On 10/5/05, Bill Clementson billclem@gmail.com wrote:
Hi Svein,
Due to a few requests lately, I thought I should mention something: This code is obsolete.
It only ever really worked on SBCL, and even there not perfectly. I'm rewriting it as something that should work everywhere, but meanwhile any non-fatal bugs are going to go unfixed - sorry about that."
Any progress in getting the SBCL stepper working yet?
It always pretty much worked in SBCL, so I assume you're referring to my new implementation. Progress? Some.
The basic design is a Lisp evaluator that communicates with the stepper UI by way of the condition system - pretty much the same setup as my last attempt, except that I'm writing said evaluator myself. I'm about halfway done with the evaluator, and haven't really tried doing anything about the UI yet; Slime's debugger works fairly well for the development bit.
*IF* I had nothing else to do, I could probably get a workable stepper (sans UI) in a week. As I do, in fact, have other things to do, I'm aiming to complete it by christmas. Partials will be unpredictable, but feel free to pull patches from darcs://brage.info/~svein/repos/slime/, which will contain semi-stable code once I've got such a thing. (At the moment it's practically empty. You could still use it as a CVS gateway, I suppose.)
I'll also announce it to slime-dev once I feel confident about letting the ravening hordes at my code. As has been mentioned, all good things come to those who wait - for a tidbit, I've got a framework for step-backwards in place.
Also, what were the problems that you had getting it to work with OpenMCL?
(defmacro step (&body body) ,@body)
That about summarizes it. I'd been using that macro as a starting point for figuring out how the implementations did it. OpenMCL apparently didn't, but maybe I didn't dig deep enough.
I'd appreciate pointers to whitepapers on Lisp evaluation and such, if you have some handy.
- Svein Ove Aas