I'm treating my PR to mypy bazel integration as dead in the water. Abandoning that approach, this seems to work reasonably well as a replacement. It's actually a slightly newer version of mypy too -- the bazel integration was stuck on an old version.
Note I don't think we're losing much: we shouldn't be writing too much python.
I'm doing this to avoid macro name conflicts, following https://google.github.io/styleguide/cppguide.html#Preprocessor_Macros: "If you do export a macro from a header, it must have a globally unique name. To achieve this, it must be named with a prefix consisting of your project's namespace name (but upper case)."
Commands run:
```
sed -i 's/\(DCHECK\|CHECK\|FATAL\|MAKE_UNIQUE_NAME\|MAKE_UNIQUE_NAME_IMPL\|RAW_EXITING_STREAM\|RETURN_IF_ERROR\|RETURN_IF_ERROR_IMPL\|ASSIGN_OR_RETURN\|ASSIGN_OR_RETURN_IMPL\|DIAGNOSTIC_KIND\|RETURN_IF_STACK_LIMITED\)(/CARBON_\1(/g' $(git ls-files *.cpp *.h *.lpp *.ypp *.def ':!third_party')
sed -i 's/#undef DIAGNOSTIC_KIND/#undef CARBON_DIAGNOSTIC_KIND/' toolchain/diagnostics/diagnostic_registry.def
```
Note this isn't *quite* everything, but it's intended to be a large pass at everything:
```
╚╡git grep '#define ' *.cpp *.h *.lpp *.ypp *.def ':!third_party' | grep -v '#define CARBON' | grep -v _H_
explorer/syntax/lexer.lpp: #define YY_USER_ACTION \
explorer/syntax/lexer.lpp: #define SIMPLE_TOKEN(name) \
explorer/syntax/lexer.lpp: #define ARG_TOKEN(name, arg) \
explorer/syntax/parse_and_lex_context.h:#define YY_DECL \
migrate_cpp/cpp_refactoring/var_decl.cpp:#define ABSTRACT_TYPE(Class, Base)
migrate_cpp/cpp_refactoring/var_decl.cpp:#define TYPE(Class, Base) \
```
We may in particular want to do a pass to clean up #ifdef guards and make them be CARBON_ rooted.
- Makes `Self` a keyword
- Inside `class Foo { `... `}`, `Self` means `Foo`
- Inside `class Foo(T:! Type, U:! Type) {` ...` }`, `Self` means `Foo(T, U)`
- `impl as Bar` means `impl Self as Bar`
- Inside `external impl Foo as Bar {`...`}`, not in a scope already defining `Self`, `Self` means `Foo`
Implemented by introducing a new kind of declaration, a `SelfDeclaration`, that is automatically added to class and impl declarations.
Co-authored-by: Jon Meow <jperkins@google.com>
This proposal:
- Adds support for interfaces requiring other types than `Self` to implement interfaces, as in:
```
interface IntLike {
impl i32 as As(Self);
// ...
}
```
- Defines requirements on how to satisfy those requirements that have a `where` clause.
- Extends `observe` declarations to include saying a type implements an interface, so code can provide a proof instead of the compiler having to perform a recursive search.
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Previously we modeled these as being class types or interface types, with special case checks to treat them as not actually being class or interface types.
A rephrasing was suggested on https://github.com/carbon-language/carbon-lang/pull/1188#discussion_r851416370 because "explorer" alone could be hard to parse; I'm suggesting just replacing with "Carbon explorer" to get precision without changing phrasing, as well as rephrasing "Executable semantic" as "Carbon explorer" (which... the former *may* have meant "executable semantics", but the difference in pluralization made it vague, so I'm not sure this is right -- but it also didn't explain _where_, so "executable semantics" seems the right reading)
Co-authored-by: Geoff Romer <gromer@google.com>
Encourage reviewers to merge when they feel okay doing so. Let reviewers make
that choice. Let authors say they'll merge themselves.
Co-authored-by: Chandler Carruth <chandlerc@gmail.com>
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Remove files accidentally duplicated in e85f45de10 between explorer/ and
executable_semantics/. Each removed file already has a corresponding file in explorer/.
Use FileCheck's `[[@LINE+n]]` mechanism to refer to the next line.
This is made awkward by a couple of things:
* We want to keep the `CHECK` lines in the original order.
* Errors are sometimes more than one line long.
The approach we use is to interleave the original lines and the check lines, putting each check line as early as possible subject to two rules:
1) Check lines never precede the 'AUTOUPDATE' line
2) Except when required by rule (1), a check line that refers to a source line by line number is never placed earlier than a source line that precedes that source line.
The actual `[[@LINE+n]]` annotations are created in a second pass after we've interleaved the lines so that we can work out the correct offsets.
Co-authored-by: Jon Meow <jperkins@google.com>
Add concrete design for interfaces for comparison.
Rename interfaces for arithmetic following current thinking in #1058.
Update rules for mixed-type comparisons for data classes following #710.
Co-authored-by: Chandler Carruth <chandlerc@gmail.com>
Right now, the text is disallowing appeals to logic (which are persuasive methods, per the linked wikipedia article). Consensus seems to be that this is unintentional.
Co-authored-by: Chandler Carruth <chandlerc@gmail.com>
The logic is using bazel Runfiles API per Jon's suggestion.
One of the scenarios used by the fuzzing framework is to place the binary and its runfiles into a folder, resulting in a directory layout like this:
[temp dir]
fuzzer
fuzzer.runfiles/
Tested:
bazel test -c opt explorer/fuzzing:explorer_fuzzer
bazel build -c opt explorer/fuzzing:explorer_fuzzer + run the binary from various locations
To make this work, introduce impl bindings into scope immediately when type-checking. We may rely on their existence before we finish type-checking the enclosing pattern.
This proposal has three main contributions:
- Types with generic parameters have an identity that consists of the types names plus the values of those parameters.
- The parameters of a type may be deduced from a function's argument.
- Types with generic parameters do not support specialization. Instead, a type can delegate to an interface to opt in to allowing specific customization points.
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Before, if the llvm hook to print a stack trace on CHECK was bound, it would trace after the message rather than before the message. I'm thinking swapping the order will generally be easier to debug.
* check that the return type is a type, add more implicit conversions
* update IsConcreteType for generic classes, add IsType predicate
* delete a comment
* trivial change
* move Witness wrt IsConcreteType and IsType
* added comments to tests
* update error line numbers
This is a workaround for #1208, allowing to remove manual tag to allow building and testing fuzzer code during precommits and in bazel test ...
Tested:
bazel test -k ...
bazel build -c opt executable_semantics/fuzzing:executable_semantics_fuzzer
bazel-bin/executable_semantics/fuzzing/executable_semantics_fuzzer
Given repeated concerns about C++ interop priorities: taking a pass at being more direct about priorities and tradeoffs.
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
I went down this route because I figured the installer binary should be in `check_deps`. The result:
- Add thin wrapper scripts for `buildozer` and `bazel`, similar to the `buildifier` wrapper that's already there.
- These are mainly to avoid deduplicating logic, and dodging intricacies of Python imports that I don't understand well through instead subprocessing.
- Fix update_roots.py (it was missing a `.parent`, must've been relocated without a re-run at some point)
- Fix fuzzer proto code to be `testonly`
- Update the `check_deps` targets
Breaks `--trace` into two flags:
- `--parser_debug`, which sets the parser debug level (which I haven't dug into piping with `--trace`, but seemed likely to be troublesome)
- `--trace_file`, which now the type checker and interpreter will use for trace information (note compile errors should use a different channel)
Most of the file edits are just mechanical testdata flag updates: `sed -i 's/--trace/--parser_debug --trace_file=-/' testdata/**/*.carbon`
To explain the output paths:
- parse/compile errors: stderr
- print() calls: stdout
- parser tracing: `--parser_debug` option, stdout (formerly stdout if `--trace`)
- type check/compile tracing: `--trace_file=<file>`, giving `-` uses stdout (formerly stdout if `--trace`)
- return code of executed Carbon code: `--trace_file` if set, stdout if not (formerly stdout always)
I think this will work on linux/mac, not windows obviously. But the essence is that running `bazel run -c opt //installers/local:install` will create:
Symlink: /usr/bin/carbon-explorer -> /usr/lib/carbon/carbon
BusyBox-style binary: /usr/lib/carbon/carbon
File: /usr/lib/carbon/data/prelude.carbon
And then just running `carbon-explorer foo.carbon` with no flags (in particular, not --prelude) will work.
Since this removes the need for prelude_file in most cases, I'm removing the relative path logic added in #1179 -- it would otherwise conflict with what I'm doing here. I *think* overall this will result in something even simpler and more reliable for compiler explorer to use, and hopefully with enough flexibility that it's easy for us to change how it's implemented later without breaking much.
Allow interfaces and implementations to be forward declared.
```
// Forward declare interface `F`
interface F;
class C {
// Forward declare `C` implements `F`
impl as F;
}
// Definitions corresponding to forward declarations
interface F { ... }
impl C as F { ... }
```
To allow members of interfaces with default definitions to be forward declared, prefix them with the keyword `default`, following #1082.
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
Co-authored-by: Chandler Carruth <chandlerc@gmail.com>
Jon Meow
Richard Smith
josh11bandJon Meow
pk19604014
Jeremy G. Siek