[plt-dev] syntax-case error messages

But Matthew, that is not the choice! The choice is between a car of
null error and the "f or something that called it or something that it
called" failed error.  There is no spec so there is no blame!

Robby

On Monday, March 29, 2010, Matthew Flatt <mflatt at cs.utah.edu> wrote:
> At Mon, 29 Mar 2010 18:58:34 -0400, Matthias Felleisen wrote:
>>
>> On Mar 29, 2010, at 6:36 PM, Robby Findler wrote:
>>
>> > By thisine of reasoning a scheme imlementation that didn't have
>> > contracts should change car-of-null error message to say "a function
>> > had an error" which seems wrong to me.
>>
>>
>> No it would be ideal if it said "function f had an error" instead of "car of
>> null mumbo jumbo".
>
> Yes, I completely agree with both of you. When I'm implementing `f', I
> want the car-of-null error. When I've given `f' to others, I want the
> "f has an error" message.
>
> That the default for `check-syntax' is the opposite of the usual one
> may be a historical accident; it may be because good syntax-error
> checking is usually a lot harder than good function-argument checking;
> or it may be that implementing a run-time "f has an error" would be
> expensive (i.e., I've never figured out how to implement something like
> that efficiently, but I've often wanted it).
>
> And I agree with Jay. At this point, though, the "bad syntax" message
> from `syntax-case' is ingrained in our code, in the sense that error
> messages would sometimes be worse for end users if we swapped the
> default. (I think that "function f had an error" would become ingrained
> in our code, too, if that were the default.)
>
>