| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
In addition to a priority, cleanup the interfaces for passing this
information as well. The pattern, if given, takes priority over the
inferred one.
We only allow Existing Instances gr ... gr | pri. for now, without
pattern, as before.
Make the API compatible to 8.5 as well.
|
|
|
|
|
|
Was PR#321: Handling of section variables in hints
|
|
Was PR#319: More error tagging, try to fix bug 5135
|
|
Was PR#337: Fix arguments
|
|
Type annotations in unrelated binders were badly interfering with
detection of recursive binders in notations.
|
|
|
|
The main point of this change is to fix #3035: Avoiding trailing
arguments in the Arguments command, and related issues occurring in
HoTT for instance. When the "assert" flag is not specified, we now
accept prefixes of the list of arguments.
The semantics of _ w.r.t. to renaming has changed. Previously, it meant
"restore the original name inferred from the type". Now it means "don't
change the current name".
The syntax of arguments is now restricted. Modifiers like /, ! and
scopes are allowed only in the first arguments list.
We also add *a lot* of missing checks on input values and fix various
bugs.
Note that this code is still way too complex for what it does, due to
the complexity of the implicit arguments, reduction behaviors and renaming
APIs.
|
|
|
|
Was PR#329: Fix #5127 Memory corruption with the VM
|
|
|
|
|
|
|
|
|
|
|
|
+ a few improvements on 5f1dd4c40 (lexing of { and }).
|
|
|
|
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.
|
|
|
|
|
|
Refine fix for bug #4763, fixing #5149
Tactic [Refine.solve_constraints] and global option
Adds a new multi-goal tactic [Refine.solve_constraints] that forces solving of
unification constraints and evar candidates to be solved. run_tactic now calls
[solve_constraints] at every [.], preserving (mostly) the 8.4/8.5 behavior of tactics.
The option allows to unset the forced solving unification constraints at
each ".", letting the user control the places where the use of
heuristics is done.
Fix test-suite files too.
|
|
Adds a compatibility flag so that the behavior of 8.5 can be obtained too.
Original commit:
unification.ml: fix for bug #4763, unif regression
Do not force all remaining conversions problems to be solved after
the _first_ solution of an evar. This was hell to track down, thanks
for the help of Maxime. contribs pass and HoTT too.
|
|
|
|
|
|
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.
|
|
due to cleared/reverted section variables.
This fixes the get_type_of but requires keeping information
around about the section hyps available in goals during resolution.
It's optimized for the non-section case (should incur no cost there),
and the case where no section variables are cleared.
|
|
|
|
This makes [refine] typecheck the term only once (instead of twice),
(Qed excluded of course). Fix test-suite file for output of constraints
accordingly.
|
|
|
|
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".
|
|
The new algorithm produces different sets of arcs than in 8.5, hence
existing developments may fail to pass now because they relied on
the (correct but more approximate) result of minimization in 8.5 which
wasn't insensitive.
The algorithm works bottom-up on the (well-founded) graph to
find lower levels that an upper level can be minimized to. We
filter said lower levels according to the lower sets of the other levels
we consider. If they appear in any of them then we don't need their
constraints. Does not seem to have a huge impact on performance in HoTT,
but this should be evaluated further.
Adapt test-suite files accordingly.
|
|
This patch converts the `search_*` functions to use an iter-style API.
Consequently, the Search Vernac will produce a message for each search
result, greatly improving roundtrip time as IDEs can effectively display
the results in a streaming way.
It also allows different printers to be used.
I didn't observe a performance difference (as things seem to be
dominated by printing and `Declaremods`).
As a minor tweak, we make search with "Output Name Only" more efficient.
Motivation:
-----------
Currently, the main search API `Search.generic_search` is an effectful,
iteration-based function:
```ocaml
val generic_search : int option -> display_function -> unit
```
This type is imposed by `Declaremods`, which exposes an effectful,
iteration-based API to traverse Coq library objects.
The `Search.search_*` functions try to offer a more functional API by
returning a list of pretty printing commands. They need to build an
internal intermediate list for that purpose.
However, this is a waste of time, as the destination of these lists is
to be flushed out by the printer right away.
|
|
|
|
When trying different possible return predicates, returns the error
given by the first try, assuming this is the most interesting one.
|
|
In order to get proper coloring, we must tag the headers of error
messages in `CError`.
This should fix bug
https://coq.inria.fr/bugs/show_bug.cgi?id=5135
However, note that this could interact badly with the richpp printing
used by the IDE. At this level, we have no clue which tag we'd like to
apply, as we know (and shouldn't) nothing about the top level backend.
Thus, for now I've selected the console printer, hoping that the
`Richpp` won't crash the IDE.
|
|
|
|
|
|
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 commit which caused the regression (5ea2539d4a) looks reasonable.
However, it happens that this commit made that unification started in
the #4955 example to follow a path with problems of the form
"proj ?x == ?x" which clearly are unsolvable (both ?x have the same
instance).
We hack the corresponding very permissive occur-check function so that
it is a bit less permissive.
One day, this occur-check function would have to be rewritten in a
more stricter way.
----------------------------------------------------------------------
Extra comments:
I could list several functions for occur check of evars.
Four of them are too strict in the sense that they do not take into
account occurrences of evars which may disappear by reduction, nor
evars which have instances made in such a way that the occur-check is
solvable (as in "fst ?x[y:=(0,0)] = ?x[y:=0]"). These are:
- Termops.occur_evar for clenv, evar_refiner, tactic unification
- syntactic check without any normalization, even on defined evars
- Evarutil.occur_evar_upto for refine and the V82 compatibility mode
- syntactic check without any normalization but inlining of defined evars
- Evarsolve.occur_evar_upto_types for evar_define
- syntactic check without any normalization but inlining of defined evars
- occur-check in the type of evars met
- optimization for not visiting several time the same evar
- Evarsolve.noccur_evar for pattern unification and for inversion of
substitution (evar/evar or evar/term problems)
- syntactic check without any normalization but inlining of defined evars
- occur-check in the type of evars met in arguments of projections
- occur-check in the type of variables met in arguments of projections
- optimization for not visiting several time the same evar
- optimization for not visiting several time the type of the same variable
- note: to go this way, it seems to me that it should check also in
all types which are a cut-formula in the expression
One is much too lax:
- Evarconv.occur_rigidly
- no recursive check except on the types of "forall" and sometimes
in the arguments of an application
- note: there is obviously a large room for refinements without
loosing solutions
|
|
`Notation ".a" := nat.' was accepted and used for printing but not
recognized in parsing. Now it does. Other examples in test-suite.
|
|
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.
|
|
|
|
The output was embedding local paths from my machine.
|
|
|
|
branches (see PR #323).
|
|
|
|
|
