| Age | Commit message (Collapse) | Author |
|---|
|
TL;DR: The binding generator is a mess as of right now. At first it was funny
(in a "this is challenging" sense) to improve on it, but this is not
sustainable.
The truth is that the current architecture of the binding generator is a huge
pile of hacks, so these few days I've been working on rewriting it with a few
goals.
1) Have the hacks as contained and identified as possible. They're sometimes
needed because how clang exposes the AST, but ideally those hacks are well
identified and don't interact randomly with each others.
As an example, in the current bindgen when scanning the parameters of a
function that references a struct clones all the struct information, then if
the struct name changes (because we mangle it), everything breaks.
2) Support extending the bindgen output without having to deal with clang. The
way I'm aiming to do this is separating completely the parsing stage from
the code generation one, and providing a single id for each item the binding
generator provides.
3) No more random mutation of the internal representation from anywhere. That
means no more Rc<RefCell<T>>, no more random circular references, no more
borrow_state... nothing.
4) No more deduplication of declarations before code generation.
Current bindgen has a stage, called `tag_dup_decl`[1], that takes care of
deduplicating declarations. That's completely buggy, and for C++ it's a
complete mess, since we YOLO modify the world.
I've managed to take rid of this using the clang canonical declaration, and
the definition, to avoid scanning any type/item twice.
5) Code generation should not modify any internal data structure. It can lookup
things, traverse whatever it needs, but not modifying randomly.
6) Each item should have a canonical name, and a single source of mangling
logic, and that should be computed from the inmutable state, at code
generation.
I've put a few canonical_name stuff in the code generation phase, but it's
still not complete, and should change if I implement namespaces.
Improvements pending until this can land:
1) Add support for missing core stuff, mainly generating functions (note that
we parse the signatures for types correctly though), bitfields, generating
C++ methods.
2) Add support for the necessary features that were added to work around some
C++ pitfalls, like opaque types, etc...
3) Add support for the sugar that Manish added recently.
4) Optionally (and I guess this can land without it, because basically nobody
uses it since it's so buggy), bring back namespace support.
These are not completely trivial, but I think I can do them quite easily with
the current architecture.
I'm putting the current state of affairs here as a request for comments... Any
thoughts? Note that there are still a few smells I want to eventually
re-redesign, like the ParseError::Recurse thing, but until that happens I'm
way happier with this kind of architecture.
I'm keeping the old `parser.rs` and `gen.rs` in tree just for reference while I
code, but they will go away.
[1]: https://github.com/Yamakaky/rust-bindgen/blob/master/src/gen.rs#L448
|
|
Skip -isystem stuff if --target is specified, and don't strip leading _ on Windows
The -isystem stuff fixes issue #53.
Stripping leading _ on Windows is weird; it needs to not be stripped on MSVC (leading @), but on gcc at one point it needed to be. But now it no longer does. I don't know. 'cargo test' succeeds with mozjs bindings generated with these fixes.
|
|
|
|
|
|
Deduplicate names of virtual overloaded methods. Fix #48
r? @emilio
|
|
|
|
Append the input header to the end of the clang args, instead of the front
r @emilio
|
|
|
|
Use docopt for argument parsing
This commit switches bindgen over to using the docopt crate for argument parsing
instead of manual argument parsing. This required two notable changes in the
arguments and flags style:
1. All flags of the form `-foo` are now of the form `--foo`.
2. We can no longer pass unknown flags straight through to clang. Instead, the
user appends `--` after the bindgen flags and input header, after which point
any more flags and arguments get collected and passed to clang. This required
changes to the test runner and the `// bindgen-flags` comments.
r? @emilio
|
|
This commit switches bindgen over to using the docopt crate for argument parsing
instead of manual argument parsing. This required two notable changes in the
arguments and flags style:
1. All flags of the form `-foo` are now of the form `--foo`.
2. We can no longer pass unknown flags straight through to clang. Instead, the
user appends `--` after the bindgen flags and input header, after which point
any more flags and arguments get collected and passed to clang. This required
changes to the test runner and the `// bindgen-flags` comments.
|
|
Add description of how to run the tests to README.md
Fixes #42
r? @emilio
|
|
Fixes #42
|
|
Expand the build instructions when using llvm 3.9
r? @emilio
|
|
|
|
|
|
Allow blacklist typedef
This allows making type aliases to be blacklisted.
|
|
|
|
Unions with destructors.
r? @Manishearth
|
|
|
|
This is kind of unfortunate, but we weren't taking into account explicit
destructors.
|
|
Add support for basic template unions.
Fixes #28.
r? @nox
cc @bholley
|
|
down the hack version.
|
|
by Rc<T>s
|
|
test: Add test for keywords from upstream.
This is the only test that's missing from upstream apart from the `va_list` one, that produces similar output but doesn't compile in any of the versions.
|
|
|
|
|
|
|
|
parameters in general that don't use the template argument.
|
|
since I introduced support for anonymous structs.
|
|
Add ignore_methods to bindgen
I'm defining a struct inline in ServoBindings.h via a macro (for the already_addrefed stuff), and I need to forbid its ignore_methods from getting generated without forbidding functions as well.
r? @bholley
|
|
|
|
Add support for 'unsafe fields'
Needed for the next step of https://github.com/servo/servo/pull/12521
These are fields which are private but have unsafe accessor functions. Since we generate bindings in a separate module, we can't touch private fields from other parts of the module (an alternative is to inject a footer with these private impls). `pub(restricted)` exists, but is not stable.
r? @emilio
|
|
|
|
|
|
|
|
Avoid recursing infinitely with CRTP while looking for a destructor.
Fixes #22
r? @nox
|
|
Fixes #22
|
|
Bump clang-sys version to 0.8.0
To include the fix for Windows build.
|
|
To include the fix for Windows build.
|
|
Add support for generating MSVC bindings (using clang)
This adds support for generating MSVC bindings, using clang's MSVC C++ ABI support. It requires a command line flag for selecting this binding due to an inability to pick up the msvc target triple.
|
|
method
|
|
There's no easy way to get the current target triple from the clang-c
bindings. Make this simple and require an option to be passed when
generating bindings for msvc, using the x86_64-pc-win32 target (or i686)
|
|
bindings
|
|
Some fixes to make it work with MSVC
The last commit -- removing clang searching code, is not closely related. The reason its being there is that the C++ headers we need in geckolib require LLVM 3.9, which is pre-release, and thus it is usually necessary to setting `LIBCLANG_PATH` manually to get the proper clang library linked.
However, the current searching code doesn't work well on Windows, because LLVM for Windows puts `libclang.dll` and `libclang.lib` in different directories, while you can only specify one of them in `LIBCLANG_PATH`.
To fix this, I submitted some code in KyleMayes/clang-sys#34 (landed in development branch), and I don't want the searching code here to introduce any conflicts with the code there.
|
|
The dependency clang-sys has been working hard to find and link with
LLVM and Clang. It doesn't make sense to duplicate that work. Also,
there could be undesired conflict if the searching code doesn't match
exactly, and different libraries are found.
|
|
This is not needed at least for Clang 3.9 (which is required to parse
stdlib headers of MSVC 2015). This hack leads to incorrect mangling
result there.
|
|
Otherwise it would be hard to write uniform blacklist header list
across platforms, especially given that on Windows, some part of
path could be using '/' while others use '\'.
|
|
Allow building bindgen with stable rust.
Fixes #13
r? @nox
cc @bholley
|
|
|
|
|
