Commit graph

3664 commits

Author SHA1 Message Date
Brian Anderson
ed6f9bb677 Add tests for unsupervise 2011-07-14 18:50:22 -07:00
Brian Anderson
c99e110119 Modify upcall_fail so that failing tasks relinquish control to the scheduler
This is sufficient to let unsupervised tasks that don't allocate fail without
crashing or leaking.
2011-07-14 18:49:18 -07:00
Brian Anderson
01a52d728c Add an unsupervise builtin and function to std::task
Calling task::unsupervise de-parents the current task, with the result
that failures do not propogate up the task tree.
2011-07-14 17:56:59 -07:00
Brian Anderson
4738cf83b0 Move the responsibility for process failure from tasks to the scheduler
When the root task fails the process fails. Failures on other tasks propagate
up the task tree. Failures on non-root tasks without parents just
(theoretically) unwind and disappear.
2011-07-14 17:44:39 -07:00
Eric Holk
827e300485 refactor: Move the task and communication-related translation functions to a new module. 2011-07-14 17:33:28 -07:00
Graydon Hoare
37b879bdf7 Fix fuzzer to compile with ivec-ized pp module. 2011-07-14 17:27:36 -07:00
Michael Sullivan
e9a6fc94d5 Make collect_upvars know about function args.
Closes #697.
2011-07-14 19:18:56 -05:00
Michael Sullivan
a3301f74f9 Generalize collect_upvars to work over any type of ast node. 2011-07-14 19:18:56 -05:00
Brian Anderson
20e94de392 Add a flag to run ignored tests. Issue #428 2011-07-14 17:13:12 -07:00
Brian Anderson
81acf69f97 Add head and tail functions to std::ivec
They even have typestate preconditions
2011-07-14 17:13:12 -07:00
Brian Anderson
139aaa1616 Add is_empty, is_not_empty preds to std::ivec 2011-07-14 17:13:12 -07:00
Eric Holk
8afb1a7c63 refactor: Move the LLVM type and constant constructors into a new module. 2011-07-14 17:08:45 -07:00
Michael Sullivan
e2fcd29bbd Consolidate some arguments in typeck by just passing the record they came from. 2011-07-14 18:45:55 -05:00
Michael Sullivan
53c4cb6a2f Factor out build_environment's code to find a variable in a function context. 2011-07-14 18:45:02 -05:00
Michael Sullivan
759317ca8b Factor out closure construction from trans_for_each. 2011-07-14 18:45:01 -05:00
Michael Sullivan
0957525cc6 Fix array out of bounds index in load_environment.
Closes #696.
2011-07-14 18:34:43 -05:00
Patrick Walton
1a6419bd73 rustc: Move metadata::encoder over to interior vectors 2011-07-14 15:39:35 -07:00
Brian Anderson
b3dee95514 Add a facility for ignoring tests. Issue #428
Adding the #[ignore] attribute will cause the test not to be run, though it
will still show up in the list of tests.
2011-07-14 13:51:30 -07:00
Graydon Hoare
49da7da441 Fix long line. 2011-07-14 12:55:17 -07:00
Rafael Ávila de Espíndola
c124a025bf Move rustc to the new llvm type system. Requires an update to llvm trunk. 2011-07-14 15:19:17 -04:00
Patrick Walton
be489ee9e2 rustc: Move much of metadata reading over to interior vectors 2011-07-14 12:00:48 -07:00
Patrick Walton
4664b67ea2 snap: Register new snapshots 2011-07-14 11:54:35 -07:00
Patrick Walton
e76f44453b rustc: First stab at interior string literals, untested as of yet 2011-07-14 11:47:20 -07:00
Brian Anderson
4038010bc6 The test runner's main returns unit, not int. Issue #428
The appropriate way to indicate failure from main is to fail.
2011-07-14 10:51:38 -07:00
Brian Anderson
f4beac4a43 Revert "Allow main to return int"
This reverts commit 8c94d8fd54.

There's no mechanism to actually return the value from main, so all this does
is allow main -> int to compile. Per #688, the program returns non-zero on
failure, so it's not obvious that this change is appropriate at this time.
2011-07-14 10:37:28 -07:00
Brian Anderson
fc67dddb9b XFAIL some tests in stage 0 2011-07-14 10:27:02 -07:00
Brian Anderson
8d2706cca8 Flag --test implies '--cfg test'. Issue #428
So certain code can be conditionally compiled only when building for testing
2011-07-14 10:27:02 -07:00
Rafael Ávila de Espíndola
7376e751ae Keep a pointer to rust_object_type. 2011-07-14 12:07:51 -04:00
Rafael Ávila de Espíndola
40f617390d Main part of the type system rewrite:
Change the type of all objects to be { {}*, {}* }.
2011-07-14 00:34:45 -04:00
Brian Anderson
8c94d8fd54 Allow main to return int 2011-07-13 19:22:53 -07:00
Tim Chevalier
f26ca025de Make resolve and the typechecker check for a main fn of the
correct type

This means if a non-library program leaves out the main program,
the error gets caught earlier than link.

Closes #626.
2011-07-13 18:30:53 -07:00
Patrick Walton
196753e4c3 rustc: Remove some interior vectors from ty, except the ones that I think were causing crashes before 2011-07-13 17:39:33 -07:00
Rafael Ávila de Espíndola
c94fc7aad8 Small cleanups that were missing in the last patch. 2011-07-13 20:21:17 -04:00
Graydon Hoare
b9635df6fe Slight enhancements to round-trip code (run kdiff3 on mismatch, remove hard-wired input dir) 2011-07-13 17:15:31 -07:00
Jesse Ruderman
7885bdc575 Add issue numbers 2011-07-13 17:15:31 -07:00
Jesse Ruderman
156458b3f6 Fuzzer: test round trips through the pretty printer and parser 2011-07-13 17:15:31 -07:00
Rafael Ávila de Espíndola
9036758191 Next step in the type system change. Add task_type to the context. 2011-07-13 20:07:16 -04:00
Patrick Walton
2a3ab10f20 rustc: Remove exterior vectors from resolve again
This reverts commit 6390c43dc4.
2011-07-13 16:44:39 -07:00
Rafael Ávila de Espíndola
3d353f76a9 First step in the typesystem rewrite:
Create the typedesc type in one place. That is where it will be named.
2011-07-13 19:37:19 -04:00
Graydon Hoare
39151f2ad8 Prohibit trailing whitespace under 'tidy' script. Clean up all caught cases. 2011-07-13 15:44:09 -07:00
Tim Chevalier
04b239f3cb Use more precise spans in error messages for bad FRU exprs
The type error message for an expression using FRU where a field
expression had the wrong type was using the span for the entire
expression. Fixed it to use the span for the individual field.

Closes #628.
2011-07-13 15:30:30 -07:00
Tim Chevalier
6b86dcde67 Add test case from issue #675. The previous fix actually fixes this too. 2011-07-13 15:04:29 -07:00
Tim Chevalier
5e1a6dac44 Parse nullary ret correctly
ret is similar to fail: if not followed by an expression, it
should be parsed as a ret without an argument. The old version would
fail if ret was followed by a close paren (for example). Fixed it.

Closes #676.
2011-07-13 15:00:59 -07:00
Tim Chevalier
0c913e63d9 Add some missing cases to token::can_begin_expr 2011-07-13 15:00:23 -07:00
Graydon Hoare
d39f84d8ec Trim a few more fail(1)s, not sure how they slipped past. 2011-07-13 14:04:38 -07:00
Graydon Hoare
336a4df778 Remove 'Nop.' comments, add emacs lines, remove obsolete file. 2011-07-13 14:03:18 -07:00
Graydon Hoare
49116adbe9 Fix compile-command lines in rt. 2011-07-13 13:51:20 -07:00
Graydon Hoare
2e2951305d Remove obsolete nargs counts from runtime. 2011-07-13 13:43:35 -07:00
Graydon Hoare
01fc165517 Attempt to correct buggy win32 timer code (causing tinderbox failures). 2011-07-13 12:25:36 -07:00
Marijn Haverbeke
ece2d27205 Feeble attempt at putting out the win tinderbox 2011-07-13 11:36:09 +02:00