Re: Converting xconq to C++

[Alan Schmitt <alan dot schmitt at inria dot fr>]

Hi,

What I'd really like to understand, if xconq is ported to another
language, is what we gain from it. If xoncq fits nicely in an object
oriented programming style, why not. But choosing which object language
to use depends on 3 things I can think of: efficiency (both of the
resulting code and of the coding process), portability, and extra
features (what librairies are available, what additional language
constructs may be used).

Let me give you an example about a not so well known programming
language that could be used: objective caml (http://caml.inria.fr). In
this language you can write object oriented, functional and imperative
code. It is ported to many systems, has many libraries (if we care about
graphics, it's possible to write tk or gtk code), a nice module system,
type inference, and many other things. It also produce fairly efficient
code (not as efficient as C, but really close ... you can look at Doug's
language shootout page, if you care:
http://www.bagley.org/~doug/shootout/craps.shtml). Anyway, my point is:
why not rewrite xconq (it seems to be quite an undertaking, but why
not), why not do it in a different programming style, but why c++ ?

Alan

* Ed . (epo001@hotmail.com) wrote:
> Hello everyone,
> 
> I started an identical thread a while back and it came to no sensible 
> conclusion. Here it is again, I find myself with time on my hands having got 
> laid off recently. I'm looking to refresh my C++ and it seems that 
> converting xconq to C++ would be an ideal project.
> 
> Equally ideal would be to have the conversion adopted by the main xconq 
> gurus (Stan, Hans and Jim). So a first question is this: is there interest 
> in having xconq ported to c++ (I'd using gcc3 under redhat 7.2). If so I 
> could liase offline with the principals to resolve design issues.
> 
> The code seems pretty well structured so the conversion should be reasonably 
> straight forward.
> 
> The rough plan is as follows:
> 
> Using mixed C/C++ programming techniques incrementally convert the C sources 
> to C++, testing as we go until a literal C++ translation of the C sources is 
> attained making explicit any object-ness of the present code. The intention 
> is to translate code not design, hence many fine details may need 
> refinement.
> 
> Release this to the developers when it is suitable for group evolution.
> 
> Use this as a code base to improve the translation object-by-object until 
> the codebase becomes fully object oriented.
> 
> Obviously details need to be sorted.
> 
> Now I am happy to embark on this, the issue is whether anyone else is 
> interested in the outcome, if so there needs to be some liasion to ensure 
> changes made the  existing code are reflected in the translation.
> 
> Please reply to the list for now and I will follow up.
> 
> Thoughts, suggestions?
> 
> Ed
> 
> _________________________________________________________________
> Send and receive Hotmail on your mobile device: http://mobile.msn.com


--
The hacker: someone who figured things out and made something cool happen.