[erlang-questions] dialyzer and re(3)

[Kostis Sagonas]

Paul Guyot wrote:
> Le 11 nov. 08 à 15:27, erlang-questions-request@REDACTED a écrit :
> 
>> While we appreciate dialyzer very much and we rely on it,
>> we are somewhat disappointed by its current limitations

Can you please be a bit more specific about "its current limitations" ?
(perhaps even off the list if you think that's more appropriate).

While we definitely do not claim that Dialyzer is perfect (in fact, we 
have a long TODO list but very little time and/or support), it's not 
certain that our appreciation of its limitations and the prioritization 
of the issues fully agrees with those of its users...

>> horrible slowness, especially when compared to ocaml compiler
>> which does type inference and checking faster by a magnitude...
> 
> I only mentioned ocaml as an illustration that the job of infering  
> types and checking them can be much faster than what dialyzer  
> currently does, which is patent for anyone who worked with ocaml.  
> Tobias argued that the job is slightly different, and I trust him on  
> that matter.

Well, you probably misinterpreted Tobias.  The job is not just slightly 
different, it's *vastly* different.  For example, in ocaml there are 
type declarations, so the types are given by the programmer and each 
value can only be part of a single type: in other words, there is no 
subtyping and e.g. unions other than those explicitly declared by the 
programmer do not exist.  Also, recursive types are regular -- for 
example, all lists are properly nil-terminated.  There are many more 
differences.  Although different people have different preferences, it's 
generally hard to compare apples and oranges objectively...

Kostis