The Coq mailing list

[Matthieu Sozeau <mattam AT mattam.org>]

Hi Beta,

  You should rather use uconstr and explicitely program which one is interpreted first and pass the sigmas around, merging is bad style. Maybe a hint about your example could help us.

Cheers,

-- Matthieu

On Wed, Apr 13, 2016 at 4:46 PM Beta Ziliani <bziliani AT famaf.unc.edu.ar> wrote:

On Wed, Apr 13, 2016 at 11:24 AM, Pierre-Marie Pédrot <pierre-marie.pedrot AT inria.fr> wrote:

On 13/04/2016 16:14, Beta Ziliani wrote:
> Mmm... Not quite. I have two open_constrs c and c' with sigmas s and s'.
> It looks like in s there are no evars from c' and viceversa :-(

This does not contradict what I said, quite the contrary. Each of your
evarmaps is an extension of the evarmap from the goal, but they are
incomparable between each other. Thus, a good generator of evar leaks.

You should try to merge them...

Sorry, meaning? I don't see anything like a merge for evars in evd.mli. Should I define my own?

 

PMP