The Coq mailing list

[Jeremy Dawson <Jeremy.Dawson AT anu.edu.au>]

Hi Gaëtan,

all: repeat solve_eqs
doesn't succeed, I don't know exactly what all : repeat solve_eqs does
(previous emails addressed this question and the answer I was given) 
though in this case it's consistent with applying
1: repeat solve_eqs (which leaves one subgoal, so subgoal 2: is the 
original subgoal 2)
then applying 2 : repeat solve_eqs.
or maybe just
1: solve_eqs. then 2: solve_eqs. (ie without "repeat", I can't tell)

It doesn't go back to try 1 : solve_eqs again.

eg see the following, where the
second application of the tactic
all : repeat solve_eqs
makes progress beyond the results of the first application of it.

2 focused subgoals
(shelved: 5)

   V : Set
   Φ2, Ψ1, Ψ2, Γ1 : list (PropF V)
   Q : PropF V
   sel1, l, l1 : list (PropF V)
   ============================
   ((Γ1 ++ Q :: sel1) ++ l ++ Φ2, Ψ1 ++ l1 ++ Ψ2) =
   (?Γ1 ++ ?Q :: ?Γ2 ++ ?Γ3, ?Δ)

subgoal 2 is:
  G ++ (?Γ1 ++ ?Γ2 ++ ?Q :: ?Γ3, ?Δ, d0) :: H6 =
  G ++ ((Γ1 ++ sel1) ++ l ++ Q :: Φ2, Ψ1 ++ l1 ++ Ψ2, d0) :: H6

gen_moveL < all : repeat solve_eqs.
1 subgoal

   V : Set
   Φ2, sel1, l : list (PropF V)
   ============================
   sel1 ++ l ++ Φ2 = (sel1 ++ l) ++ Φ2

gen_moveL < all : repeat solve_eqs.
This subproof is complete, but there are some unfocused goals.
Try unfocusing with "}".

Cheers,

Jeremy


On 01/25/2019 08:53 PM, Gaëtan Gilbert wrote:
> Note that repeat runs the inner tactic focused on each goal separately
> so "repeat (all: solve_eqs)" would (if the syntax allowed it) be the 
> same as "repeat (only 1: solve_eqs)" which is also the same as "repeat 
> solve_eqs", and "repeat (only 2: solve_eqs)" is the same as "repeat 
> fail" ie idtac.
> 
> I feel like "all: repeat solve_eqs" should work given what you describe, 
> can you explain how it fails?
> 
> Gaëtan Gilbert
> 
> On 25/01/2019 06:54, Jeremy Dawson wrote:
>>
>> Hi Cyprian,
>>
>> What I wanted is what I would express as
>> repeat (all : solve_eqs)
>> but it seems as though that form is forbidden.
>>
>> The various things I tried as indicated in my earlier email were simply
>> looking for alternatives that work.
>>
>> Detailed background to what I want: solve_eqs doesn't seem to enlarge
>> the number of goals in the cases where I want to use it, and sometimes
>> solves a goal, sometimes only after solve_eqs has instantiated an evar
>> in another goal, sometimes only after being repeatedly called on the
>> same goal (where each time it simplifies the goal)
>>
>> Cheers,
>>
>> Jeremy
>>
>>
>> On 01/25/2019 10:00 AM, Cyprien Mangin wrote:
>>> What you probably meant is something like "all : repeat solve_eqs"