On Wed, Jun 08, 2011 at 02:00:03PM +0200, Ariane van der Steldt wrote: > On Wed, Jun 08, 2011 at 02:24:14AM +0200, Thomas de Grivel wrote: > > Just like C, OCaml has skills in its semantics, but they both suck at > > grammar : parsing ml requires a full lex/yacc above and below gcc. Not > > really checked code but if the bug hasnt shown it must not exist. Or in the > > next release. > > Ah, OCaml is cool because you have not seen a bug in a compiler...
There are lots of good points and bad points in Ocaml. One thing you won't find is a bug in its Garbage-Collector (that was Damien Doligez's thesis, and he had to fix all the bugs in it to finish the proof that it was bug-free). As far as I know, similar work has not been done in the remaining parts of OCaml. So there are definitely bugs in it. Just because no-one hasn't found them doesn't mean they don't exist (and there are few system hackers as brilliant as Damien). There are less cool parts of ocaml development, like feuds between various sets of developers, but I won't go there, it's just the nitty gritty of a real life language, thus very much imperfect. Of course, ocaml type system sucks. But so do all type systems, since type-inference is a semi-decidable problem, and all "smart" languages out there are making-do by using a "suitable" approximation of the best matching type for a given construct... Well, if de Grivel likes ocaml so much, why doesn't he improve our port ? like figure out why it doesn't run on hppa (at a wild guess, ocaml assumes things about stack orientation...), or to polish the compiler so that it actually compiles on, say, arm, for instance...
