- 27 8月, 2012 1 次提交
-
-
由 Patrick Walton 提交于
Previously, LLVM was generating a ton of byte-by-byte copies, leading to huge numbers of vregs and bloating the code. Now, using memset, the code becomes a nice series of SSE moves instead.
-
- 26 8月, 2012 1 次提交
-
-
由 Patrick Walton 提交于
This is needed for a lot of Apple libraries, as Apple tends to put a lot of globals in dynamic libraries.
-
- 25 8月, 2012 8 次提交
-
-
由 Tim Chevalier 提交于
-
由 Tim Chevalier 提交于
-
由 Graydon Hoare 提交于
-
由 Niko Matsakis 提交于
-
由 Elliott Slaughter 提交于
-
由 Elliott Slaughter 提交于
-
由 Elliott Slaughter 提交于
-
由 Elliott Slaughter 提交于
-
- 24 8月, 2012 2 次提交
-
-
由 Michael Sullivan 提交于
-
由 Paul Stansifer 提交于
-
- 23 8月, 2012 2 次提交
-
-
由 Niko Matsakis 提交于
A similar approach could be used for type parameters. Fixes #2282.
-
由 Paul Stansifer 提交于
-
- 22 8月, 2012 2 次提交
-
-
由 Niko Matsakis 提交于
There used to be two distinct code paths. Now there is one.
-
由 Niko Matsakis 提交于
-
- 18 8月, 2012 5 次提交
-
-
由 Michael Sullivan 提交于
-
由 Patrick Walton 提交于
-
由 Niko Matsakis 提交于
this will require more temporaries, but is probably less magical. also, it means that borrowck matches trans better, so fewer crashes. bonus. Finally, stop warning about implicit copies when we are actually borrowing. Also, one test (vec-res-add) stopped failing due to #2587, and hence I added an xfail-test. Fixes #3217, #2977, #3067
-
由 Patrick Walton 提交于
-
由 Brian Anderson 提交于
-
- 16 8月, 2012 5 次提交
-
-
由 Patrick Walton 提交于
-
由 Patrick Walton 提交于
-
由 Patrick Walton 提交于
-
由 Patrick Walton 提交于
There is also code for ~Trait and &Trait, but these are currently (incorrectly) synonyms for "as @Trait" and "as &Trait".
-
由 Brian Anderson 提交于
-
- 15 8月, 2012 2 次提交
-
-
由 Graydon Hoare 提交于
-
由 Lindsey Kuper 提交于
-
- 14 8月, 2012 3 次提交
-
-
由 Patrick Walton 提交于
-
由 Lindsey Kuper 提交于
-
由 Lindsey Kuper 提交于
-
- 11 8月, 2012 4 次提交
-
-
由 Patrick Walton 提交于
This reverts commit 0101125a.
-
由 Patrick Walton 提交于
-
由 Lindsey Kuper 提交于
-
由 Niko Matsakis 提交于
-
- 10 8月, 2012 2 次提交
-
-
由 Patrick Walton 提交于
-
由 Graydon Hoare 提交于
-
- 09 8月, 2012 3 次提交
-
-
由 Brian Anderson 提交于
-
由 Patrick Walton 提交于
-
由 Tim Chevalier 提交于
The first case was getting treated like a variable binding, meaning that if e had a destructor, it wouldn't run until the end of the enclosing scope. To me it seems less confusing for let _ = e; and e; to work exactly the same way, so now, the destructor for e runs immediately in both cases.
-