ATS resources

[H Zhang]

(This was first posted in another thread as a reply although I intended it to start a new thread.)

Hi,

I found Chris Double's blogs very helpful with real world usage examples of ATS:
http://bluishcoder.co.nz/tags/ats/

I am very interested in ATS but I am trying to figure out how exactly it fits with my needs. I think some high level guidelines on how a project can be done in ATS will really help newcomers and may even benefit the developers as it may force you to think more about how ATS will be used in real projects.

For my own preference for explorative coding I like languages that are expressive and that have REPL support. I currently use Prolog. ATS has an ML-like part that is expressive but does not support REPL.

That leaves the potential usage model of coding in ATS for performance critical libraries. Again if possible I still would like to code in the more expressive ML style, but the library code needs to work with other high level languages that have their own garbage collectors. Does that mean that ML style is out of question if I want to write an FLI extension for, say, Prolog? If ML style is still possible how does one go about boxing/unboxing data structures from ATS ML style to the target language and interacting with host GC? Or is the realistic choice to code in C style? I would love to see some worked out examples on interfacing ATS (ML style -- for C style presumable one can just follow the FLI of the targets) to Prolog or Scheme/Lisp, writing performance critical libraries in ATS and working with the FLIs of the target languages.

Sincerely,
Haitao

On Thursday, January 9, 2014 8:59:33 AM UTC-8, gmhwxi wrote:

Right now, a serious problem with ATS is that there is very little
on-line source of information for people interested in learning it. So it is
my hope that everyone learning ATS right now could also help his or her
fellow learners.

By the way, encoding proofs of theorems inside ATS is an excellent way to
gain understanding of the theorems. For instance, I had great difficulty with
Yoneda Lemma in category theory. Once I was able to implement it in ATS
a couple days ago, I felt that I had gained solid understanding of it. Actually,
I right now feel that Yoneda Lemma is so "trivial" :)

[Brandon Barker]

I just want to summarize what I've picked up on the list related to this question. ATS is a complex language, and trying to use all the features on a first pass is extremely difficult, even for expert ATS users. Starting with an ML style is perhaps a good first option, but you may want to make it more procedural (C-style) to get performance, and bring in linear types, ATS/LF, and dependent types to guarantee your code is working correctly later on. Of course, procedural style and linear types may be integrated at the same stage I imagine.

I think Hongwei had a tutorial posted on a mailing list a few months ago where he went from ML style, to procedural, to procedural + linear along these lines. There is probably more than one that has at least two of these steps. This doesn't touch on abstract types and templates which will also likely be important parts of projects; the former in particular seems to be important for program design.

As for a REPL interface, I'm also interested in having this, and would be happy to contribute what I can to making it happen. I'm wondering if using CINT as a backend would be viable, but I don't have enough experience to know how the GC might work in this situation.

Brandon Barker
brandon...@gmail.com

--
You received this message because you are subscribed to the Google Groups "ats-lang-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ats-lang-user...@googlegroups.com.
To post to this group, send email to ats-lan...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/ats-lang-users/c8d27f1b-874d-42ef-8fa7-172dcccc0e0b%40googlegroups.com.

[H Zhang]

Brandon,

Thanks for the comments. It would be nice if we can write ML style code that can be made binary compatible with foreign language data structures or have a nice way to model foreign language data structures even while using ML style. Also there is the question of working with the foreign language GC.

I started a separate thread on an REPL for ATS. From what I read about CINT I hear it is a rather hairy hack due to C semantics.

Haitao

[gmhwxi]

This all depends how Prolog creates/manipulates heap-allocated values.
If a library for Prolog can be implemented in ATS in ML-style, then it should already be possible
to do it in ML. So I have doubts about the viability of such an approach.

However, one possiblity is that you first implement in ML-style but use a separate heap to store
ATS values; this will create memory-leaks; then you employ  linear types to eliminate these leaks.

If you have a concrete case, I will be happy to take a look.