.bat --no-gui
> > > > and the resulting output is given below.
> > > > --------------------------
> > > > octave:6> plottest
> > > >    *** Building graphs
> > > > error: graphics_toolkit: qt toolkit is not available
> > > > error: called from
> > > >     graphics_toolkit at line 81 column 5
> > > >     plottest at line 2 column 3
> > > > octave:6>
> > > > --------------------------------
> > > >
> > > > The common property of the situations in which it works is that a new
> > > > window is launched for all of them, and in the case where it does not
> > > > work Octave "borrows" the commandline window it was launched from.
> > > >
> > > > Unfortunately for me this is a show-stopper for use of 5.2.0 so I will
> > > > be happy to run other tests as needed to help track down the problem.
> > > >
> > > > Cheers... Ian
> > >
> > > "octave.bat --no-gui" runs octave-cli.exe which isn't linked against Qt. 
> > > So you won't be able to use "graphics_toolkit ('qt')" with it.
> >
> > Hi Markus, Thanks for the clear definitive answer.
> >
> > > Use one of the other ways you've shown to start Octave to use 
> > > "graphics_toolkit ('qt')".
> >
> > Unfortunately, as the other ways produce a second window when launched
> > from and existing commandline window they are not a solution for me
> > because my user workflow runs in a windows commandline window, and the
> > user needs to see the output of his commands (some of which use
> > Octave) in the window.
> >
> > I switched to qt from gnuplot as recommended a few months ago on this
> > list because in 5.2.0, gnuplot and fltk both produce warning messages
> > as shown in the following output:
> > --------------------------
> > octave:4> plottest
> > Plot a graph using gnuplot
> > DEBUG: FC_WEIGHT didn't match
> > Elapsed time is 1.02322 seconds.
> > Graph plotted
> > octave:5> plottest
> > Plot a graph using fltk
> > DEBUG: FC_WEIGHT didn't match
> > Elapsed time is 1.0631 seconds.
> > Graph plotted
> > --------------------------
> >
> > Given the choice between staying on 4.4.1 and using Qt vs moving to
> > 5.2.0 and getting spurious DEBUG: messages I will go with the second
> > choice and tell users to ignore the messages.
> >
> > Will octave-cli.exe at some point be linked against Qt?
>
> I don't think this will happen because it is basically the main difference 
> between octave-cli and octave-gui.
>
> Markus

OK. Thanks again. So the final follow-up question is, is there work
going on to avoid emitting the spurious DEBUG message?

Cheers... Ian