Alex Crichton [Mon, 10 Mar 2014 22:10:56 +0000 (15:10 -0700)]
green: Fix a scheduler assertion on yielding
This commit fixes a small bug in the green scheduler where a scheduler task
calling `maybe_yield` would trip the assertion that `self.yield_check_count > 0`
This behavior was seen when a scheduler task was scheduled many times
successively, sending messages in a loop (via the channel `send` method), which
in turn invokes `maybe_yield`. Yielding on a sched task doesn't make sense
because as soon as it's done it will implicitly do a yield, and for this reason
the yield check is just skipped if it's a sched task.
I am unable to create a reliable test for this behavior, as there's no direct
way to have control over the scheduler tasks.
cc #12666, I discovered this when investigating that issue
Alex Crichton [Mon, 10 Mar 2014 06:20:05 +0000 (23:20 -0700)]
std: Relax an assertion in oneshot selection
The assertion was erroneously ensuring that there was no data on the port when
the port had selection aborted on it. This assertion was written in error
because it's possible for data to be waiting on a port, even after it was
disconnected. When aborting selection, if we see that there's data on the port,
then we return true that data is available on the port.
bors [Wed, 12 Mar 2014 13:27:03 +0000 (06:27 -0700)]
auto merge of #12807 : pnkfelix/rust/fsk-issue5121-fns-with-early-lifetime-params, r=pnkfelix
Fix issue #5121: Add proper support for early/late distinction for lifetime bindings.
There are some little refactoring cleanups as separate commits; the real meat that has the actual fix is in the final commit.
The original author of the work was @nikomatsakis; I have reviewed it, revised it slightly, refactored it into these separate commits, and done some rebasing work.
Changed lists of lifetimes in ast and ty to use Vec instead of OptVec.
There is a broader revision (that does this across the board) pending
in #12675, but that is awaiting the arrival of more data (to decide
whether to keep OptVec alive by using a non-Vec internally).
For this code, the representation of lifetime lists needs to be the
same in both ScopeChain and in the ast and ty structures. So it
seemed cleanest to just use `vec_ng::Vec`, now that it has a cheaper
empty representation than the current `vec` code.
bors [Wed, 12 Mar 2014 03:51:56 +0000 (20:51 -0700)]
auto merge of #12774 : alexcrichton/rust/proc-bounds, r=pcwalton
This is needed to make progress on #10296 as the default bounds will no longer
include Send. I believe that this was the originally intended syntax for procs,
and it just hasn't been necessary up until now.
bors [Wed, 12 Mar 2014 02:31:57 +0000 (19:31 -0700)]
auto merge of #12650 : huonw/rust/librand, r=alexcrichton
Move std::rand to a separate rand crate
This functionality is not super-core and so doesn't need to be included
in std. It's possible that std may need rand (it does a little bit now,
for io::test) in which case the functionality required could be moved to
a secret hidden module and reexposed by librand.
Unfortunately, using #[deprecated] here is hard: there's too much to
mock to make it feasible, since we have to ensure that programs still
typecheck to reach the linting phase.
Also, deprecates/removes `rand::rng` (this time using `#[deprecated]`), since it's too easy to accidentally use inside a loop, making things very slow (have to read randomness from the OS and seed the RNG each time.)
Alex Crichton [Sun, 9 Mar 2014 02:21:01 +0000 (18:21 -0800)]
syntax: Add support for trait bounds on procs
This is needed to make progress on #10296 as the default bounds will no longer
include Send. I believe that this was the originally intended syntax for procs,
and it just hasn't been necessary up until now.
Huon Wilson [Sun, 2 Mar 2014 01:59:35 +0000 (12:59 +1100)]
rand: deprecate `rng`.
This should be called far less than it is because it does expensive OS
interactions and seeding of the internal RNG, `task_rng` amortises this
cost. The main problem is the name is so short and suggestive.
The direct equivalent is `StdRng::new`, which does precisely the same
thing.
The deprecation will make migrating away from the function easier.
Huon Wilson [Sat, 1 Mar 2014 13:36:33 +0000 (00:36 +1100)]
Remove the dependence of std::io::test on rand.
This replaces it with a manual "task rng" using XorShift and a crappy
seeding mechanism. Theoretically good enough for the purposes
though (unique for tests).
Huon Wilson [Sun, 2 Mar 2014 00:23:04 +0000 (11:23 +1100)]
std: Move rand to librand.
This functionality is not super-core and so doesn't need to be included
in std. It's possible that std may need rand (it does a little bit now,
for io::test) in which case the functionality required could be moved to
a secret hidden module and reexposed by librand.
Unfortunately, using #[deprecated] here is hard: there's too much to
mock to make it feasible, since we have to ensure that programs still
typecheck to reach the linting phase.
bors [Tue, 11 Mar 2014 19:36:58 +0000 (12:36 -0700)]
auto merge of #12783 : adrientetar/rust/more-docs, r=alexcrichton
- remove `node.js` dep., it has no effect as of #12747 (1)
- switch between LaTeX compilers, some cleanups
- CSS: fixup the print stylesheet, refactor highlighting code (2)
(1): `prep.js` outputs its own HTML directives, which `pandoc` cannot recognize when converting the document into LaTeX (this is why the PDF docs have never been highlighted as of now).
Note that if we were to add the `.rust` class to snippets, we could probably use pandoc's native highlighting capatibilities i.e. Kate ([here is](http://adrientetar.github.io/rust-tuts/tutorial/tutorial.pdf) an example of that).
(2): the only real highlighting change is for lifetimes which are now brown instead of red, the rest is just refactor of twos shades of red that look the same.
Also I made numbers highlighting for src in rustdoc a tint more clear so that it is less bothering.
bors [Tue, 11 Mar 2014 18:17:01 +0000 (11:17 -0700)]
auto merge of #12780 : zslayton/rust/json-nav, r=alexcrichton
This is my first non-docs contribution to Rust, so please let me know what I can fix. I probably should've submitted this to the mailing list first for comments, but it didn't take too long to implement so I figured I'd just give it a shot.
These changes are modeled loosely on the [JsonNode API](http://jackson.codehaus.org/1.7.9/javadoc/org/codehaus/jackson/JsonNode.html) provided by the [Jackson JSON processor](http://jackson.codehaus.org/).
Many common use cases for parsing JSON involve pulling one or more fields out of an object, however deeply nested. At present, this requires writing a pyramid of match statements. The added methods in this PR aim to make this a more painless process.
Accessing "successful":
```rust
let example_json : Json = from_str("...above json...").unwrap();
let was_successful: Option<bool> = example_json.find(&~"successful").and_then(|j| j.as_boolean());
```
Accessing "status":
```rust
let example_json : Json = from_str("...above json...").unwrap();
let status_code : Option<f64> = example_json.find(&~"status").and_then(|j| j.as_number());
```
Accessing "vehicles":
```rust
let example_json : Json = from_str("...above json...").unwrap();
let vehicle_list: Option<List> = example_json.search(&~"vehicles").and_then(|j| j.as_list());
```
Accessing "vehicles" with an explicit path:
```rust
let example_json : Json = from_str("...above json...").unwrap();
let vehicle_list: Option<List> = example_json.find_path(&[&~"content", &~"vehicles"]).and_then(|j| j.as_list());
```
Accessing "error", which might be null or a string:
```rust
let example_json : Json = from_str("...above json...").unwrap();
let error: Option<Json> = example_json.find(&~"error");
if error.is_null() { // This would be nicer as a match, I'm just illustrating the boolean test methods
println!("Error is null, everything's fine.");
} else if error.is_str(){
println!("Something went wrong: {}", error.as_string().unwrap());
}
```
Some notes:
* Macros would help to eliminate some of the repetitiveness of the implementation, but I couldn't use them due to #4621. (**Edit**: There is no longer repetitive impl. Methods were simplified to make them more composable.)
* Would it be better to name methods after the Json enum type (e.g. `get_string`) or the associated Rust built-in? (e.g. `get_str`)
* TreeMap requires its keys to be &~str. Because of this, all of the new methods required &~str for their parameters. I'm uncertain what the best approach to fixing this is: neither demanding an owned pointer nor allocating within the methods to appease TreeMap's find() seems desirable. If I were able to take &str, people could put together paths easily with `"foo.bar.baz".split('.').collect();` (**Edit**: Follow on investigation into making TreeMap able to search by Equiv would be worthwhile.)
* At the moment, the `find_<sometype>` methods all find the first match for the provided key and attempt to return that value if it's of the specified type. This makes sense to me, but it's possible that users would interpret a call to `find_boolean("successful")` as looking for the first "successful" item that was a boolean rather than looking for the first "successful" and returning None if it isn't boolean. (**Edit**: No longer relevant.)
I hope this is helpful. Any feedback is appreciated!
debuginfo: Improve commandline option handling for debuginfo (fixes #12811)
The `-g` flag does not take an argument anymore while the argument to `--debuginfo` becomes mandatory. This change makes it possible again to run the compiler like this:
`rustc -g ./file.rs`
This did not work before because `./file.rs` was misinterpreted as the argument to `-g`. In order to get limited debuginfo, one now has to use `--debuginfo=1`.
bors [Tue, 11 Mar 2014 16:56:57 +0000 (09:56 -0700)]
auto merge of #12556 : alexcrichton/rust/weak-linkage, r=brson
It is often convenient to have forms of weak linkage or other various types of
linkage. Sadly, just using these flavors of linkage are not compatible with
Rust's typesystem and how it considers some pointers to be non-null.
As a compromise, this commit adds support for weak linkage to external symbols,
but it requires that this is only placed on extern statics of type `*T`.
Codegen-wise, we get translations like:
// generated IR
@foo = extern_weak global i32
@_some_internal_symbol = internal global *i32 @foo
```
All references to the rust value of `foo` then reference `_some_internal_symbol`
instead of the symbol `_foo` itself. This allows us to guarantee that the
address of `foo` will never be null while the value may sometimes be null.
An example was implemented in `std::rt::thread` to determine if
`__pthread_get_minstack()` is available at runtime, and a test is checked in to
use it for a static value as well. Function pointers a little odd because you
still need to transmute the pointer value to a function pointer, but it's
thankfully better than not having this capability at all.
Thanks to @bnoordhuis for the original patch, most of this work is still his!
zslayton [Sun, 9 Mar 2014 06:30:27 +0000 (01:30 -0500)]
Added convenience methods and accompanying tests to the Json class.
Fixed some styling issues with trailing whitespace.
- Removed redundant functions.
- Renamed `get` to `find`
- Renamed `get_path` to `find_path`
- Renamed `find` to `search`
- Changed as_object and as_list to return Object and List
rather than the underlying implementation types
of TreeMap<~str,Json> and ~[Json]
- Refactored find_path to use a fold() instead of recursion
Formatting fixes.
Fixed spacing, deleted comment.
Added convenience methods and accompanying tests to the Json class.
Alex Crichton [Wed, 26 Feb 2014 00:15:10 +0000 (16:15 -0800)]
rustc: Support various flavors of linkages
It is often convenient to have forms of weak linkage or other various types of
linkage. Sadly, just using these flavors of linkage are not compatible with
Rust's typesystem and how it considers some pointers to be non-null.
As a compromise, this commit adds support for weak linkage to external symbols,
but it requires that this is only placed on extern statics of type `*T`.
Codegen-wise, we get translations like:
// generated IR
@foo = extern_weak global i32
@_some_internal_symbol = internal global *i32 @foo
All references to the rust value of `foo` then reference `_some_internal_symbol`
instead of the symbol `_foo` itself. This allows us to guarantee that the
address of `foo` will never be null while the value may sometimes be null.
An example was implemented in `std::rt::thread` to determine if
`__pthread_get_minstack()` is available at runtime, and a test is checked in to
use it for a static value as well. Function pointers a little odd because you
still need to transmute the pointer value to a function pointer, but it's
thankfully better than not having this capability at all.
Guillaume Pinot [Sat, 8 Mar 2014 15:56:07 +0000 (16:56 +0100)]
fix a bug in shootout-reverse-complement, official tests should pass with it
In the "reverse-complement" loop, if there is an odd number of element,
we forget to complement the element in the middle. For example, if the
input is "ggg", the result before the fix is "CgC" instead of "CCC".
This is because of this bug that the official shootout says that the rust
version is in "Bad Output". This commit should fix this error.
bors [Tue, 11 Mar 2014 07:32:04 +0000 (00:32 -0700)]
auto merge of #12617 : sfackler/rust/item-modifier, r=alexcrichton
Where ItemDecorator creates new items given a single item, ItemModifier
alters the tagged item in place. The expansion rules for this are a bit
weird, but I think are the most reasonable option available.
When an item is expanded, all ItemModifier attributes are stripped from
it and the item is folded through all ItemModifiers. At that point, the
process repeats until there are no ItemModifiers in the new item.
Steven Fackler [Fri, 28 Feb 2014 07:49:25 +0000 (23:49 -0800)]
Add an ItemModifier syntax extension type
Where ItemDecorator creates new items given a single item, ItemModifier
alters the tagged item in place. The expansion rules for this are a bit
weird, but I think are the most reasonable option available.
When an item is expanded, all ItemModifier attributes are stripped from
it and the item is folded through all ItemModifiers. At that point, the
process repeats until there are no ItemModifiers in the new item.
bors [Tue, 11 Mar 2014 05:42:02 +0000 (22:42 -0700)]
auto merge of #12793 : brson/rust/installer, r=alexcrichton
Work towards #9876.
Several minor things here:
* Fix the `need_ok` function in `configure`
* Install man pages with non-executable permissions
* Use the correct directory for man pages when installing (this was a recent regression)
* Put all distributables in a new `dist/` directory in the build directory (there are soon to be significantly more of these)
Finally, this also creates a new, more precise way to install and uninstall Rust's files, the `install.sh` script, and creates a build target (currently `dist-tar-bins`) that creates a binary tarball containing all the installable files, boilerplate and license docs, and `install.sh`.
This binary tarball is the lowest-common denominator way to install Rust on Unix. We'll use it as the default installer on Linux (OS X will use .pkg).
## How `install.sh` works
* First, the makefiles (`prepare.mk` and `dist.mk`) put all the stuff that needs to be installed in a new directory in `dist/`.
* Then it puts `install.sh` in that same directory and a list of all the files to install at `rustlib/manifest`.
* Then the directory can be packaged and distributed.
* When `install.sh` runs it does some sanity checking then copies everything in the manifest to the install prefix, then copies the manifest as well.
* When `install.sh` runs again in the future it first looks for the existing manifest at the install prefix, and if it exists deletes everything in it. This is how the core distribution is upgraded - cargo is responsible for the rest.
* `install.sh --uninstall` will uninstall Rust
## Future work:
* Modify `install.sh` to accept `--man-dir` etc
* Rewrite `install.mk` to delegate to `install.sh`
* Investigate how `install.sh` does or doesn't work with .pkg on Mac
* Modify `dist.mk` to create `.pkg` files for all hosts
* Possibly use [makeself](http://www.megastep.org/makeself/) to create self-extracting installers
* Modify dist-snap bots run on mac as well, uploading binary tarballs and .pkg files for the four combos of linux, mac, x86, and x86_64.
* Adjust build system to be able to augment versions with '-nightly'
* Adjust build system to name dist artifacts without version numbers e.g. `rust-nightly-...pkg`. This is so we don't leave a huge trail of old nightly binaries on S3 - they just get overwritten.
* Create new dist-nightly builder
* Give the build master a new cron job to push to dist-nightly every night
* Add docs to distributables
* Update README.md to reflect the new reality
* Modernize the website to promote new installers
bors [Tue, 11 Mar 2014 01:52:03 +0000 (18:52 -0700)]
auto merge of #12652 : rcxdude/rust/hexfloatext, r=alexcrichton
Closes #1433. Implemented after suggestion by @cmr in #12323
This is slightly less flexible than the implementation in #12323 (binary and octal floats aren't supported, nor are underscores in the literal), but is cleaner in that it doesn't modify the core grammar, or require odd syntax for the number itself. The missing features could be added back with relatively little effort (the main awkwardness is parsing the string. Is there a good approach for this in the stdlib currently?)
bors [Mon, 10 Mar 2014 20:42:05 +0000 (13:42 -0700)]
auto merge of #12733 : edwardw/rust/rw-liveness, r=nikomatsakis
- Repurposes `MoveData.assignee_ids` to mean only `=` but not `+=`, so
that borrowck effectively classifies all expressions into assignees,
uses or both.
- Removes two `span_err` in liveness analysis, which are now borrowck's
responsibilities.
bors [Mon, 10 Mar 2014 19:27:05 +0000 (12:27 -0700)]
auto merge of #12670 : dmski/rust/master, r=nick29581
CodeMap.span_to_* perform a lookup of a BytePos(sp.hi), which lands into the next filemap if the last byte of range denoted by Span is also the last byte of the filemap, which results in ICEs or incorrect error reports.
Example:
````
pub fn main() {
let mut num = 3;
let refe = &mut num;
*refe = 5;
println!("{}", num);
}````
(note the empty line in the beginning and the absence of newline at the end)
The above would have caused ICE when trying to report where "refe" borrow ends.
The above without an empty line in the beginning would have reported borrow end to be the first line.
Most probably, this is also responsible for (at least some occurrences of) issue #8256.
The issue is fixed by always adding a newline at the end of non-empty filemaps in case there isn't a new line there already.
Dmitry Promsky [Mon, 3 Mar 2014 10:44:43 +0000 (14:44 +0400)]
syntax: fixed ICEs and incorrect line nums when reporting Spans at the end of the file.
CodeMap.span_to_* perform a lookup of a BytePos(sp.hi), which lands into the next filemap if the last byte of range denoted by Span is also the last byte of the filemap, which results in ICEs or incorrect error reports.
Example:
````
pub fn main() {
let mut num = 3;
let refe = &mut num;
*refe = 5;
println!("{}", num);
}````
(note the empty line in the beginning and the absence of newline at the end)
The above would have caused ICE when trying to report where "refe" borrow ends.
The above without an empty line in the beginning would have reported borrow end to be the first line.
Most probably, this is also responsible for (at least some occurrences of) issue #8256.
The issue is fixed by always adding a newline at the end of non-empty filemaps in case there isn't a new line there already.
Edward Wang [Thu, 6 Mar 2014 14:58:34 +0000 (22:58 +0800)]
borrowck: classify expressions as assignees, uses or both
- Repurposes `MoveData.assignee_ids` to mean only `=` but not `+=`, so
that borrowck effectively classifies all expressions into assignees,
uses or both.
- Removes two `span_err` in liveness analysis, which are now borrowck's
responsibilities.
Adrien Tétar [Sun, 9 Mar 2014 10:06:03 +0000 (11:06 +0100)]
doc: remove node.js dependency
`prep.js` outputs its own HTML directives, which `pandoc` cannot
recognize when converting the document into LaTeX (this is why the
PDF docs have never been highlighted as of now).
Note that if we were to add the `.rust` class to snippets, we could
probably use pandoc's native highlighting capatibilities i.e. Kate.
bors [Sun, 9 Mar 2014 10:06:42 +0000 (03:06 -0700)]
auto merge of #12747 : huonw/rust/rustdoc-markdown, r=alexcrichton
This gives Rustdoc the ability to render our guides, tutorial and manual and modifies the those documents (minor modifications) and makefiles to achieve this.
See commits for more details (especially the makefile rewrite).
Huon Wilson [Sun, 9 Mar 2014 04:54:16 +0000 (15:54 +1100)]
mk: only build PDFs of the manual and tutorial.
This restores the old behaviour (as compared to building PDF versions of
all standalone docs), because some of the guides use unicode characters,
which seems to make pdftex unhappy.
Huon Wilson [Sun, 9 Mar 2014 03:14:07 +0000 (14:14 +1100)]
docs: render rustdoc docs with rustdoc, hack around sundown code-fence
parsing limitations.
Sundown parses
```
~~~
as a valid codeblock (i.e. mismatching delimiters), which made using
rustdoc on its own documentation impossible (since it used nested
codeblocks to demonstrate how testable codesnippets worked).
This modifies those snippets so that they're delimited by indentation,
but this then means they're tested by `rustdoc --test` & rendered as
Rust code (because there's no way to add `notrust` to
indentation-delimited code blocks). A comment is added to stop the
compiler reading the text too closely, but this unfortunately has to be
visible in the final docs, since that's the text on which the
highlighting happens.
Huon Wilson [Sun, 9 Mar 2014 01:50:45 +0000 (12:50 +1100)]
tutorial: hack a code snippet to make it compile.
This is meant to be compiling a crate, but the crate_id attribute seems
to be upsetting it if the attribute is actually on the crate. I.e. this
makes this test compile by putting the crate_id attribute on a function
and so it's ignored. Such a hack. :(
Huon Wilson [Sun, 9 Mar 2014 03:55:20 +0000 (14:55 +1100)]
mk: rename `check-...-doc-<crate>` to `check-...-doc-crate-<crate>`.
E.g. this stops check-...-doc rules for `rustdoc.md` and `librustdoc`
from stamping on each other, so that they are correctly built and
tested. (Previously only the rustdoc crate was tested.)
Huon Wilson [Sat, 8 Mar 2014 14:41:31 +0000 (01:41 +1100)]
mk: rewrite the documentation handling.
This converts it to be very similar to crates.mk, with a single list of
the documentation items creating all the necessary bits and pieces.
Changes include:
- rustdoc is used to render HTML & test standalone docs
- documentation building now obeys NO_REBUILD=1
- testing standalone docs now obeys NO_REBUILD=1
- L10N is slightly less broken (in particular, it shares dependencies
and code with the rest of the code)
- PDFs can be built for all documentation items, not just tutorial and
manual
- removes the obsolete & unused extract-tests.py script
- adjust the CSS for standalone docs to use the rustdoc syntax
highlighting
Huon Wilson [Sat, 8 Mar 2014 11:05:20 +0000 (22:05 +1100)]
docs: adjust code blocks to pass with rustdoc.
The changes are basically just because rustdoc runs tests/rendering on
more snippets by default (i.e. everything without a `notrust` tag), and
not anything significant.
Huon Wilson [Sat, 8 Mar 2014 10:30:43 +0000 (21:30 +1100)]
rustdoc: adding some common feature gates when testing a markdown file.
The manual, tutorial and guides need the feature gates quite often,
unfortunately, so this is the low-cost path to migrating to use
rustdoc. This is only activated for pure-Markdown files.
Huon Wilson [Fri, 7 Mar 2014 14:24:54 +0000 (01:24 +1100)]
rustdoc: hardcode each header as a link.
This avoids having to include JS in the guide/tutorial/manual pages just
to get the headers being links. The on-hover behaviour showing the
little section marker § is preserved, because that gives a useful hint
that the heading is a link.
Huon Wilson [Fri, 7 Mar 2014 14:13:17 +0000 (01:13 +1100)]
rustdoc: add table-of-contents recording & rendering, use it with plain
markdown files.
This means that
# Foo
## Bar
# Baz
### Qux
## Quz
Gets a TOC like
1 Foo
1.1 Bar
2 Baz
2.0.1 Qux
2.1 Quz
This functionality is only used when rendering a single markdown file,
never on an individual module, although it could very feasibly be
extended to allow modules to opt-in to a table of contents (std::fmt
comes to mind).
Huon Wilson [Fri, 7 Mar 2014 03:31:41 +0000 (14:31 +1100)]
rustdoc: run on plain Markdown files.
This theoretically gives rustdoc the ability to render our guides,
tutorial and manual (not in practice, since the files themselves need to
be adjusted slightly to use Sundown-compatible functionality).
bors [Sun, 9 Mar 2014 04:51:48 +0000 (20:51 -0800)]
auto merge of #12758 : rgawdzik/rust/master, r=alexcrichton
Refactored get_metadata_section to return a Result<MetadataBlob,~str> instead of a Option<MetadataBlob>. This provides more clarity to the user through the debug output when using --ls.
This is kind of a continuation of my original closed pull request 2 months ago (#11544), but I think the time-span constitutes a new pull request.
bors [Sun, 9 Mar 2014 01:31:57 +0000 (17:31 -0800)]
auto merge of #12759 : lucab/rust/char-doc, r=alexcrichton
This is mostly a reaction to #12730. If we are going to keep calling them `char`, at least make it clear that they aren't characters but codepoint/scalar.
Luca Bruno [Fri, 7 Mar 2014 22:53:34 +0000 (23:53 +0100)]
doc: don't refer to 'char' as characters
This seems to be causing some confusion among users. Rust's char are
not 8bit characters, but 32bit UCS-4 codepoint without surrogates
(Unicode Scalar Values as per Unicode glossary).
Make the doc more explicit about it.
bors [Sat, 8 Mar 2014 01:16:52 +0000 (17:16 -0800)]
auto merge of #12752 : edwardw/rust/rdoc, r=alexcrichton
A structure's definition and implementation may be cross-module. If the
implementing module is parsed before defining module, the fully
qualified name of the structure won't be present for the implementation
to use when being indexed. So caches such 'orphan' implementation and
indexes it at the end of crate parsing.
Edward Wang [Fri, 7 Mar 2014 09:26:06 +0000 (17:26 +0800)]
Index cross-mod type definition and implementation properly in rustdoc
A structure's definition and implementation may be cross-module. If the
implementing module is parsed before defining module, the fully
qualified name of the structure won't be present for the implementation
to use when being indexed. So caches such 'orphan' implementation and
indexes it at the end of crate parsing.
bors [Fri, 7 Mar 2014 02:16:39 +0000 (18:16 -0800)]
auto merge of #12635 : alexcrichton/rust/speedy-hash, r=brson
This leverages the new hashing framework and hashmap implementation to provide a
much speedier hashing algorithm for node ids and def ids. The hash algorithm
used is currentl FNV hashing, but it's quite easy to swap out.
I originally implemented hashing as the identity function, but this actually
ended up in slowing down rustc compiling libstd from 8s to 13s. I would suspect
that this is a result of a large number of collisions.
With FNV hashing, we get these timings (compiling with --no-trans, in seconds):