| Age | Commit message (Collapse) | Author |
|---|
|
When typing a "with clause fails, type classes are used to possibly
help to insert coercions. If this heuristic fails, do not consider it
anymore to be the best failure since it has made type classes choices
which may be inconsistent with other constraints and on which no
backtracking is possible anymore (see new example in test suite file
4782.v).
This does not mean that using type classes at this point is good. It
may find an instance which help to find a coercion, but which might
still be a choice of instance and coercion which is incompatible with
other constraints.
I tend to think that a convenient way to go to deal with the absence
of backtracking in inserting coercions would be to have special
For the record, here is a some comments of what happens regarding
f9695eb4b and 827663982.
In the presence of an instance (x:=t) given in a "with" clause, with
t:T, and x expected of type T', the situation is the following:
Before f9695eb4b:
- If T and T' are closed and T <= T' is not satisfiable (no coercion
or not convertible), the test for possible insertion of a coercion
is postponed to w_merge, even though there is no way to get more
information since T ant T' are closed. As a result, t may be
ill-typed and the unification may try to unify ill-formed terms,
leading to #4872.
- If T and T' are not closed and contains evars of type a type class,
inference of type classes is tried. If it fails, e.g. because a
wrong type class instance is found, it was postponed to w_merge as
above, and the test for coercion is retried now interleaved with
type classes.
After f9695eb4b and 827663982e:
- If T and T' are closed and T <= T' is not satisfiable (no coercion
or not convertible), the test for possible insertion of a coercion
is an immediate failure. This fixes #4872.
- However, If T and T' are not closed and contains evars of type a
type class, inference of type classes is tried. If it gives closed
terms and fails, this is immediate failure without backtracking on
type classes, resulting in the problem added here to file 4872.v.
The current fix does not consider the result of the use of type
classes while trying to insert a coercion to be the last word on
it. So, it fails with an error which is not the error for conversion
of closed terms (ConversionFailed), therefore in a way expected by
f9695eb4b and 827663982e, and the "with" typing problem is then
postponed again.
|
|
Instead of relaunching the coqtop process and then open the warning window,
we rather fire the warning and wait for the user to press the OK button before
doing anything.
|
|
|
|
It seems like this code was copy-pasted from kernel/inductive.ml. It was
already dubious enough in the kernel. It feels completely wrong in the
checker.
|
|
with recent Coq
|
|
Goal 0=0 -> true=true.
intro H; rewrite H1.
was highlighting H1 but
Goal 0=0 -> true=true.
intro H; rewrite H.
was only highlighting the whole "intro H; rewrite H".
|
|
With par: the scenario is this one:
coqide --- master ---- proof worker 1 (has no par: steps)
---- proof worker 2 (has a par: step)
---- tac worker 2.1
---- tac worker 2.2
---- tac worker 2.3
Actor 2 installs a remote counter for universe levels, that are
requested to master. Multiple threads dealing with actors 2.x
may need to get values from that counter at the same time.
Long story short, in this complex scenario a mutex was missing
and the control threads for 2.x were accessing the counter (hence
reading/writing to the single socket connecting 2 with master at the
same time, "corrupting" the data flow).
A better solution would be to have a way to generate unique fresh universe
levels locally to a worker.
|
|
|
|
|
|
|
|
|
|
Check that the polymorphic status of everything that
is parameterized in nested sections is coherent.
|
|
This avoids postponing constraints which will surely produce
an occur-check and allow to backtrack on first-order unifications
producing those constraints directly (e.g. to apply eta).
(fixes HoTT/HoTT with 8.5).
|
|
non-recursive notations (#4815).
|
|
Function default_fail was always part of an ise_try. Its associated
error message was anyway thrown away. It is then irrelevant and could
be made simpler.
|
|
The tentative fix in f9695eb4b (which I was afraid it might be too
strong, since it was implying failing more often) indeed broke other
things (see #4813).
|
|
conversion on closed terms.
This will be useful to discriminate problems involving the "with"
clause and which fails by lack of information or for deeper reasons.
|
|
This can happen with the "with" clause (see e.g. #4782), but also with
recursive calls in first-order unification (e.g. "?n a a b = f a" when
a matching between "b" and "a" is tried before expanding f).
|
|
|
|
Trying to now catch all unification errors, but without a clear view
at whether some errors could be tolerated at the point of checking the
type of the binding.
|
|
Don't know however what should be the right guard to this try. Now
using catchable_exception, but even in 8.4, Failure was caught, which
is strange.
|
|
|
|
Typically, a problem of the form "?x args = match ?y with ... end" was
a failure even if miller-unification was applicable.
|
|
|
|
Renouncing to bypass the library path given by "camlp5o -where" what
we assume to be the default library location, considering that
-usecamlp5dir is here to deal with the non-standard installation
layout.
Renouncing to find camlp5 libraries in a subdirectory of the ocaml
library directory since we now know that camlp5o is found and that we
have a priori to trust option -where of camlp5o.
Additionally falling back on looking for camlp4 if a camlp5 library is
found but no camlp5 binary. Also using camlp5o as a reference since
after all this is camlp5o that we need.
In particular, this fixes situations where -usecamlp5dir is given but
"camlp5o -where" contradicts it.
If something has to be checked on windows, please tell.
|
|
Now, type_scope is consistently used whether it is an hypothesis or
the conclusion, and consistently not used when in "context".
The question of a compatibility support, e.g., as suggested by Jason,
using a scope is still open though.
See reports #3050 and #4398.
|
|
on the contrary of message given in 23041481f, while it introduces a
square time complexity of the size of the goal in subterm finding.
|
|
The exact and e_exact tactics were not registering the universes and
constraints of the hint correctly. Now using the same connect_hint_clenv
as unify_resolve, they do. Also correct the implementation of
prepare_hint to normalize the universes of the hint before abstracting
its undefined universes. They are going to be refreshed at each
application. This means that eauto using term can
use multiple different instantiations of the universes of term
if term introduces new universes. If we want just one instantiation
then the term should be abbreviated in the goal first.
|
|
The error message was not only causing coqtop to exit, but also coqide to
crash, which led to a rather poor user experience. Since the code does not
actually care about the file extension, this commit just removes the test.
|
|
We delay the externalization of application arguments in Constrextern,
so that they only get computed when they are actually explicitly displayed.
|
|
The COQLIBS variable contains some -Q and -I options, which are not
supported by the checker. So this commit introduces a COQCHKLIBS variable
that contains the proper options for coqchk. For the sake of homogeneity,
the COQDOCLIBS variable is also preprocessed in the same way. This means
that both variables have the same value, but they are kept separate in
case the user would like to override one and not the other.
This commit also removes some deprecated options from "coqchk --help".
They are not removed from coqchk itself to preserve backward compatibility
in the branch.
An open question is whether coqchk should support dummy options such as -Q
(interpreted as -R) or -I (ignored).
|
|
Tactic Notation "f" constr(x) := apply x. Ltac g x := f x.
Goal False.
g I. (* Was printing Top.Top#<>#1 *)
idtac; f I. (* Was not properly locating error *)
This is a "a minima" fix.
This a different fix than in trunk, so the merge will have to take the
trunk version.
|
|
Since d09def34, only the summary of libraries was included in the checksum, not
the actual content of the library. This quick fix performs the checking of the
checksum immediately, even if the loading is delayed.
|
|
|
|
#4770).
This also fixes comments not being properly skipped when looking for eol.
|
|
As a side effect, there should be a small speedup when ignoring comments.
This commit also fixes two bugs related to handling "$$" and "#" in
comments.
|
|
|
|
This brings a 10x speedup for going at the end of large .v files.
|
|
If the second allocation causes a collection of the minor heap, the first
allocation will be freed, thus causing a memory corruption.
Note: it only happens when computing the native projection of an opaque
value while the minor heap is almost full.
|
|
This reverts commit a66b57ba4bba866bb626bde2b6fe3b762347eb3e.
|
|
|
|
Some dubious evarmap manipulation is going on in destruct because of the
use of clenv primitives. Here, building a clenv was introducing new evars
that were not taken into account in the remainder of the tactic. We plug
them back using a local workaround.
Eventually, this code should be replaced by an evar-based one, but meanwhile,
we rely on what is probably a hack.
|
|
Command line options to be dropped got outdated after vi -> vio renaming.
This made the par: goal selector do not work in conjunction with -quick.
|
|
Return the most appropriate evar_map for commands that can run on
non-focused proofs (like Check, Show and debug printers) so that
universes and existentials are printed correctly (they are global
to the proof). The API is backwards compatible.
|
|
|
|
- In Program, a check_evars_are_solved was introduced
too early. Program does it's checking of evars by itself.
- In Function, the universe environments were not threaded
correctly.
|
|
|
|
Use the compatibility match construction to extract the compatibility
constant associated to a primitive projection.
|
|
Fix handling of non-polymorphic hints built from polymorphic values, or
simply producing new universes. We have to record the side effects of
global hints built from constrs which are not polymorphic when they
declare global universes, which might need to be discharged at the end
of sections too. Also issue a warning when a Hint is declared for a
polymorphic reference but the Hint isn't polymorphic itself (this used
to produce an anomaly). For [using] hints, treat all lemmas as
polymorphic, refreshing their universes at each use, as is done for
their existentials (also used to produce an anomaly).
|
|
Anomaly: Uncaught exception Unix.Unix_error(Unix.EACCES, "open", "lia.cache").
Please report.
|
