Workaround the python vscode extension's polyfill
Fixes#13442
`String.replaceAll` and `String.replace` behave the same when given a (/g-flagged) Regex, so fix is very simple.
linker: Fix weak lang item linking with combination windows-gnu + LLD + LTO
In https://github.com/rust-lang/rust/pull/100404 this logic was originally disabled for MSVC due to issues with LTO, but the same issues appear on windows-gnu with LLD because that LLD uses the same underlying logic as MSVC LLD, just with re-syntaxed command line options.
So this PR just disables it for LTO builds in general.
Fix bug introduced by #9386#9386 introduced a potential out-of-bounds array access. Specifically, a location returned by `local_assignments` could have [`location.statement_index` equal to `mir.basic_blocks[location.block].statements.len()`](b8a9a507bf/clippy_utils/src/mir/mod.rs (L129)), in which case the location would refer to the block terminator:
b8a9a507bf/clippy_lints/src/dereference.rs (L1204-L1206)
I suspect the bug is not triggerable now, because of checks leading up to where it occurs. But a future code change could make it triggerable. Hence, it should be fixed.
r? `@Jarcho`
changelog: none
feat: add multiple getters mode in `generate_getter`
This commit adds two modes to generate_getter action.
First, the plain old working on single fields.
Second, working on a selected range of fields.
Should partially solve #13246
If this gets approved will create a separate PR for setters version of the same
### Points to help in review:
- `generate_getter_from_record_info` contains code which is mostly taken from assist before refactor
- Same goes for `parse_record_fields`
- There are changes in other assists, as one of the methods in utils named `find_struct_impl` is changed, before it used to accept a single `fn_name`, now it takes a list of function names to check against. All old impls are updated to create a small list to pass their single element.
### Assumptions:
- If any of the fields have an implementation, the action will quit.
Magic functions for writing to stdout/stderr.
This enables I/O in no_std contexts (or, really, any Miri-specific OS-independent context). Combined with the `abort` intrinsic it should allow a reasonable test framework in no_std.
**Question for maintainers:** So, the `no_std` panic test needs work, for two reasons:
- First, its stdout includes Miri's whole message about the abort intrinsic having been used. I guess whatever panic handler you use in `std` contexts exits cleanly without triggering this message. Comparing the entire output with backtrace as golden seems fragile.
- Second, likely for the same reason, the test framework appears to expect the test to exit successfully, when in fact it exits with status 1 due to the abort. This means the test doesn't actually pass right now.
What shall I do there?
The canonical form of a usub.with.overflow check in LLVM are
separate sub + icmp instructions, rather than a usub.with.overflow
intrinsic. Using usub.with.overflow will generally result in worse
optimization potential.
The backend will attempt to form usub.with.overflow when it comes
to actual instruction selection. This is not fully reliable, but
I believe this is a better tradeoff than using the intrinsic in
IR.
Fixes#103285.
Fix the bug of next_point in source_map
There is a bug in `next_point`, the new span won't move to next position when be called in the first time.
For this reason, our current code is working like this:
1. When we really want to move to the next position, we called two times of `next_point`
2. Some code which use `next_point` actually done the same thing with `shrink_to_hi`
This fix make sure when `next_point` is called, span will move with the width at least 1, and also work correctly in the scenario of multiple bytes.
Ref: https://github.com/rust-lang/rust/pull/103140#discussion_r997710998
r? `@davidtwco`
fix: Don't catch the server activation error
We are are rethrowing and showing errors higher up in the call stack already. This just ate the error hiding the stacktrace unnecessarily.
The primary reason for this is that make can result in a substantial
under utilization of parallelism, mostly due to the submake structure
preventing good dependency tracking and scheduling.
In f758c7b2a78 (Debian 6 doesn't have ninja, so use make for the dist builds)
llvm.ninja was disabled due to lack of distro package. This is no longer the
case with the CentOS 7 base, so bring ninja back for a performance boost.
rustdoc: remove no-op CSS `nav.sub { font-size: 1rem }`
This rule originated as a `font-size: 16px`, when body had `font-size: 13px` set in 4fd061c426902b0904c65e64a3780b21f9ab3afb.
It remained even when body's font size was bumped up to 16px, 4d5f4ff5e9297dcad21612f9dd20ae4598b5b7e2, making the rule a no-op, and was carried forward when it was converted to 1rem in cc18120425a5c571a968d850c75cc935a8321136.
Deny hashing ty/re/ct inference variables
cc `@cjgillot` and https://github.com/rust-lang/rust/pull/102695#issuecomment-1275706528
r? `@lcnr`
best reviewed one commit at a time, mostly because the second commit that fixes `ClosureOutlivesRequirement` is mostly noise because of losing its `<'tcx>` lifetime parameter.
