Remove __extensions__ from method symbols as well as the meth_XXX. The XXX is
now used to append a few characters at the end of the name of the symbol.
Closes#6602
As with the previous commit, this is targeted at removing the possibility of
collisions between statics. The main use case here is when there's a
type-parametric function with an inner static that's compiled as a library.
Before this commit, any impl would generate a path item of "__extensions__".
This changes this identifier to be a "pretty name", which is either the last
element of the path of the trait implemented or the last element of the type's
path that's being implemented. That doesn't quite cut it though, so the (trait,
type) pair is hashed and again used to append information to the symbol.
Essentially, __extensions__ was removed for something nicer for debugging, and
then some more information was added to symbol name by including a hash of the
trait being implemented and type it's being implemented for. This should prevent
colliding names for inner statics in regular functions with similar names.
Before, the path name for all items defined in methods of traits and impls never
took into account the name of the method. This meant that if you had two statics
of the same name in two different methods the statics would end up having the
same symbol named (even after mangling) because the path components leading to
the symbol were exactly the same (just __extensions__ and the static name).
It turns out that if you add the symbol "A" twice to LLVM, it automatically
makes the second one "A1" instead of "A". What this meant is that in local crate
compilations we never found this bug. Even across crates, this was never a
problem. The problem arises when you have generic methods that don't get
generated at compile-time of a library. If the statics were re-added to LLVM by
a client crate of a library in a different order, you would reference different
constants (the integer suffixes wouldn't be guaranteed to be the same).
This fixes the problem by adding the method name to symbol path when building
the ast_map. In doing so, two symbols in two different methods are disambiguated
against.
I think this fixes 8898. I'm a bit unsure about how slices and owned pointers to vectors-of-unit already worked, but since they were already working, I didn't mess with them.
In this commit I:
- removed unneeded heap allocations
- added extra whitespace to crowded expressions
- and removed unneeded syntax
Also, CC @bblum although this change is fairly unobjectionable.
Summary:
-removed "ne" methods in libstd and librustpkg
-made default "ne" be inlined
-made one of the "eq" methods in librustpkg follow more standard parameter naming convention
The only changes to the default passes is that O1 now doesn't run the inline
pass, just always-inline with lifetime intrinsics. O2 also now has a threshold
of 225 instead of 275. Otherwise the default passes being run is the same.
I've also added a few more options for configuring the pass pipeline. Namely you
can now specify arguments to LLVM directly via the `--llvm-args` command line
option which operates similarly to `--passes`. I also added the ability to turn
off pre-population of the pass manager in case you want to run *only* your own
passes.
I would consider this as closing #8890. I don't think that we should change the default inlining threshold because LLVM/clang will probably have chosen those numbers more carefully than we would. Regardless, here's the performance numbers from this commit:
```
$ ./x86_64-apple-darwin/stage0/bin/rustc ./gistfile1.rs --test --opt-level=3 -o before
warning: no debug symbols in executable (-arch x86_64)
$ ./before --bench
running 1 test
test bench::aes_bench_x8 ... bench: 1602 ns/iter (+/- 66) = 7990 MB/s
test result: ok. 0 passed; 0 failed; 0 ignored; 1 measured
$ ./x86_64-apple-darwin/stage1/bin/rustc ./gistfile1.rs --test --opt-level=3 -o after
warning: no debug symbols in executable (-arch x86_64)
$ ./after --bench
running 1 test
test bench::aes_bench_x8 ... bench: 2103 ns/iter (+/- 175) = 6086 MB/s
test result: ok. 0 passed; 0 failed; 0 ignored; 1 measured
$ ./x86_64-apple-darwin/stage1/bin/rustc ./gistfile1.rs --test --opt-level=3 -o after --llvm-args '-inline-threshold=225'
warning: no debug symbols in executable (-arch x86_64)
$ ./after --bench
running 1 test
test bench::aes_bench_x8 ... bench: 1600 ns/iter (+/- 71) = 8000 MB/s
test result: ok. 0 passed; 0 failed; 0 ignored; 1 measured
```
The only changes to the default passes is that O1 now doesn't run the inline
pass, just always-inline with lifetime intrinsics. O2 also now has a threshold
of 225 instead of 275. Otherwise the default passes being run is the same.
I've also added a few more options for configuring the pass pipeline. Namely you
can now specify arguments to LLVM directly via the `--llvm-args` command line
option which operates similarly to `--passes`. I also added the ability to turn
off pre-population of the pass manager in case you want to run *only* your own
passes.
r? @brson
@metajack requested the ability to violate the "only workspaces can be in the RUST_PATH" rule for the purpose of bootstrapping Servo without having to restructure all the directories. This patch gives rustpkg the ability to find sources if a directory in the RUST_PATH directly contains one of rustpkg's "special" files (lib.rs, main.rs, bench.rs, or test.rs), even if it's not a workspace. In this case, it puts the build artifacts in the first workspace in the RUST_PATH.
I'm not sure whether or not it's a good idea to keep this feature in rustpkg permanently. Thus, I added a flag, ```--use-rust-path-hack```, and only enabled it when the flag is set.
This commit adds a rustpkg flag, --rust-path-hack, that allows
rustpkg to *search* inside package directories if they appear in
the RUST_PATH, while *building* libraries and executables into a
different target directory.
This behavior is hidden behind a flag because I believe we only
want to support it temporarily, to make it easier to port servo to
rustpkg.
This commit also includes a fix for how rustpkg fetches sources
from git repositories -- it uses a temporary directory as the target
when invoking `git clone`, then moves that directory into the workspace
if the clone was successful. (The old behavior was that when the
`git clone` failed, the empty target directory would be left lying
around anyway.)
