- Added detection of unformed usage with `return var`, control-flow statements and member access.
- Refactored the work flow: made flow facts a class and moved operations of flow facts into the class.
- Added, cleaned and renamed test cases. Some test cases for the dynamic unformed check were suppressed by the static check. They are now added back.
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Summary:
Fixes a small compilation error where an llvm::Error variable was being
returned by copy rather than by move. The llvm::Error copy constructor
is deleted.
Co-authored-by: ergawy <kareem.ergawy@guardsquare.com>
To pick up proto updates (field renames etc.) and new carbon source samples added to testdata.
```
$ rm explorer/fuzzing/fuzzer_corpus/*
$ explorer/fuzzing/regen_corpus.py
```
This is being done because Projects v1 requires repo write access, a serious limitation for letting people use it. Projects v2 isn't a great option because it lacks event support. Labels are pretty stable in GitHub, so this switches to that.
Note this assumes we're fine renaming "decision: accepted" -> "proposal accepted", etc. There are two reasons for this:
1) To make it clear that this is a proposal-specific label, versus something like an issue for leads label.
2) Removing the colon because it was causing trouble with yaml syntax.
This also adds the "proposal draft" label, mainly to complete the taxonomy.
I was considering whether this should be a proposal itself, but it feels like maybe it's not necessary because it's a fairly low-key infrastructure change, and I'm not sure how much people were relying on the project board anyways.
I tested this in a personal repo, basically just poking at https://github.com/jonmeow/test/pull/2
Also references were added to clarify that some features are not provisional.
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
Co-authored-by: Geoff Romer <gromer@google.com>
Note, partly writing this down to make sure everyone's in agreement with what's written.
Co-authored-by: Geoff Romer <gromer@google.com>
Co-authored-by: Chandler Carruth <chandlerc@gmail.com>
MAP_POPULATE is a Linux mmap flag that optionally assists read-ahead on
the mapping. It is nonstandard and omission does not affect the
underlying mapping, so fall back to not including the flag, rather than
the reverse.
Part of #1902Fixes#1904
Regarding #1904, using #access-requests because I've decided I feel weird using any other channel for this (including #proposal-prs-and-process)
This will allow us to automatically create and maintain specific repos
based on the main repository here, pulling key files like the license
and other infrastructure and pushing them systematically to a narrow
repo. Things like editor plugins that are best packaged and installed
from a separate repos can still be developed in a central place, even
potentially sharing common things like grammars where useful.
Currently this will maintain a Vim plugin repository out of the
`utils/vim` directory, but can be easily expanded for other systems.
Trying to do this without affecting too much. I think we've generally considered this to be part of "refactoring tools", since large-scale refactoring is assisted by automatic formatting, but formatting is visible enough that it's probably worth addressing directly.
Fixes https://github.com/carbon-language/carbon-lang/issues/1656
The most significant change here is that explorer now uses the chosen spelling
rather than the old `Bool` spelling. Also update a few documentation examples
and some skeletal design docs to use the chosen spelling.
The toolchain embeds the assumption that `clang` and `llvm-ar` are
adjacent, which may not be true on all host platforms. Moreover,
in #1842 and #1843, we test if Homebrow LLVM is in `PATH` by checking
that `llvm-ar` is adjacent to `clang` and if it isn't, we `fail()` the
build, even if `llvm-ar` *is* in `PATH`.
Instead, actually check `PATH` with `repository_ctx.which`. This however
necessitates the assumption that `llvm-ar` and other LLVM binutils
are adjacent, and subsequently that `clang` and `ld.lld` are adjacent.
It appears that we don't seem to always be using these tools, but we
should avoid embedding wrong assumptions regardless.
Update docs to reflect this change.
- An introprocedural forward analysis that checks the may-be-formed states on local variables.
- Returns compilation error on usage of must-be-unformed variables.
- Implemented as a pass of `ASTNode` traversal.
- Currently supports detection of: function parameter, return expression and rhs of assign.
Provides a first implementation iteration towards class prefix and extension, along with user-friendly error regarding missing implementation for #1881
**Explorer behavior**
For class prefix `base` and `abstract`:
```
Class prefixes `base` and `abstract` are not supported yet
```
For extension with `extends`:
```
Class extension with `extends` is not supported yet
```
**Motivation**
* Provides user-friendly error for these unsupported features
* First implementation increment in supporting class prefix and extension.
We shouldn't allow insults even when they're directed against people who aren't "participants", or when they're directed against ideas (because that indirectly attacks the people holding those ideas).
Auto-update tries to add `CHECK: Nice!` or `CHECK: HALLO WELT` to the test case. As what's being printed is non-deterministic, the auto-update should be disabled.
The original sentence was grammatically unsound, specified the three options of copy, move or pointer binding in parenthesis and some words around it.
Co-authored-by: Jon Ross-Perkins <jperkins@google.com>
pseyfert
Zenong Zhang
Adrien Leravat
Kareem Ergawyandergawy
Enzo Venturi
pk19604014
Geoff Romer
Krishna Agarwal
Dennis Onyeka
Asmi AgarwalandJon Ross-Perkins
3405691582
cabmeurer
Metarsit
Shakeel MahateandJon Ross-Perkins
MarchMore