Re: converting the sockets and instrument-control packages to mercurial

[Olaf Till]

On Fri, Jul 12, 2013 at 10:57:46AM +0200, Stefan Mahr wrote:
> > I'm converting the instrument-control package to a mercurial
> > repository. I'm thinking of keeping the instrument-control and sockets
> > packages in the same repository. The reason behind this is that the
> > functionality of the sockets package is part of the instrument control
> > toolbox in matlab, so the two packages should be merged anyway. In
> > keeping the history of both together now, means it will be easier to
> > conserve their history when such merge is performed.
> > 
> > Does this decision seems reasonable to everybody else?
> 
> The sockets package is a straight interface to raw sockets. AFAIK Matlab has 
> no direct interface to raw sockets. It makes not much sense to me to merge 
> something from sockets to instrument-control, since instrument-control needs 
> some higher level extensions to be compatible to matlab.

Just an additional note, without knowing whether it is useful in the
decision:

It could be sensible for a sockets package to associate socket
descriptors with Octaves oct_iostream descriptors, to be able to use
them from the interpreter with Octaves existing functions
(e.g. Octaves fprintf). AFAIK the current sockets package doesn't do
this. The parallel package does this association for its sockets, but
not with the intention to make them generally usable from the
interpreter.

Olaf

-- 
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net

signature.asc
Description: Digital signature