| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
Type annotations in unrelated binders were badly interfering with
detection of recursive binders in notations.
|
|
Was PR#338: Remove warning now that info_auto is fixed.
|
|
Removes a warning dating from 8.5 signaling that info_auto
and info_trivial are broken and advising to use Info 1 auto
instead. Now, these tactics are fixed and thus they can be
used again. They do not do exactly the same thing as
Info 1 auto and may be more useful for the learner.
|
|
This reverts commit c9c54122d1d9493a965b483939e119d52121d5a6.
This behavior of refine has changed three times in recent years, so
let's take the time to make up our mind and wait for a major release.
Btw, onhyps=None is not a sane way to express that a tactic should be
applied to all hypotheses.
|
|
|
|
|
|
Was PR#233: Fix a bug in error printing of unif constraints
|
|
Was PR#329: Fix #5127 Memory corruption with the VM
|
|
Was PR#326: Extend documentation of auto
|
|
|
|
The bytecode interpreter ensures that the stack space available at some
points is above a static threshold. However, arbitrary large stack space
can be needed between two check points, leading to segmentation faults
in some cases.
We track the use of stack space at compilation time and add
an instruction to ensure greater stack capacity when required. This is
inspired from OCaml's PR#339 and PR#7168.
Patch written with Benjamin Grégoire.
|
|
|
|
|
|
New fix. A bit less consistent with the spirit of this Makefile as we
list prerequisite files explicitly, but avoid to redo subsystems at
each call to "make" as with previous fix.
Other solutions from experts are welcome.
|
|
This assertion checked that two arguments in the same position were equal,
but in fact, since one might have already been reduced, they are only
convertible (which is too costly to check in an assertion).
|
|
When inserting a character in an already processed buffer, a message is
sent to Coq so that the proof is backtracked and the character is
inserted. If a second character is inserted while Coq is still busy with
the first message, the action is canceled, but the signal is no longer
dropped so that the second character is properly inserted.
|
|
|
|
|
|
Conversion problems are in a de Bruijn environment that may
include references to unbound rels (because of the way evars
are created), we patch the env to named all de Bruijn variables
so that error printing does not raise an anomaly. Also fix a minor
printing bug in unsatisfiable constraints error reporting.
HoTT_coq_117.v tests all of this.
|
|
This reverts commit 0b417c12eb10bb29bcee04384b6c0855cb9de73a.
A good fix requires to review a bit the design of unification constraint
postponement, which we do in 8.6. We leave things as they are in 8.5 for
compatibility.
|
|
Was PR#328: Change the order of arguments of fig2dev.
|
|
|
|
If an existing evar was cleared in pretyping (typically while
processing "ltac:"), it created an evar considered as new. Updating
them instead along the "cleared" flag.
If correct, I suspect similar treatment should be done for refining
along "change", "rename" and "move".
|
|
These variants are from 8.3 but were never documented except in CHANGES.
|
|
The doc of auto hints now mention again that sometimes a hint
will be used with simple apply and sometimes it will be used
with exact.
It does not try to be fully precise in that we don't
necessarily want to document the behaviors of auto that
we might like to change.
See also the discussion on commit 9227d6e.
|
|
For some reason, with my version of transfig (which seems to be the latest),
the order of arguments of the fig2dev command matters: -L png must come
before -m 2. I suppose that this fix shouldn't break things for others.
|
|
This way of giving the summary avoids redundancy as much as possible.
It is helpful for the auto-completion of Company-Coq of @cpitclaudel.
|
|
Now that this tactic has been fixed (commit 58d1381),
it needed to get documented.
|
|
Improve the description of what auto/eauto do.
These two tactics rely on the simple version of apply/eapply.
Since this simple version is available to the end user,
it is better to mention it. See also the confusion that such
description can create in the thread "Understanding auto" on Coq-Club.
|
|
I do not know how to provide a proper test in 8.5, as the location on my
machine appears in the error printed when loading the file. Adding a Fail
on the End command does not help much either, because it simply does not
print anything.
Do not merge this commit in 8.6, we still want a test there.
|
|
The fix solves the original bug report but it only turns the Not_found
into a normal error in the alternative example by Guillaume. See
test-suite file for comments on how to eventually improve the
situation and find a solution in Guillaume's example too.
|
|
|
|
|
|
|
|
"msg_debug (mt())" is not identity, so we return back to how it was
done in 8.4, contracting a repeated pattern-matching into one.
|
|
Header was missing in last commit.
One day, it would be nice to unify the display of "debug auto" and
"debug eauto"...
|
|
|
|
This happens when recursive notations are used to define recursive
notations.
|
|
Was PR#286: Fix the definition of if_then_else for tactics with multiple
success.
|
|
|
|
|
|
|
|
Previously, some splipped through and were caught by unrelated calls to
typeclass resolution.
|
|
side_effects. Partial solution to the handling of side effects
in proofview.
|
|
|
|
In `Ftactic` the number of results could desynchronise with the number
of goals when some goals were solved by side effect in a different
branch of a `DISPATCH`.
See [coq-bugs#4416](https://coq.inria.fr/bugs/show_bug.cgi?id=4416).
|
|
More precisely, commands that calls parse_entry put the lexer in an
inconsistent state, breaking the lexing of bullet which relies on it.
(Not to be pushed to v8.6 which has a better fix).
|
