[erlang-questions] dialyzer working with lfe, joxa, et. al?

[Robert Virding]

As I see it there are two separate parts to making dialyzer be able to handle other languages:

1. Modify dialyzer to be able to handle language specific plugins.
2. Write the language specific plugins.

It is clearly not the dialyzer implementor's job to do 2., that should be left to the various language implementers. But 1. need only be done once if it done properly and generically. It is 1. which I mean would be better if the dialyzer implementers did it as it only needs doing once and they would do it much faster and better than anyone else.

That was all.

Robert

----- Original Message -----
> From: "Kostis Sagonas" <kostis@REDACTED>
> To: "Robert Virding" <robert.virding@REDACTED>
> Cc: "erlang-questions" <erlang-questions@REDACTED>
> Sent: Sunday, 10 March, 2013 11:17:45 AM
> Subject: Re: [erlang-questions] dialyzer working with lfe, joxa, et. al?
> 
> On 03/10/2013 01:57 AM, Robert Virding wrote:
> >> From: "Kostis Sagonas"<kostis@REDACTED>
> >>
> >> On 03/07/2013 06:48 PM, Robert Virding wrote:
> >>> Keep them crossed. :-) It's definitely not impossible, it "just"
> >>> needs the dialyzer people to implement hooks to which we (LFE,
> >>> Joxa, ...) can write plugins. It that were to happen I might even
> >>> add type specifications to LFE.
> >>
> >> Even disregarding for the moment the question why would "the
> >> dialyzer
> >> people" care about lfe, joxa, et al., (dialyzer: Discrepancy
> >> Analyzer
> >> of
> >> *Erlang* code) if you are not more specific than the above wish
> >> how
> >> exactly do you expect them to know about the details of possible
> >> hooks
> >> that are needed by non-standard uses of dialyzer by other more
> >> exotic
> >> languages out there?
> >
> > Well, if you want to be really strict then dialyzer works on *Core
> > erlang* code. This is what dialyzer means by Erlang.
> 
> No, it's not. Just because dialyzer does its analysis on Core Erlang,
> this does not mean that it was ever designed to analyze languages
> other
> than Erlang.
> 
> I grant you that dialyzer's analysis is in principle applicable to
> other
> languages that compile to Core Erlang, but I think that my previous
> mail
> makes it clear that I also think that the responsibility for doing
> any
> changes that are required lies with the developers of languages who
> are
> interested in having such a tool for their language.
> 
> It's probably irrational to expect that non-users of a language care
> about the tools that are available for it more than the community of
> that language, isn't it?
> 
> Kostis
>