| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This is the continuation of #244, we now deprecate `CErrors.error`,
the single entry point in Coq is `user_err`.
The rationale is to allow for easier grepping, and to ease a future
cleanup of error messages. In particular, we would like to
systematically classify all error messages raised by Coq and be sure
they are properly documented.
We restore the two functions removed in #244 to improve compatibility,
but mark them deprecated.
|
|
We move Coqlib to library in preparation for the late binding of
Gallina-level references. Placing `Coqlib` in `library/` is convenient
as some components such as pretyping need to depend on it.
By moving we lose the ability to locate references by syntactic
abbreviations, but IMHO it makes to require ML code to refer to
a true constant instead of an abbreviation/notation.
Unfortunately this change means that we break the `Coqlib`
API (providing a compatibility function is not possible), however we
do so for a good reason.
The main changes are:
- move `Coqlib` to `library/`.
- remove reference -> term from `Coqlib`. In particular, clients will
have different needs with regards to universes/evar_maps, so we
force them to call the (not very safe) `Universes.constr_of_global`
explicitly so the users are marked.
- move late binding of impossible case from `Termops` to
`pretying/Evarconv`. Remove hook.
- `Coqlib.find_reference` doesn't support syntactic abbreviations
anymore.
- remove duplication of `Coqlib` code in `Program`.
- remove duplication of `Coqlib` code in `Ltac.Rewrite`.
- A special note about bug 5066 and commit 6e87877 . This case
illustrates the danger of duplication in the code base; the solution
chosen there was to transform the not-found anomaly into an error
message, however the general policy was far from clear. The long
term solution is indeed make `find_reference` emit `Not_found` and
let the client handle the error maybe non-fatally. (so they can test
for constants.
|
|
We remove redundant functions `coq_constant`, `gen_reference`, and
`gen_constant`.
This is a first step towards a lazy binding of libraries references.
We have also chosen to untangle `constr` from `Coqlib`, as how to
instantiate the reference (in particular wrt universes) is a
client-side issue. (The client may want to provide an `evar_map` ?)
c.f. #186
|
|
|
|
|
|
|
|
|
|
|
|
We use heaps instead of continuously adding elements to an ordered list,
which was quadratic in the worst case.
As a byproduct, this solves bug #5359, which was due to a stack overflow on
big lists.
|
|
Several cleanups were performed.
1. Removal of dead code lurking around.
2. Removal of global variables used to pass arguments to functions, as well as
unnecessary mutable state here and there. We rely on state-passing and
encapsulated mutable state.
3. Removal of crazy reference manipulation and its replacement with proper list
handling, as well as cleaning up the source and taking advantage of invariants.
This should solve algorithmic limitations of the previous code.
4. Opacification of some structures to have a clearer idea of the code
requirements.
5. Cleaning of debug printing functions. We thunk the computation of the
debugging data, whose computation can be costly for no reason, and we rely
on Feedback-based interaction instead of Printf-debugging.
|
|
|
|
|
|
This is cumbersome, because now code may fail at link time if it's not
referring to the correct module name. Therefore, one has to add corresponding
open statements a the top of every file depending on a Ltac module. This
includes seemingly unrelated files that use EXTEND statements.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This fixes the stack overflow part of the bug, even if the tactic is still quite
slow. The offending functions have been written in a tail-recursive way.
|
|
|
|
|
|
we use a hashtable to reduce the complexity of creating
a duplicate-free list.
|
|
the function in_ideal of ideal.ml supposes the list of polynomials
does not contain zero and is duplicate free.
I force this invariant in the call of in_ideal in nsatz.ml4
the function clean_pol returns the reduced list plus a list of
booleans that indicates which polynomials have been deleted
the function expand_pol translates back the certificate of
the reduced to list to the complete list thanks to the list
of booleans.
The fix is quadratic with respect to the input list which should
be ok for reasonable usage of nsatz.
If there is some performance issue we could improve the in_pol
function.
|
|
changing
set (x := val)
into
let x := fresh "x" in
set (x := val)
|
|
module)
For the moment, there is an Error module in compilers-lib/ocamlbytecomp.cm(x)a
|
|
For now, the pack name reuse the previous .cma name of the plugin,
(extraction_plugin, etc).
The earlier .mllib files in plugins are now named .mlpack.
They are also handled by bin/ocamllibdep, just as .mllib.
We've slightly modified ocamllibdep to help setting the -for-pack
options: in *.mlpack.d files, there are some extra variables such as
foo/bar_FORPACK := -for-pack Baz
when foo/bar.ml is mentioned in baz.mlpack.
When a plugin is calling a function from another plugin, the name
need to be qualified (Foo_plugin.Bar.baz instead of Bar.baz).
Btw, we discard the generated files plugins/*/*_mod.ml, they are
obsolete now, replaced by DECLARE PLUGIN.
Nota: there's a potential problem in the micromega directory,
some .ml files are linked both in micromega_plugin and in csdpcert.
And we now compile these files with a -for-pack, even if they are
not packed in the case of csdpcert. In practice, csdpcert seems
to work well, but we should verify with OCaml experts.
|
|
|
|
|
|
|
|
|
|
|
|
corresponding Declare ML Module command. This changes essentially two
things:
1. ML plugins are forced to use the DECLARE PLUGIN statement before any
TACTIC EXTEND statement. The plugin name must be exactly the string passed to
the Declare ML Module command.
2. ML tactics are only made available after the Coq module that does the
corresponding Declare ML Module is imported. This may break a few things,
as it already broke quite some uses of omega in the stdlib.
|
|
The removed code isn't used locally and isn't exported in the signature
|
|
On the compilation of Coq, we can see an increase of ~20% compile time on
my completely non-scientific tests. Hopefully this can be fixed.
There are a lot of low hanging fruits, but this is an iso-functionality commit.
With a few exceptions which were not necessary for the compilation of the theories:
- The declarative mode is not yet ported
- The timeout tactical is currently deactivated because it needs some subtle
I/O. The framework is ready to handle it, but I haven't done it yet.
- For much the same reason, the ltac debugger is unplugged. It will be more
difficult, but will eventually be back.
A few comments:
I occasionnally used a coercion from [unit Proofview.tactic] to the old
[Prooftype.tactic]. It should work smoothely, but loses any backtracking
information: the coerced tactics has at most one success.
- It is used in autorewrite (it shouldn't be a problem there). Autorewrite's
code is fairly old and tricky
- It is used in eauto, mostly for "Hint Extern". It may be an issue as time goes
as we might want to have various success in a "Hint Extern". But it would
require a heavy port of eauto.ml4
- It is used in typeclass eauto, but with a little help from Matthieu, it should
be easy to port the whole thing to the new tactic engine, actually simplifying
the code.
- It is used in fourier. I believe it to be inocuous.
- It is used in firstorder and congruence. I think it's ok. Their code is
somewhat intricate and I'm not sure they would be easy to actually port.
- It is used heavily in Function. And honestly, I have no idea whether it can do
harm or not.
Updates:
(11 June 2013) Pierre-Marie Pédrot contributed the rebase over his new stream based
architecture for Ltac matching (r16533), which avoid painfully and expensively
working around the exception-throwing control flow of the previous API.
(11 October 2013) Rebasing over recent commits (somewhere in r16721-r16730)
rendered a major bug in my implementation of Tacticals.New.tclREPEAT_MAIN
apparent. It caused Field_theory.v to loop. The bug made rewrite !lemma,
rewrite ?lemma and autorewrite incorrect (tclREPEAT_MAIN was essentially
tclREPEAT, causing rewrites to be tried in the side-conditions of conditional
rewrites as well). The new implementation makes Coq faster, but it is
pretty much impossible to tell if it is significant at all.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16967 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16806 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
with OCaml 3.12.1).
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16787 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
The problem occurs when a customized ring/field structure
declared with a so-called "morphism" (see 24.5 in the manual) tactic
allowing to reify (numerical) constants efficiently.
When declaring a ring/field structure, the user can provide a cast
function phi in order to express numerical constants in another type than
the carrier of the ring. This is useful for instance when the ring is
abstract (like the type R of reals) and one needs to express constants
to large to be parsed in unary representation (for instance using a
phi : Z -> R).
Formerly, the completeness of the tactic required (phi 1) (resp. (phi 0))
to be convertible to 1 (resp. 0), which is not the case when phi is
opaque. This was not documented untill recently
but I moreover think this is also not desirable
since the user can have good reasons to work with such an opaque case phi.
Hence this commit:
- adds two constructors to PExpr and FExpr for a correct reification
- unplugs the optimizations in reification: optimizing reification
is much less efficient than using a cast known to the tactic.
TODO : It would probably be worth declaring IZR as a cast in the ring/field
tactics provided for Reals in the std lib.
The completeness of the tactic formerly relied on the fact that
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16730 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
problem are cleared (in case they denote applicative terms with at
least 6 arguments). However some of them are used for type class
instance inference (e.g. hypotheses of type Integral_domain).
This commits prevents clearing these hypotheses.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16469 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16292 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16283 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16129 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16104 85f007b7-540e-0410-9357-904b9bb8a0f7
|
