[ previous ] [ next ] [ threads ]
 
 From:  Peter Curran <peter at closeconsultants dot com>
 To:  m0n0wall dash dev at lists dot m0n0 dot ch
 Subject:  Re: [m0n0wall-dev] Re: [m0n0wall] Survey results
 Date:  Tue, 1 Nov 2005 19:23:18 +0000
On Tuesday 01 November 2005 18:40, Joe Nardone wrote:
> > There's a D-compiler for FreeBSD in the ports (see [4]). Has anyone
> > experiences in using D for projects?
>
> I'm going to interject here from a pragmatist's point of view. Pick
> something commonplace. Something that has popular support, general industry
> trust, and a wide variety of tools. This means, no D, no Erlang, etc. etc.
>
> Think very hard about the implications of choosing Java -- the fact that
> you need a runtime for whatever platforms you want to run on is a definite
> limiting factor. It still may be the best choice (I don't think so) but
> know the implications. It also has definite impacts on the memory
> footprint, which the survey indicates could be a problem.
>
> I'd still say C++ for the back-end server. There is a higher risk here than
> with something like Perl or Python but on the other hand there is a LOT of
> C++ code out there begging to be re-used, both within other projects and
> free-standings libs (e.g. Boost).
>
> Joe

Can we explore what we expect this back-end to do?

I have followed the arguments for and against the various options, but I have a strong feeling that
we are not going to get anywhere until we can decide the requirement in more detail.  Chris Buechler
posted a requirements list for the OS - what about something similar for the back-end system.

For example, we had a discussion about OO a few days ago.  Lots of people for and against, but
nobody actually said why it would be a good idea from a functionality viewpoint.

If we are just talking about OO because its cool then I suggest that is not a good start!

My reason for suggesting OO is that with a very well designed (and documented) pattern it should be
pretty simple to add functionality without breaking what is there already.  From my own (pretty
limited) exposure to OO I know that a) it works; b) it is very difficult to do properly without an
exceptionally good design model.

Is it worth the effort?  Can we collectively generate the aforementioned design?

If the concensus is that this cannot be done, then the choice of an OO system is pretty pointless. 
A poorly designed OO environment is a barrier to progress.

Peter


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.