| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Otherwise, the interpreter sees already unified evars as accumulators
rather than actual constants, thus preventing the computations from
progressing.
This was caused by 6b61b63bb8626827708024cbea1312a703a54124, which removed
evar normalization. The effect went unnoticed because the computed term is
still convertible to the reduced term, except that it is the lazy
machinery that ends up reducing it, rather than the bytecode one.
So, performances became abysmal, seemingly at random.
|
|
|
|
We store bound variable names instead of functions for both branches and
predicate, and we furthermore add the parameters in the node. Let bindings
are not taken into account and require an environment lookup for retrieval.
|
|
We simply use evars instance in the right order while reading back VM values.
|
|
|
|
No need to store the case_info, all the data is reconstructible from the
context. Furthermore, this reconstruction is performed in a context where
we already access the environment, so performance is not at stake.
Hopefully this will also reduce the number of globally allocated VM values,
since the switch representation now only depends on the shape of the inductive
type.
|
|
Reviewed-by: ppedrot
|
|
|
|
Persistent arrays expose a functional interface but are implemented
using an imperative data structure. The OCaml implementation is based on
Jean-Christophe Filliâtre's.
Co-authored-by: Benjamin Grégoire <Benjamin.Gregoire@inria.fr>
Co-authored-by: Gaëtan Gilbert <gaetan.gilbert@skyskimmer.net>
|
|
|
|
This corresponds more naturally to the use we make of them, as we don't need
fast indexation but we instead keep pushing terms on top of them.
|
|
|
|
Add headers to a few files which were missing them.
|
|
|
|
We also remove trailing whitespace.
Script used:
```bash
for i in `find . -name '*.ml' -or -name '*.mli' -or -name '*.mlg'`; do expand -i "$i" | sponge "$i"; sed -e's/[[:space:]]*$//' -i.bak "$i"; done
```
|
|
* This commit add float instructions to the VM, their encoding in bytecode
and the interpretation of primitive float values after the reduction.
* The flag '-std=c99' could be added to the C compiler flags to ensure
that float computation strictly follows the norm (ie. i387 80-bits
format is not used as an optimization).
Actually, we use '-fexcess-precision=standard' instead of '-std=c99'
because the latter would disable GNU asm used in the VM.
|
|
|
|
|
|
Kernel should be mostly correct, higher levels do random stuff at
times.
|
|
It used to simply remember the normal form of the type of the constructor.
This is somewhat problematic as this is ambiguous in presence of
let-bindings. Rather, we store this data in a fully expanded way, relying
on rel_contexts.
Probably fixes a crapload of bugs with inductive types containing
let-bindings, but it seems that not many were reported in the bugtracker.
|
|
This work makes it possible to take advantage of a compact
representation for integers in the entire system, as opposed to only
in some reduction machines. It is useful for heavily computational
applications, where even constructing terms is not possible without such
a representation.
Concretely, it replaces part of the retroknowledge machinery with
a primitive construction for integers in terms, and introduces a kind of
FFI which maps constants to operators (on integers). Properties of these
operators are expressed as explicit axioms, whereas they were hidden in
the retroknowledge-based approach.
This has been presented at the Coq workshop and some Coq Working Groups,
and has been used by various groups for STM trace checking,
computational analysis, etc.
Contributions by Guillaume Bertholon and Pierre Roux <Pierre.Roux@onera.fr>
Co-authored-by: Benjamin Grégoire <Benjamin.Gregoire@inria.fr>
Co-authored-by: Vincent Laporte <Vincent.Laporte@fondation-inria.fr>
|
|
This is a pre-requisite to use automated formatting tools such as
`ocamlformat`, also, there were quite a few places where the comments
had basically no effect, thus it was confusing for the developer.
p.s: Reading some comments was a lot of fun :)
|
|
|
|
|
|
|
|
|
|
There were actually two broken things with VM + evars, the fixes are:
- Do not pass let-bound arguments to evars.
- Use the right order for evar arguments.
Native compilation seems to be suffering from the same shortcomings, I will
open a separate bug and adapt the PR.
|
|
In #7607, dead code that used to handle non-dependent return predicates
was removed. This made the reification branch expecting non-functions
in predicates dead code. We fix this by using an assert instead.
|
|
compilers
|
|
|
|
I believe this is legacy code due to a previous, more complex
representation of return predicates in the kernel.
|
|
Note that the normalization of the context of the return predicate was
not done by the vm but by the lazy machine. The patch also "fixes" an
anomaly in the case of an arity which was not in canonical form as in:
Inductive A : nat -> id (nat->Type) := .
Eval vm_compute in fun x => match x in A y z return y = z with end.
|
|
We now have only two notions of environments in the kernel: env and
safe_env.
|
|
We bootstrap the circular evar_map <-> econstr dependency by moving
the internal EConstr.API module to Evd.MiniEConstr. Then we make the
Evd functions use econstr.
|
|
We forbid calling `EConstr.to_constr` on terms that are not evar-free,
as to progress towards enforcing the invariant that `Constr.t` is
evar-free. [c.f. #6308]
Due to compatibility constraints we provide an optional parameter to
`to_constr`, `abort` which can be used to overcome this restriction
until we fix all parts of the code.
Now, grepping for `~abort:false` should return the questionable
parts of the system.
Not a lot of places had to be fixed, some comments:
- problems with the interface due to `Evd/Constr` [`Evd.define` being
the prime example] do seem real!
- inductives also look bad with regards to `Constr/EConstr`.
- code in plugins needs work.
A notable user of this "feature" is `Obligations/Program` that seem to
like to generate kernel-level entries with free evars, then to scan
them and workaround this problem by generating constants.
|
|
|
|
|
|
We simply treat them as as an application of an atom to its instance,
and in the decompilation phase we reconstruct the instance from the stack.
This grants wish BZ#5659.
|
|
This simplifies the representation of values, and brings it closer to
the ones of the native compiler.
|
|
|
|
|
|
We separate functions dealing with VM values (vmvalues.ml) and
interfaces of the bytecode interpreter (vm.ml). Only the former relies
on untyped constructions.
This also makes the VM architecture closer to the one of native_compute,
another patch could probably try to share more code between the two for
conversion and reification (not trivial, though).
This is also preliminary work for integers and arrays.
|
|
|
|
The patch has three parts:
- Introduction of a configure flag `-bytecode-compiler (yes|no)`
(due to static initialization this is a configure-time option)
- Installing the hooks that register the VM with the pretyper and the
kernel conditionally on the flag.
- Replacing the normalization function in `Redexpr` by compute if the
VM is disabled.
We also rename `Coq_config.no_native_compiler` to `native_compiler`
and `Flags.native_compiler` to `output_native_objects` [see #4607].
|
|
We do up to `Term` which is the main bulk of the changes.
|
|
Instead of returning either an instance or the set of constraints, we rather
return the corresponding abstracted context. We also push back all uses of
abstraction-breaking calls from these functions out of the kernel.
|
|
|
|
|
|
It stores both universe constraints and subtyping information for
blocks of inductive declarations.
At this stage the there is no inference or checking implemented. The
subtyping information simply encodes equality of levels for the condition of
subtyping.
|
