Re: Octave Gui Bug (Maybe)

[Daniel J Sebald]

On 06/05/2013 03:09 PM, John Donoghue wrote:
[snip]
>     >  On 06/05/2013 12:50 PM, Patrick Martin wrote:
>     >
>     > > Hello,
>     > >
>     > > I pulled down the most recent source for Octave and it built
>     > > successfully. When I run octave, it correctly pulls up the gui
>     and the
>     > > command line works fine. However, the buttons for opening a
>     script file,
>     > > or calling "edit somthing.m" nothing happens. Did I miss a library
>     > > during compile time? Should I grab a different branch to compile?
[snip]

> Out of interest sake, if you click on the "new" file button on the
> editor, does it display a new file window? I am wondering whether
> QScintilla was enabled?
> John

Good point.  Alternately, try something like this in the build directory:

[sebald@ build1]$ grep QSCINTILLA config.h
#define HAVE_QSCINTILLA 1


In any case, I've looked into this further and the way that things are 
set up is sort of a bug.  It may well be that Patrick's system doesn't 
have QScintilla, but Octave GUI should fall back on the editor that is 
shown in

> > edit get all
ans =

  scalar structure containing the fields:

    EDITOR = emacs %s
    HOME = /home/sebald/octave
    AUTHOR = Daniel J Sebald
    EMAIL = [](0x0)
    LICENSE = GPL
    MODE = async
    EDITINPLACE =  1

if that is true.  However, I know it isn't doing that.  Here is the hunk 
of code for opening the file via "status = __octave_link_edit_file__ 
(file);"

bool
octave_qt_link::do_edit_file (const std::string& file)
{
  emit edit_file_signal (QString::fromStdString (file));

  return true;
}

It is returning true in all cases, even if QScintilla editor isn't 
present.  The minimal fix should be something like:

bool
octave_qt_link::do_edit_file (const std::string& file)
{
#ifdef HAVE_QSCINTILLA
  emit edit_file_signal (QString::fromStdString (file));
  return true;
#else
  return false;
#endif
}

But even that isn't proper in some respects.  The way this should work 
is that the routine should emit the signal and then go into a wait state 
until it gets a wakeup and info about the success of the opening of the 
file from some other signal.  If QScintilla isn't present that editor 
slot should be a rudimentary thing that simply sets the result to false 
then wakes the worker (Octave) process.

Torsten, if you are following, would you like to look into this? 
Otherwise I will a changeset this weekend.

Dan