[plt-dev] an api for check syntax & module browser

From: Robby Findler (robby at eecs.northwestern.edu)
Date: Thu Sep 10 09:55:33 EDT 2009

That sounds like a great thing (I know you've mentioned it to me
privately before, but it was nice to get a little more detail now)!

That also suggests an indirect use of the check syntax refactoring I
had in mind, namely building a test suite that checks to see that they
behave the same way.

Robby

On Thu, Sep 10, 2009 at 2:43 AM, Philippe Meunier <meunier at ccs.neu.edu> wrote:
> Ryan Culpepper wrote:
>> I envision two APIs. One handles source code analysis, possibly
>> including expansion, builds def-use relations, etc. The other handles
>> annotations to code in editors, allows adding mouseover callbacks to
>> regions of code, adding right-click menu items, etc.
>
> FYI, I'm working on an abstract interpretation framework for DrScheme
> that will eventually provide APIs for that stuff, including the GUI
> part (in fact the GUI part of MrFlow already had a basic version of an
> API that provided the mouseover callbacks (to draw flow arrows) and
> the right-click menu items, it's just that MrFlow was that API's only
> customer and that the API had only minimal documentation).
>
> Most of (all?) the "client programs" you list would then be plug-in
> analyses for the framework.  In turn there's fairly simple DrScheme
> tool that takes the framework and all its analyses and plugs the whole
> thing into DrScheme (in theory the framework can be plugged into any
> program you want, but I haven't actually tried it yet).  In parallel
> to working on the framework proper, I'm also developing two client
> analyses, mostly to check that I'm abstracting the framework at the
> right level, and, interestingly, one of them is an attempt at
> re-implementing Robby's syntax checker (the other client is a tracing
> evaluator, and there's in fact a third client I'm working on, which
> one day should become MrFlow 2.0, but that one is harder to develop
> than the other two).
>
> Don't hold your breath though, I've got about 2000 lines of code so
> far and I'm only just about to be able to analyze the lambda calculus,
> so it'll be quite a while before I have something that can more or
> less match Robby's tool.  Progress is steady but slow, there's just a
> ton of infrastructure code to write (anybody wants to write a
> hashconsing library that can handle cyclic data structures, unions,
> subtyping (and least upper bounds in general), and that's reasonably
> efficient and GC-friendly?)
>
> Philippe
>
>
> _________________________________________________
>  For list-related administrative tasks:
>  http://list.cs.brown.edu/mailman/listinfo/plt-dev
>


Posted on the dev mailing list.