Fixes#19759
I'm not going to bother to do more than this, as it'll end up getting re-done as part of the reference work, but at least it's correct now.
Fix for `error: functions used as tests must have signature fn() -> ()` and `error: functions used as benches must have signature `fn(&mut Bencher) -> ()` in case of explicit return type declaration.
While trying to experiment with changes for some other issues, I noticed that the test for #15149 was failing because I have `/tmp` mounted as `noexec` on my Linux box, and that test tries to run out of a temporary directory. This may not be the most common case, but it's not rare by any means, because executing from a world-writable directory is a security problem. (For this reason, some kernel options/mods such as grsecurity also can prevent this on Linux.) I instead copy the executable to a directory created in the build tree, following the example of the `process-spawn-with-unicode-params` test.
After I made that change, I noticed that I'd made a mistake, but the test was still passing, because the "parent" process was not actually checking the status of the "child" process, meaning that the assertion in the child could never cause the overall test to fail. (I don't know if this has always been the case, or if it has something to do with either Windows or a change in the semantics of `spawn`.) So I fixed the test so that it would fail correctly, then fixed my original mistake so that it would pass again.
The one big problem with this is that I haven't set up any machines of my own so that I can build on Windows, which is the platform this test was targeted at in the first place! That might take a while to address on my end. So I need someone else to check this on Windows.
The base64 support can be trivially removed (there are no in-tree users) and the regex support is a whopping 4k lines of code to maintain for a few non-critical uses in-tree. This commit migrates all current users in-tree away from regexes to custom matching code.
The most critical application affected by this migration is that the testing framework no longer considers filter arguments as regexes, but rather just a substring matching. It is expected that more featureful testing frameworks can evolve outside of the in-tree libtest version over time which can properly depend on libregex from crates.io.
[breaking-change]
The regex library was largely used for non-critical aspects of the compiler and
various external tooling. The library at this point is duplicated with its
out-of-tree counterpart and as such imposes a bit of a maintenance overhead as
well as compile time hit for the compiler itself.
The last major user of the regex library is the libtest library, using regexes
for filters when running tests. This removal means that the filtering has gone
back to substring matching rather than using regexes.
This PR is intended as alternative to #20958. It fixes the same grammar inconsistencies, but does not increase the operator precedence of `..`, leaving it at the same level as the assignment operator.
For previous discussion, see #20811 and #20958.
Grammar changes:
* allow `for _ in 1..i {}` (fixes#20241)
* allow `for _ in 1.. {}` as infinite loop
* prevent use of range notation in contexts where only operators of high precedence are expected (fixes#20811)
Parser code cleanup:
* remove `RESTRICTION_NO_DOTS`
* make `AS_PREC` const and follow naming convention
* make `min_prec` inclusive
r? nikomatsakis
This pull request add tooltips to most links of sidebar.
The tooltips display "summary line" of items' document.
Some lengthy/annoying raw markdown code are eliminated, such as links and headers.
- `[Rust](http://rust-lang.org)` displays as `Rust` (no URLs)
- `# header` displays as `header` (no `#`s)
Some inline spans, e.g. ``` `code` ``` and ```*emphasis*```, are kept as they are, for better readable.
I've make sure `&` `'` `"` `<` and `>` are properly displayed in tooltips, for example, `&'a Option<T>`.
Online preview: http://liigo.com/tmp/tooltips/std/index.html
@alexcrichton @steveklabnik since you have reviewed my previous ([v1](https://github.com/rust-lang/rust/pull/13014),[v2](https://github.com/rust-lang/rust/pull/16448)) PRs of this serise, which have been closed for technical reasons. Thank you.
Lifetime elision documentation was reference a previously existing function
that doesn't exist. After talking with Steve Klabnik, I confirmed the correct
function to be referenced and updated documentation accordingly.
When I wrote this code, my janky shim to verify the constants didn't
work as intended.
This fixes everything I've run into since merge, which is hopefully
everything.
Make compiletest set an environment variable so that on Windows 32-bit test executables containing certain substrings (e.g. 'install', 'setup', 'update', 'patch') are not blocked by UAC's Installer Detection Technology.
This should fix issue #20797 (but I don't want to close it automatically).
As the actual fix is very small this would be a perfect candidate for a rollup.
Compiling won't produce an executable just yet because (as stated in the next
paragraph) there are errors. By removing this sentance, the reader won't get
confused when they expect a successful compile i.e. if they don't read ahead one
paragraph, they are going to be checking their code and wondering why it's not
compiling.
Previous wording wasn’t clear about its actual behaviour. It could be
interpreted as answering either:
* Can current thread panic?
* Is current thread unwinding because of panic?
r? @steveklabnik
Not sure on what *exactly* should be said here, but I think this is the most important bit. This PR also establishes conventions for describing performance minimally.
I suggest to describe preformance for individual methods we use a `# Performance` heading. Not sure if we should have
```
# Performance: O(1)
details details
```
or
```
# Performance:
O(1)
details details
```
Since I think most methods don't need discussion, the former seems more resonable. But it's kind of weird to have info "in" the heading.
r? @steveklabnik
