From: Jesse van den Kieboom <troplosti@...>
Date: 2004-09-16T02:22:10+09:00
Subject: Re: Truly modular

Op wo 15-09-2004, om 18:19 schreef Robert Klemme:
> "Jesse van den Kieboom" <troplosti@orcaweb.cjb.net> schrieb im Newsbeitrag
> news:1095262552.10900.12.camel@ayla...
> > Hi,
> >
> > As Ruby claims to be very modular I'd like to do the following (I tried
> > some things but I think I'm going in the wrong directions).
> >
> > I have my main Ruby application. The applications need to be extendible.
> > My idea is to write extra .rb files and put them in one directory. The
> > main app should somehow load those files to extend functionality. on
> > runtime the files can changes and the main app should be able to reload
> > the extensions. Also new files can be added and these should be reloaded
> > too.
> >
> > I don't know the right approach to this problem. The external .rb files
> > somehow need to have access to parts of the main application and the
> > main application obviously needs to call functions from the external
> > files.
> >
> > Users can communicatie with the main app telling it there are new
> > external files to be loaded, the name of the files and maybe function
> > names to be called or whatever. The idea is that I don't want to restart
> > the main application but do want to extend it on the flow.
> >
> > Somebody with some great ideas how to achieve this?
> 
> Great?  Dunno...  As far as I can see here are two problems:
> 
>  i) file loading with respect to changes

This is not the main issue though. If I have to I can manually tell the
application to reload a certain file. Or use GnomeVFS for instance.

>  ii) communication between the code in the files and your app

> I can suggest something for i) (see below) but for ii) I'd have to know
> more about the nature of the app.

My application gets data through a TCPSocket. All the data for example
is prefixed by a certain command. I would like to have some sort of
modular support for these commands which means that I can dynamically
add a certain action for a new command. My guess was that this could
best be done by loading external files, let the files register some
command/actionfunction and have some sort of communication between the
files code and the main app code. The communication is the problem.


Jesse van den Kieboom