Various FIXME comments added around to denote copies which when removed cause
the compiler to segfault at some point before stage2. None of these copies
should even be necessary.
- Most functions that used to return `~[~str]` for a list of substrings got turned into iterators over `&str` slices
- Some cleanup of apis, docs and code layout
The basic problem was that vtables were not being resolved. The fix uncovered another issue, which was that the syntax extensions were not creating method calls properly and were relying on outdated code in typeck, so I fixed that too.
Resolves issues #3888, #4036, #4492.
r? @catamorphism
and from typeck, which is verboten. We are supposed to write inference results
into the FnCtxt and then these get copied over in writeback. Add assertions
that no inference by-products are added to this table.
Fixes#3888Fixes#4036Fixes#4492
Added notes explaining how [expr, ..expr] form is used, targeted at
individuals like me who thought it was more general and handled
dynamic repeat expressions. (I left a TODO for this section in a
comment, but perhaps that is bad form for the manual...)
Added example of `do` syntax with a function of arity > 1; yes, one
should be able to derive this from the text above it, but it is still
a useful detail to compare and contrast against the arity == 1 case.
Added example of using for expression over a uint range, since someone
who is most used to write `for(int i; i < lim; i++) { ... }` will
likely want to know how to translate that form (regardless of whether
it happens to be good style or not for their use-case).
Added note about the semi-strange meaning of "fixed size" of vectors
in the vector type section.
In order to mitigate typo of target-triples, error notification of unsupported target triples which defined in mk/platform.mk added.
minor fix for arm-linux-androideabi added.
This currently requires workarounds for the borrow checker not being flow-sensitive for `LinearMap` and `TrieMap`, but it can already be expressed for `TreeMap` and `SmallIntMap` without that.
Kills some warnings, and implements str::each_char_reverse so that it actually iterates. The test case wasn't detecting a failure, since the loop body was never executed.
`let v = [24, ..1000];` now more or less emits the same IR as:
```Rust
let mut i = 0;
while i < 1000 {
v[i] = 24;
i += 1;
}
```
LLVM will still turn it into a memset if possible with optimization on.
The reasoning for doing it this way is that it's much easier to transition method-by-method to the `Map` API than trying to do the migration all at once.
I found an issue unrelated to my changes in one of the run-fail tests - if it uses `LinearMap`, it still fails but exits with 0. I xfailed it for now and opened [an issue](https://github.com/mozilla/rust/issues/5512), because it's not caused by these changes.
