Re: Howto make EEGLab compatible...

[John W. Eaton]

On 29-Nov-2006, Andreas Romeyke wrote:

| John W. Eaton wrote:
| >
| > OK, you are certainly encouraged to run that and report problems you
| > find.  Submitting fixes is even better.
| The most problems are related to EEGLab, sure. Because some of the
| Octave-developers deny to be more compatible with Matlab I am not able
| to decide what is a real problem by Octave or a compatibility issue.

Compatibility issues are often treated as bugs.

| Only if the result of specific mathematical, isolated operations will
| be wrong I had a chance to commit a bugreport, right?

You can submit a bug report for whatever doesn't seem to work
correctly.

| > | The main problems are related to graphical user interfaces and the
| > | rotten syntax checking by Matlab.
| >
| > What do you mean by "rotten syntax checking by Matlab"?
| 
| Matlab interpretes the m-files and checks them only in interpreting
| state. Octave checks the m-files first, checks the called functions
| within and then it will start  the  interpretation-process.

Octave does not parse all called functions before starting to
interpret.  It only parses functions as needed.  Try the following:

  f.m:
  function f ()
  if (nargin == 1)
    g
  endif

  g.m:
  function g ()
    disp ("foo-i-hithere")

  f
  who
  f (1)
  who

You shouldn't see g in the output of who until it is called.

| > | But in reality
| > | it is only possible to support primary Matlab and secondary Octave and
| > | therefore we need more or less compatibility between them.
| >
| > Why is that?
| Because most developers are only familiar with Matlab-Syntax.

I think the syntax is essentially the same.

| also work on universities and are mostly not affected with license
| fees so there is not enough pressure to switch over to Octave. The
| migration of EEGLab from Matlab to Octave is a long term process and a
| lot more of compatibility will make things easier.

OK.

| > | To avoid unnecessary work we are very interested in your comments and
| > | thoughts about this  work and your  roadmap to Octave, especially in
| > | supporting GUI-funtionality.
| >
| > So are you saying "write a Matlab-compatible GUI for us and we will be
| > happy"?
| 
| It seems the easiest way to me, is not it?

Yes, well, for you...

| But I am very happy if any(!) GUI-functionality will be added soon. So
| we are able to build wrappers to use GUI with Octave, too.

I think you would want the GUI features to be mostly compatible.  Of
course, someone has to do the work.

jwe