Opinion on edit.m and doc.m

[John Swensen]

I have been working a little more on OctaveDE lately and trying to add 
functionality.  One of the features I have been looking at lately is allowing 
the user to open a file from the command line using the 'edit' command.  
However, it seems from looking at edit.m that it is coded to call system() to 
open the editor of choice.  However, for OctaveDE the IDE and Octave are 
actually in the same process and it doesn't seem to make sense to have to make 
a call to system() to accomplish the behavior.  I suppose I could either create 
something akin to emacsclient over sockets/IPC or write something to a file and 
have OctaveDE be checking for a change in that specific file.

A similar situation exists for the 'doc' command.  OctaveDE has a WebKit based 
documentation browser with full text search capabilities.  It would be nice to 
be able to open a specific function in the documentation browser with the 'doc' 
command, but 'doc' currently works only with the GNU Info browser.

As of right now, all the interaction between OctaveDE and Octave is done 
through a class I call octave_server.  This class uses the readline idle loop 
to process these interactions in a thread-safe manner.  I would prefer to make 
changes to doc.m and edit.m to allow the interaction between OctaveDE and 
Octave to be consistent and wholly located through this interface.  But, before 
I start working on a patchset I wanted to see if anyone has a preferences, 
suggestions, or alternative that I haven't thought of.

John Swensen