[racket-dev] tool transition leads to fragile middle-ground

On Feb 10, 2011, at 9:28 AM, Robby Findler wrote:

> Is it possible you're misparsing the error message? The message (I
> think!) says that the unit expects its import to be drscheme:tool^ but
> that it is invoked with something else (drracket:tool^ in this case).
> 
> (Not that this defends anything about the head-scratching-ness, but
> just to clarify what the message is trying to say.)

Yep, you're totally right.

Thanks!

John

> 
> Robby
> 
> On Thu, Feb 10, 2011 at 11:14 AM, John Clements
> <clements at brinckerhoff.org> wrote:
>> Updating tools from drscheme->drrracket involves two steps:
>> 
>> 1) changing the tool to refer to drracket, and
>> 2) changing the info.rkt file to refer to drracket-tools rather than tools.
>> 
>> Unfortunately, if you do #2 before #1, you get a real head-scratcher of an error message:
>> 
>> define-unit-binding: unit argument expects an untagged import with signature drscheme:tool^, which this usage context does not supply
>> 
>> The baffling thing about this is that drscheme:tool^ *is* the signature that this thing supplies.  I took a look at the tool infrastructure, and there's some regexp-rewriting stuff that scared me.  I conjecture that the (correct) error message is one that refers to the drracket:tool^ signature, but that it's getting rewritten into one that refers to drscheme:tool^.
>> 
>> I'm not sure how many people are going to fall into this hole, but the error message here steers you in the wrong direction.
>> 
>> John
>> 
>> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4669 bytes
Desc: not available
URL: <http://lists.racket-lang.org/dev/archive/attachments/20110210/de833c68/attachment.p7s>