On May 28, 2013, at 7:00 AM, Piotr Klibert wrote: > Hello, > > I like an idea of setting breakpoint and when it is reached during runtime > dropping into the interactive shell. It's done in Python with pdb and in one > of the Schemes (Gambit?). I tried, but I couldn't make anything similar for > Racket. > > At first I looked at namespaces and (read-eval-print-loop), but that didn't > work. I could "load" the top-level module namespace, but I couldn't hook into > the namespace inside the function (if it even exists as a namespace). > > Then I tried to look inside the GUI debugger, and I realized that it uses a > macro which inserts some function call before and after every expression. The > debugger code was too hard for me to comprehend fully and I couldn't make it > work the way I wanted, too. > > It was some time ago, so sorry for the lack of specifics, but my question is: > is it possible, and how, to do something like this in Racket: > > (define (fun) > (define a 9) > (define b '(a b c)) > (let > ([d display]) > (set-trace)) > (displayln "We're here!")) > > (fun) ;; starts repl, where > > > a > 9 > >(d b) > (a b c)> (continue) > We're here!
Yep, you can definitely do this. To show this, I've just built a simple proof-of-concept command-line debugger; it's barely functional (NPI), but there shouldn't be any huge conceptual hurdles (and I don't have time to spend more time on it right now (unless I get a lot of interest :))). I've pushed "command-line-debugger-example.rkt" in the stepper collection to the plt repo. As an alternative to compiling the whole tree (or waiting for a nightly build), I've also created a gist: https://gist.github.com/jbclements/5671991 You can copy this file into the "collects/stepper" directory. Here's an example of using it (this example also appears in the file itself): Put the following code in /tmp/foo.rkt: #lang racket (define (f x) (+ x 14)) (f 9) Then, you might run it from the command line like this: jclements-09740:~/plt/collects/stepper clements> racket Welcome to Racket v5.3.4.7. > (require "command-line-debugger-example.rkt") > (run-program "/tmp/foo.rkt") > *ding* a new step has arrived: '#s(step #f expr-finished-break ((#<procedure:...ate/annotate.rkt:1061:62> #f #<procedure>))) (srcloc) step-srcloc: no source contexts in this step context...: /Users/clements/plt/collects/racket/private/misc.rkt:87:7 > (continue) > *ding* a new step has arrived: '#s(step (#<procedure> #<procedure> #<procedure>) normal-break #f) (srcloc) #<syntax:5:1 f> > (continue) > *ding* a new step has arrived: '#s(step (#<procedure> #<procedure> #<procedure>) result-value-break (#<procedure>)) (srcloc) #<syntax:5:1 f> > (continue) > *ding* a new step has arrived: '#s(step (#<procedure> #<procedure>) normal-break #f) (srcloc) #<syntax:5:0 (#%app f (quote 9))> > (exit) Note that while I spent the five minutes required to extract some semblance of source locations, I have not spent the additional five minutes required to expose variable bindings… I'd be happy to do this, if people are actually interested in it. Best, John ____________________ Racket Users list: http://lists.racket-lang.org/users
