The Coq mailing list

[Julien Tesson <julien.tesson AT univ-orleans.fr>]

Le 12/10/2012 15:27, AUGER Cédric a écrit :
> Le Fri, 12 Oct 2012 15:00:21 +0200,
> Julien Tesson 
> <julien.tesson AT univ-orleans.fr>
>  a écrit :
>
>> Le 11/10/2012 10:47, Arnaud Spiwack a écrit :
>>> Yes.
>>>
>>> It's occasionally useful to be able to compute with existential
>>> variables.
>>>
>>> Why is it strange ?
>> It looked strange for me at first because I would have expected the
>> same behaviour (in term of failure) between  [Eval compute in a] and
>> [Definition x := Eval compute in a], and because when I use type
>> classes, I expect that the instance search mechanism warn me when it
>> failed to find a matching instance.
>>
>> This new behaviour seems interesting  but I found it surprising at
>> first.
>>
>> Regards,
>> Julien
> I guess "Definition x := Eval compute in a" could be made to work by
> automatically generalizing existentials, buth then I fear it would
> become hard to predict what really will x be.
I'm not sure it would be a good idea.
In the cases where I use type classes, I really want that the definition 
fails if there exists no instance which give a definition for the field I'm 
using.