[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: QT Octave installation

From: Jordi Gutiérrez Hermoso
Subject: Re: QT Octave installation
Date: Tue, 14 Feb 2012 13:05:28 -0500

2012/2/14 Judd Storrs <address@hidden>:
> 2012/2/14 Jordi Gutiérrez Hermoso <address@hidden>:
>> However, the flaw runs deeper: it doesn't work with fltk plotting. Try
>> it: "graphics_toolkit fltk;" now try to plot something. QtOctave will
>> hang. Same with GUI Octave, same with Xoctave (and Octave workshop,
>> and YAOG, and all those other historical efforts). We need a tighter
>> coupling between Octave and its GUI, but no GUI author has wanted to
>> do that. They all think of the problem too simply. They think of
>> Octave as some foreign program and that communication is simply text
>> goes in, text comes out. Octave needs more interaction that, for
>> example, for being able to fully reproduce the interactive plots that
>> Matlab users come to expect.
> Does the fltk backend allow plots to be interacted with while the
> interpreter is busy with calculations? Perhaps the design flaw with
> fltk windows is elsewhere.

You're right, this is also a problem, which we can hopefully fix when
we move completely away from fltk (it was always meant to be a
temporary GUI toolkit, but we've kept it around for too long).

But plots still need a lot more interaction than a pipe can offer. You
should be able to modify objec4ts and variables from the plot window.
This is a semi-frequent request of GUI users. The plot window should
really share memory with Octave, especially if large amounts of data
is being exchanged, same as the GUI. Passing large amounts of data
through a pipe is slow and brittle. Not to mention that it's not
uncommon to exhaust memory when using Octave, so that memory shouldn't
be copied to two different programs.

I still think that the GUI in Savannah is doing the right thing.

- Jordi G. H.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]