Re: Editing files, indexing operations...

[Paul Kienzle]

James Frye wrote:

> On Mon, 23 Dec 2002, John W. Eaton wrote:
>
>  
>
> > Why not just fix things so that Octave is compatible with Matlab?
> >    
>
> Well, that would certainly be my preference :-)  Then the programs I have,
> which do work in Matlab, would also work in octave without me having to
> fiddle around with them at all, and I could get back to what I'm supposed
> to be doing.  And if you could have it done sometime in the next few days,
> please :-)
>
> However, the message I got was that octave wouldn't allow a whole program
> (or multiple functions) to be packaged in one file, with the implication
> that this WAS a case of octave doing something better than Matlab. 
That certainly wasn't my intent.

> Which
> would seem to leave me SOL as far as using octave: I could introduce minor
> fixes such as the end in array index stuff without anyone complaining, but
> if I did something as major as breaking all the scripts into multiple
> files, the people around here would like me a lot less than they
> do now :-)
If you automate the process, you only have to break the scripts into 
multiple files for the purpose of
running octave.  Put it in a Makefile, and have it spit the split files 
out to a separate directory which is
in the octave path but not the matlab path.  You have one extra step 
after saving the file, which isn't
so bad in the short term.  This script should be easy enough for someone 
familiar with perl.  Any
volunteers?

While you are at it, you can collect all the functions in the .private 
subdirectory, tag them with the
directory name of the parent, copy them to your octave path, and modify 
all the files in octave path
so that any identifier which matches the name of a file in the .private 
directory is replaced with a
tagged version of the identifier.  Even if the identifier is a variable 
instead of a function name, the
tagged version will still work.

There is some special pragmas for marking functions as commands in the 
latest CVS. Some
matlab code depends on this.  Running a special script in the directory 
creates a PKGADD
file which gets executed when the directory is first added to your path. 
I don't know what
happens if you change the contents of PKGADD in the middle of the 
session.  I suppose
you would have to reexecute it with something like 
source(fille_in_loadpath('package/PKGADD')).

Long term we should have a way to handle functions which are private to 
a file and those that are
private to a package.  

I'm not so keen on multiple levels of nesting --- pascal had it, but it 
never did much for me.  Of
course, that was first year and projects were smaller, but even now I 
get along just fine with
functions that are private to a file in C.  I also don't see a 
convenient syntax for multiple levels
of nesting.  The current matlab syntax doesn't separate the multiple 
functions in the file with
end statements, so if octave had multi-level nesting only the first sub 
function would be visible
to the main function.  I don't know the semantics of scope within 
matlab, but I'm guessing
that you can call any function within a single m-file from within any 
other function defined in
the same file.

Paul Kienzle
address@hidden




-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------