Bug#467563: camlp5 strict mode?

[Stéphane Glondu <glondu@debian.org>]

Le 01/03/2012 13:51, Hendrik Tews a écrit :
>    There are other reverse build-dependencies: matita, ssreflect,
>    geneweb... and maybe others (transitively). Someone has to test.
> 
> ulex0.8 does not build with a strict camlp5:
> 
>     ocamlc -a -o pa_ulex.cma -pp 'camlp5o pa_extend.cmo q_MLast.cmo' -I +camlp5 cset.ml ulex.mli ulex.ml pa_ulex.ml
>     File "pa_ulex.ml", line 152, characters 17-22:
>     Error: This expression has type (MLast.patt * 'a option * MLast.expr) list
>            but an expression was expected of type
>              (MLast.patt * MLast.expr option Ploc.vala * MLast.expr) list
> 
> 
> matita depends on ulex0.8. Does this fact already rule out the
> option of supporting strict camlp5 only? 

I would greatly prefer "fixing" whatever doesn't work with camlp5
strict. AFAIU, strict mode is superior and the only benefit of
transitional mode is backward compatibility. However, I guess we can
keep both if it's not too much burden.

Concerning ulex0.8/matita, a perhaps saner alternative is to make matita
work with latest ulex and get rid of ulex0.8. I guess this means porting
it to camlp4... I've got no idea how much work is that, though.

> ledit, geneweb and coq build without problems with strict camlp5.

Did you test with the latest version (6.04)? Thanks a lot for your great
work.


Cheers,

-- 
Stéphane