Re: LPeg pattern optimizer

[Julien Desgats <julien+lua@...>]

Le Sun, 01 Mar 2015 23:11:39 GMT,
Gavin Wraith <gavin@wra1th.plus.com> a écrit :

> In message <20150301222653.1608daaf@desgats.fr>
>           Julien <julien+lua@desgats.fr> wrote:
> 
> > Le Sun, 1 Mar 2015 14:56:30 -0500,
> > Daurnimator <quae@daurnimator.com> a écrit :
> >> 
> >> Nice work!
> 
> Seconded.
> 
> By the way, in part-2-ast-transformation, I think you meant to write
> "traversing" or maybe "translating", as "traducing" in English
> nowadays means "betraying" - 'trahison' n'est pas 'traduction'. It is
> an interesting etymological nexus between French and English :)
> 

I didn't knew about that, fixed (along with some other
typos), thank you :)

> The problem with the "Algebra of Regular Expressions" is that though
> it is finitely generated it has not been proved to be finitely
> presented 
> - i.e there is probably no finite set of identitiess from which all
> identities may be deduced. Though in your case it is the algebra of
> PEGs rather than of regular expressions, and reductions, rather than 
> identities, that are the pertinent objects of study.
> 

Actually, I have to admit that I did this work using TDD and benchmarks
rather than formal demonstration (which is a field that I don't know
well).