[plt-scheme] Proposal for s-expr tracing format

From: Matthias Felleisen (matthias at ccs.neu.edu)
Date: Mon Aug 13 22:00:37 EDT 2007

On Aug 13, 2007, at 9:53 PM, John Clements wrote:

>
> On Aug 13, 2007, at 11:51 AM, Shriram Krishnamurthi wrote:
>
>>> John's suggestion has a certain appeal -- [...]
>>
>> It's definitely a *cute* suggestion.  But it has appeal for no more
>> than a moment.
>>
>> To accept this "solution" would be to agree that continuations are so
>> abnormal that programmers should never be able to work conveniently
>> with them.  That is, we would be expressly penalizing them: first
>> creating the incentive of being able to walk the trace using
>> s-expression walkers, then perversely taking it away by making the
>> parens not match and the walkers therefore no longer work.
>>
>> I'm not sure I'd want to send out that message about continuations.
>> If anything, a TRACE done right could be a handy tool to
>> *understanding* continuations!  (Until the Stepper handles
>> continuations, that is, but I think that's slated for about 2025?)
>
> If you like the suggestion, you can credit me with it; all I  
> observed, though, was that the parens would be naturally unbalanced  
> in the presence of call/cc.  My initial thought, quite frankly, was  
> that this observation constituted support for the continued use of  
> vertical bars or other non-parenthesis indicators.

I second this. Though I myself have wanted to navigate the output of  
a trace with meta-f etc. But that was such a long time ago. I just  
don't ever need trace anymore. -- Matthias



Posted on the users mailing list.