The Coq mailing list

[Beta Ziliani <beta AT mpi-sws.org>]

Thanks PMP and Matthieu for their inputs. For the record, what I did in the end was to use uconstr:

TACTIC EXTEND munify_tac
| ["munify" uconstr(c) uconstr(c') ] -> ...
END

and to define the following function to understand the closed_glob_term coming from uconstr:

let understand env sigma {Glob_term.closure=closure;term=term} =
  let open Pretyping in
  let flags = all_no_fail_flags in
  let lvar = { empty_lvar with
               ltac_constrs = closure.Glob_term.typed;
               ltac_uconstrs = closure.Glob_term.untyped;
               ltac_idents = closure.Glob_term.idents;
             } in
  understand_ltac flags env sigma lvar WithoutTypeConstraint term

​

On Wed, Apr 13, 2016 at 12:48 PM, Beta Ziliani <beta AT mpi-sws.org> wrote:

I think this will work for me. I'm building the tactic to unify two arguments using Unicoq.

Written from mobile. Excuse my limited communication.

On Apr 13, 2016 12:23 PM, "Matthieu Sozeau" <mattam AT mattam.org> wrote:

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