profile
viewpoint
Emilio Jesús Gallego Arias ejgallego Paris https://www.irif.fr/~gallego/ “La perfection est atteinte, non pas lorsqu'il n'y a plus rien à ajouter, mais lorsqu'il n'y a plus rien à retirer.”

Deducteam/lambdapi 156

Proof assistant based on the λΠ-calculus modulo rewriting

ejgallego/coq-serapi 99

Coq Protocol Playground with Se(xp)rialization of Internal Structures.

ejgallego/coq-plugin-template 13

Template of Coq Plugin using the Dune build system, and showcasing some advanced features

ejgallego/CodeMirror-TeX-input 3

TeX-style unicode input for CodeMirror

deian/lio-lecture 1

Building simplified LIO lecture

ejgallego/coqide-exp 1

Experiments with split coqide

ejgallego/coqide2-split 1

Frozen CoqIDE gtk2 version to build against Coq 8.10

brando90/coq-serapi-python 0

API for Coq using SerAPI

Deducteam/atom-dedukti 0

Atom mode for Dedukti (Lambdapi)

ejgallego/aac-tactics 0

This Coq plugin provides tactics for rewriting universally quantified equations, modulo associative (and possibly commutative) operators.

issue commentgarrigue/lablgtk

lablgtk fails on multicore due to use of naked pointers

@kit-ty-kate raised this issue again, so I proposed we coordinate a possible effort here.

talex5

comment created time in 3 days

pull request commentjscoq/addon-mathcomp-extra

Adding more extras

Note that for mathcomp analysis you will need hierarchy-builder as well

Ups, that may go beyond the amount of time I have for it.

All these plugins are, IMHO, an engineering nightmare.

CohenCyril

comment created time in 3 days

push eventejgallego/coq

Hugo Herbelin

commit sha 1ca6daf280492909c362ff7b4112616f64ba1423

Notations for numerals are treated natively and should not modify the parser. For instance: Notation "0" := tt. Check 0%nat. should not return "tt".

view details

Ali Caglayan

commit sha a793e055518c647c82b1be7f02728faa4844701f

Added test for Scheme names.

view details

coqbot-app[bot]

commit sha bd097a48a542f94b0d3120749792e58673ec242c

Merge PR #14730: Remove the eta flag from CClosure redflags. Reviewed-by: gares Ack-by: SkySkimmer Co-authored-by: SkySkimmer <SkySkimmer@users.noreply.github.com>

view details

coqbot-app[bot]

commit sha a29e3eae1932309a340f834e4989237ec83654aa

Merge PR #14742: Output test for `Scheme` command(s). Reviewed-by: SkySkimmer Co-authored-by: SkySkimmer <SkySkimmer@users.noreply.github.com>

view details

coqbot-app[bot]

commit sha 8c4e6dff14e551fb54595335d90d14c1d697b362

Merge PR #14713: Fixes #14708: anomalies in applied notation variable in match patterns Reviewed-by: SkySkimmer Co-authored-by: SkySkimmer <SkySkimmer@users.noreply.github.com>

view details

Ali Caglayan

commit sha 1286c91f168964a5ae46bc05653402d9421fc62a

Add category-theory to ci.

view details

Pierre-Marie Pédrot

commit sha 3fd62a8b1006e82bf1942e6a5316f7d2977b941d

Merge PR #14598: Explicit Projection nodes in concrete syntax Ack-by: Janno Reviewed-by: gares Reviewed-by: jfehrle Reviewed-by: ppedrot

view details

Gaëtan Gilbert

commit sha c43979a45e45cc8285ace507f2ae5ffcfc978896

Replace confusing Coercion.inh_coerce_to_prod by new Coercion.remove_subset No actual coercion is happening, so this shouldn't be called inh_coerce_*

view details

Gaëtan Gilbert

commit sha 73cc3e38be86386b575c7924a68a14655de791bc

Don't early error in pretype_lambda when unification can succeed

view details

Gaëtan Gilbert

commit sha ad573e287d07de40ecb5f6a6c5a0fd70e531fa3d

More use of unification in inh_app_fun

view details

Gaëtan Gilbert

commit sha a9146af280bb7518933ed61e83a8483cc61600c8

Add tests for canonical structures vs lambdas and applications While we're here convert the test files to modules. Co-authored-by: Janno <Janno@users.noreply.github.com>

view details

Pierre-Marie Pédrot

commit sha e7b938359c9410f7c650bb28cb0baa468c6a1858

Faster algorithm for addition of coercions. Instead of crawling a set of pairs, we maintain an indexing structure for each component of the pair and use the corresponding subset when needed. The underlying algorithm is still quadratic in the worst case but it should be faster on average.

view details

Pierre-Marie Pédrot

commit sha eca9e617b471a08eb94abb8367af5c6d29b7e00c

On-the-fly η-reduction of dnet patterns and terms. Due to the way opaque terms are handled, dnet pattern matching was failing to recognize that the η-expansion of a term was matching that term, in both directions. To solve this, we simply η-reduce both terms and patterns before adding and retrieving them from the dnet. Fixes #14731: Term matching in discriminated hints is not up-to η-expansion.

view details

Pierre-Marie Pédrot

commit sha dea79e1d9afcb91632fa007f027d59b88b37307b

Remove the Lambda node in Btermdn patterns. Statically recognizing lambda abstractions would lead to unexpected failures in complex cases involving η-expansions with holes and implicit unfolding. To preserve backwards compatibility, we simply consider that lambda abstractions match anything. It does not seem to have any negative performance impact and is enough to make strict variable recognition work in #14722.

view details

Pierre-Marie Pédrot

commit sha 77d2f425b071fc0463718377ca002edf0903a0b8

Adding changelog.

view details

Pierre-Marie Pédrot

commit sha 4913281c96b15a8f0c65f26d7e08af49ef2befad

Adding a test-case from @Janno.

view details

Pierre-Marie Pédrot

commit sha 1cd865e7e19a780091add8deecca2a3569d24e40

Merge PR #14715: More comprehensive use of unification when pretyping lambdas and applications Ack-by: Janno Reviewed-by: ppedrot

view details

Gaëtan Gilbert

commit sha 58ab8c34b2b1eeaa4b3fc11f6169783b6a0dbb74

new_instance functions: locality argument is not optional

view details

Gaëtan Gilbert

commit sha 2f16cb1cb1384ab0483ec5c21b18a0443fff4958

Use type hint_locality in classes.ml

view details

Gaëtan Gilbert

commit sha 73789bcfa595fc87055fb8194311545b5d54be4a

Handle default locality at attribute time Fix #14704

view details

push time in 3 days

pull request commentjscoq/addon-mathcomp-extra

Adding more extras

none of them have dune build files

Ok thanks, I'll add them as PRs

CohenCyril

comment created time in 3 days

GollumEvent
GollumEvent

pull request commentjscoq/addon-mathcomp-extra

Adding more extras

Do you know if they have Dune build files? This would make integration very easy.

I can write the dune files, where's the repos?

CohenCyril

comment created time in 3 days

GollumEvent
GollumEvent
PullRequestEvent

issue commentcoq/coq

VSCoq and Coq master

Yeah we can revert the coqtail overlay, that maybe is easier ?

mattam82

comment created time in 3 days

issue commentcoq/coq

VSCoq and Coq master

Why don't we make the new add call a different one and we keep compat for now?

I have ideas if you are interested on how to avoid this in the future

mattam82

comment created time in 3 days

push eventejgallego/pycoq

Emilio Jesus Gallego Arias

commit sha 3cf1f4cfddb914901f1ce2a487a6839dfd7500b5

[readme] Fix bug on run instructions. Closes #18

view details

push time in 4 days

issue closedejgallego/pycoq

Trouble installing

I came across this project while searching for a way to parse coq source documents using Python (there is no pre-existing way for that, is there?).

But had trouble installing. I guess it's something trivial, but couldn't figure it out.

I tried

$ opam install --deps-only coq-serapi
Nothing to do.

$ opam install pythonlib
[NOTE] Package pythonlib is already installed (current version is v0.14.0).

$ make && dune build examples/test.py && dune exec -- python3 _build/default/examples/test.py
targets {build,python,test,clean}
Traceback (most recent call last):
  File "_build/default/examples/test.py", line 1, in <module>
    import pycoq, coq
ModuleNotFoundError: No module named 'pycoq'

The project at https://pypi.org/project/pycoq/ is not same as this, right?

closed time in 4 days

ju-sh

issue commentejgallego/pycoq

Trouble installing

Hi @ju-sh , yes, you are in the right place I think! API is pretty experimental but should work for you.

Sorry about the bug, instructions were out of date, first command needs to be make install now.

[I'm not sure how to best call pip3]

The project at https://pypi.org/project/pycoq/ is not same as this, right?

Nope, that unfortunate, we named this before learning about it; tho the two projects are similar and in a sense this superseedes that one so we hope to fix the naming issue before a first release.

ju-sh

comment created time in 4 days

pull request commentcoq/coq

Load plugins with findlib

So I don't think there is nothing to worry there.

Build will fail due to the missing dep.

gares

comment created time in 5 days

pull request commentcoq/coq

Load plugins with findlib

Actually, there could be a very easy fix to dune just by adding a dep to the meta file for the library, note that in (coq.theory ... (libraries )) we already use library names.

However, this is the main problem that findlib requires the plugin to be present in the installed layout, however we will only inject the dependency for the location of the library which is usually on the build side, so I guess that's the main issue the sites loader overcomes.

gares

comment created time in 5 days

pull request commentcoq/coq

Load plugins with findlib

What's the template exactly?

https://github.com/ejgallego/coq-plugin-template

gares

comment created time in 5 days

PullRequestReviewEvent

pull request commentcoq/coq

Send debugger init even when no breakpoints are set

@coqbot: merge now

jfehrle

comment created time in 6 days

PullRequestReviewEvent

pull request commentcoq/coq

Send debugger init even when no breakpoints are set

Yup I am in the states and was lucky to see some friends, happy holidays!

Ok so let's push this then.

jfehrle

comment created time in 6 days

pull request commentcoq/coq

Fix "make test-suite" with legacy build

@ejgallego I can't reproduce that error

I still get these errors on a fresh rebase of this on master and OCaml 4.12.1

==========> TESTING bugs/closed/bug_7854.v <==========
File "./bugs/closed/bug_7854.v", line 10, characters 0-64:
Error:
Dynlink error: error loading shared library: Dynlink.Error (Dynlink.Cannot_open_dll "Failure(\"/tmp/Coq_native21ac23/Coq_native5b78f1.cmxs: failed to map segment from shared object\")")

0m0.000000s 0m0.000000s
0m0.230000s 0m0.140000s

Ummm, let me do a fresh switch opam

SkySkimmer

comment created time in 6 days

pull request commentcoq/coq

Send debugger init even when no breakpoints are set

Ping @ejgallego Hellooo out there. Anybody home?

My belly is still full of Turkey , please bear with me!

Well, I'm unsure I follow the logic here, your explanation confused me even more. As you point out, not having uniformity is not the best thing to do, now the state machine in the server side has to work even when the protocol doesn't send the "see my breakpoints" part.

The alternative fix seems actually more conservative to me that this convoluted conditional logic, but if you prefer that I guess we can go ahead.

jfehrle

comment created time in 6 days

issue commentProofGeneral/PG

Dune support

@Matafou not that I know of yet, that's mostly for object files tho so it is hard to observe.

ejgallego

comment created time in 6 days

issue commentcoq/coq

Private (or "abstract") types should not be allowed to be marhsalled

@ejgallego What should be done to resolve this?

That requires developer discussion; usually you would expect types that can be saved to disk to implement a Serializable.S interface that Coq would provide, but there are non-trivial issues to overcome. Maybe we could introduce a Marshall.S interface?

I dunno, there are many use cases here, including differential computation.

ejgallego

comment created time in 6 days

issue commentProofGeneral/PG

Dune support

merlin just strips the _build part from filesnames, it is a hack but so far it has worked well [until dune rpc is more usable I guess]

ejgallego

comment created time in 6 days

delete branch ejgallego/coq

delete branch : configure+positive_warnings

delete time in 6 days

PullRequestReviewEvent
PullRequestReviewEvent
more