Ctypes: adds a stop-gap measure deps fields#5346
Conversation
025978d to
4b9df71
Compare
|
There is also a fix for allowing the use of |
I dit not look in depth at the new But one thing we should probably be careful about is, when building foreign stubs in C++, to have the flags Initial issue: #4846 |
4b9df71 to
7f9731b
Compare
|
Thank you for the confirmation that there is nothing to add for C. I think that Ctypes only create C files (that don't use C++ datastructures) so even if one create a stub for a C++ library I think we don't need to do anything differently than for other libraries . The C++ library is a library dependency like gmp could be one. |
christinerose
left a comment
There was a problem hiding this comment.
Just a few minor things in syntax, a spelling error, and suggested formatting for ctypes library name.
|
NTS: add |
christinerose
left a comment
There was a problem hiding this comment.
Here are a few minor suggestions for subject/verb agreement and fixing a typo. Hope it helps!
4c9fdc4 to
eec0f4e
Compare
eec0f4e to
72e53b3
Compare
christinerose
left a comment
There was a problem hiding this comment.
Just a few minor syntax / verb agreement suggestions
|
@mbacarella can you review this MR? |
lgtm aside from one question: why do you need to instantiate and pass around a sandbox to support this deps feature? Shouldn't whatever you depend on get surfaced from the sandbox if it's listed as a target? |
Signed-off-by: François Bobot <francois.bobot@cea.fr>
before writing them to sexp Signed-off-by: François Bobot <francois.bobot@cea.fr>
Signed-off-by: François Bobot <francois.bobot@cea.fr>
Co-authored-by: Christine Rose <christinerose@users.noreply.github.com>
Co-authored-by: Christine Rose <christinerose@users.noreply.github.com>
Co-authored-by: Christine Rose <christinerose@users.noreply.github.com>
Co-authored-by: Christine Rose <christinerose@users.noreply.github.com>
9d00eaf to
5c25ecf
Compare
|
The passing around of the sandbox is just because the deps field generally allows to specify the kind of sandbox one desire. |
…ne-site, dune-rpc, dune-rpc-lwt, dune-private-libs, dune-glob, dune-configurator, dune-build-info and dune-action-plugin (3.1.0) CHANGES: - Add `sourcehut` as an option for defining project sources in dune-project files. For example, `(source (sourcehut user/repo))`. (ocaml/dune#5564, @rgrinberg) - Add `dune coq top` command for running a Coq toplevel (ocaml/dune#5457, @rlepigre) - Fix dune exec dumping database in wrong directory (ocaml/dune#5544, @bobot) - Always output absolute paths for locations in RPC reported diagnostics (ocaml/dune#5539, @rgrinberg) - Add `(deps <deps>)` in ctype field (ocaml/dune#5346, @bobot) - Add `(include <file>)` constructor to dependency specifications. This can be used to introduce dynamic dependencies (ocaml/dune#5442, @anmonteiro) - Ensure that `dune describe` computes a transitively closed set of libraries (ocaml/dune#5395, @esope) - Add direct dependencies to $ dune describe output (ocaml/dune#5412, @esope) - Show auto-detected concurrency on Windows too (ocaml/dune#5502, @MisterDA) - Fix operations that remove folders with absolute path. This happens when using esy (ocaml/dune#5507, @EduardoRFS) - Dune will not fail if some directories are non-empty when uninstalling. (ocaml/dune#5543, fixes ocaml/dune#5542, @nojb) - `coqdep` now depends only on the filesystem layout of the .v files, and not on their contents (ocaml/dune#5547, helps with ocaml/dune#5100, @ejgallego) - The mdx stanza 0.2 can now be used with `(implicit_transitive_deps false)` (ocaml/dune#5558, fixes ocaml/dune#5499, @emillon) - Fix missing parenthesis in printing of corresponding terminal command for `(with-outputs-to )` (ocaml/dune#5551, fixes ocaml/dune#5546, @Alizter)
The deps fields added to the ctypes sub-stanza, is not meant to be in a final version of the ctypes feature. But I think it could help to experiment with it by fixing missing dependencies. Cf #5325
There is also a dirty hack for #5190 so that I can test it.
CC @mbacarella