Right now if you have concurrent builds of two libraries in the same directory
(such as rustc's bootstrapping process), it's possible that two libraries will
stomp over each others' metadata, producing corrupt rlibs.
By placing the metadata file in a tempdir we're guranteed to not conflict with
ay other builds happening concurrently. Normally this isn't a problem because
output filenames are scoped to the name of the crate, but metadata is special in
that it has the same name across all crates.
I personally do not have huge amounts of experience in this area, so there's likely a thing or two wrong around the edges. I tried to just copy what libuv is doing as closely as possible with a few tweaks in a few places, but all of the `std::io::net::udp` tests are now run in both native and green settings so the published functionality is all being tested.
This PR adds `std::unsafe::intrinsics::{volatile_load,volatile_store}`, which map to LLVM's `load volatile` and `store volatile` operations correspondingly.
This would fix#11172.
I have addressed several uncertainties with this PR in the line comments.
The comments have more information as to why this is done, but the basic idea is
that finding an exported trait is actually a fairly difficult problem. The true
answer lies in whether a trait is ever referenced from another exported method,
and right now this kind of analysis doesn't exist, so the conservative answer of
"yes" is always returned to answer whether a trait is exported.
Closes#11224Closes#11225
The `--dep-info` command line option allows a nice way to generate make-style dependencies, but it currently does so alongside building of the output binary. This isn't a problem for make, as it mixes dependency graph generation and actual building, but it is problematic for other tools (e.g. CMake) which keep them separate.
To play more nicely with those tools, I've moved the --dep-info output from phase 6 (linking) up to after phase 2 (expansion of macros). Also, since there was no prior option to do so, I added a command line switch (`--no-analysis`) to stop compilation just before phase 3 (type-checking) which speeds this up even further.
Here's the beginning of a CMake function which is enabled by this change:
~~~cmake
function(get_rust_deps root_file out_var)
execute_process(COMMAND rustc ${RUSTC_FLAGS} --no-analysis --dep-info "${CMAKE_BINARY_DIR}/.deps" "${root_file}")
# Read and parse the dependency information
file(READ "${CMAKE_BINARY_DIR}/.deps" crate_deps)
file(REMOVE "${CMAKE_BINARY_DIR}/.deps")
# parsing follows...
~~~
forward-to-word is undefined, and so Emacs would throw errors in
rust-align-to-expr-after-brace. This change yields the expected
behavior discussed in the issue.
Right now on linux, an empty executable with LTO still depends on librt becaues
of the clock_gettime function in rust_builtin.o, but this commit moves this
dependency into a rust function which is subject to elimination via LTO.
At the same time, this also drops libstd's dependency on librt on unices that
are not OSX because the library is only used by extra::time (and now the
dependency is listed in that module instead).
I'd really like to be able to do something like
struct MapChain<'next, K, V> {
info: BlockInfo,
map: HashMap<K, V>,
next: Option<&'next mut MapChain<'next, K, V>
}
but I can't get the lifetimes to work out.
Right now on linux, an empty executable with LTO still depends on librt becaues
of the clock_gettime function in rust_builtin.o, but this commit moves this
dependency into a rust function which is subject to elimination via LTO.
At the same time, this also drops libstd's dependency on librt on unices that
are not OSX because the library is only used by extra::time (and now the
dependency is listed in that module instead).
Right now if you have concurrent builds of two libraries in the same directory
(such as rustc's bootstrapping process), it's possible that two libraries will
stomp over each others' metadata, producing corrupt rlibs.
By placing the metadata file in a tempdir we're guranteed to not conflict with
ay other builds happening concurrently. Normally this isn't a problem because
output filenames are scoped to the name of the crate, but metadata is special in
that it has the same name across all crates.
Currently any line starting with `#` is filtered from the output,
including line like `#[deriving]`; this patch makes it so lines are only
filtered when followed by a space similar to the current behaviour of
the tutorial/manual tester.
* Pass `&ExtCtxt` instead of `@ExtCtxt`.
* Stop passing duplicate parameters around in `expand`.
* Make `ast_fold` methods take `&mut self`.
After these, it should be possible to remove the `@mut` boxes from `ExtCtxt` altogether, though #11167 is doing some of that so I'm holding off on that for now. This will probably conflict with that PR, so I'm guessing that one will have to be rebased on top of the other.
r? @pcwalton
