Re: [Texmacs-dev] Current work, nogencc and tree in generic programming

[Joris van der Hoeven]

On Sat, 27 Apr 2002, David Allouche wrote:

> While you were away, I ported TeXmacs to plain C++. See:
> http://ddaa10.free.fr/texmacs/patch/1.0.0.1
> The release 1.0.0.1-nogencc-0.5 is functionnaly equivalent to the official 
> 1.0.0.1 release, but the compilation and installation system is broken: 
> currently it only allows MIXED compilation and is only tested on 
> i386-gnu-linux.

I am not fundamentally opposed to this idea, but you will have to test
this no-gencc version of TeXmacs on all architectures which work at
this moment first. When I have time, I will also check whether there
are no hidden performance losses (or gains; either in time or space).

> By the way, I do not really see the use of the DYNAMIC compilation style. And 
> I think that, by default, TeXmacs should be compiled with -O2 optimization 
> since that brings a good 30% speed-up.

I am not sure; we have to make detailed performance compairisons.

> I am currently working on the ttree structure. I am porting most of the code 
> of the tree class into a new base template class basic_tree<tree_traits>. 
> This additional genericity should help reduce the amount of code duplication 
> in tree, ttree and rewriter for no runtime cost.

Hmmm, we'll see; remember that we might replace tree -> ttree
at the end anyway.