The Coq mailing list

[Matthieu Sozeau <mattam AT mattam.org>]

Hi Vadim,

On Wed, Aug 17, 2016 at 1:26 AM Vadim Zaliva <vzaliva AT cmu.edu> wrote:

On Tue, Aug 16, 2016 at 3:20 PM, Jonathan Leivent <jonikelee AT gmail.com> wrote:

[setoid_rewrite <-compose_assoc at 2] works in Vadim's example.  The way that setoid_rewrite matches patterns is different from the way that rewrite matches patterns.  I think that setoid_rewrite's method is more intuitive.

That's very interesting. Thanks! I think it would good to mention this in the documentation. Confusingly it mentions that 'rewrite ... at ..." uses Setoid equality, but apparently that does not mean that it matches patterns in the same way as 'setoid_rewrite'.

Indeed, it's again the difference between n-th occurrence of the first instance of the lemma and n-th occurrence found for any instance of the lemma.

 

My remaining problem is that I am working in environment where I am already using setoid equality and thus my 'default_relation' is re-defined ('equiv' not 'eq'), while `comp_assoc` is defined for 'eq'. So I need to find a way to use `setoid_rewrite` tactics but with 'eq' relation. You can specify one for 'setoid_replace' tactic ('using relation' argument), but unfortunately not for `setoid_rewrite`!

I'm not sure why you would need to specify a relation for setoid_rewrite, as it is always inferred from the lemma itself. Do you have an example?

Best,

-- Matthieu