Michael Bebenita
dbcc9b36b1
Let circular buffers actually grow to max sz, reset _next when resizing.
2010-07-28 20:30:29 -07:00
Michael Bebenita
c1795a1046
Change unread-on-destroy condition for circular buffer to merely a warning.
2010-07-28 20:30:29 -07:00
Michael Bebenita
c5e0ea7276
Add comment explaining NULL case in circular_buffer::enqueue and add logging to ::dequeue.
2010-07-28 20:30:29 -07:00
Michael Bebenita
d6cba83322
Add peek method to circular buffer.
2010-07-28 20:30:29 -07:00
Michael Bebenita
17fa6bfd9a
Change _unit_sz to unit_sz and make public in circular_buffer.
2010-07-28 20:30:28 -07:00
Michael Bebenita
60b0486c11
Make circular buffer use only power-of-two sizes, cheaper arithmetic.
2010-07-28 20:30:28 -07:00
Michael Bebenita
d786469785
Add a warning interface to rust_srv.
2010-07-28 20:30:28 -07:00
Michael Bebenita
6afb6c767e
Move allocation-tracking into rust_srv.
2010-07-28 20:30:28 -07:00
Michael Bebenita
712249d6b8
Log a separator when logging switches between threads.
2010-07-28 20:30:28 -07:00
Michael Bebenita
7ead107290
array_list improvements.
2010-07-28 20:30:28 -07:00
Roy Frostig
dfcf21ca82
Another deque test commented out.
2010-07-28 18:14:33 -07:00
Roy Frostig
3c22e79b77
Burn the mac and win trees less. Deque test is XFAILed, despite doing fine on Linux.
2010-07-28 17:50:07 -07:00
Roy Frostig
c489af8e1a
Test the deque some more.
2010-07-28 17:31:41 -07:00
Roy Frostig
237b9d4a64
Test the deque more and fix uncovered off-by-one bug.
2010-07-28 16:34:22 -07:00
Roy Frostig
f282c5ccc0
Get slots in trans_tag using Semant tables. Closes #133 .
2010-07-28 15:04:58 -07:00
Roy Frostig
a9ad2e98e3
Refer to issue #136 at the offending source point.
2010-07-28 14:11:56 -07:00
Roy Frostig
596d19e2ea
Test the deque a bit. Give it a get-by-index method. Fix two uncovered state-calculation bugs --- one decently, the other with an ugly hack. Bug on the latter coming right up.
2010-07-28 14:00:44 -07:00
Graydon Hoare
8030757624
Switch machine-type lexemes to use suffixes. Remove support for foo(bar) as a cast notation. Closes #129 .
2010-07-27 19:21:51 -07:00
Graydon Hoare
6662aeb779
Calm some LLVM indigestion of last change.
2010-07-27 17:23:41 -07:00
Graydon Hoare
4d31cf1dc5
Distill semantics of use-def maps to fewer and more-obvious words.
...
- Remove redundant uses of 'resolve' and 'referent' in semant.
- Use defn, defn_id, lval, lval_base more consistently.
- Make associated query functions more consistent.
- Closes #127 .
2010-07-27 11:19:43 -07:00
Graydon Hoare
b7d7f70d09
Slight shaving on RA, no more optimizing today. It's fast enough for now.
2010-07-26 16:55:54 -07:00
Roy Frostig
1a61fb8eb7
Adjust testcase to cover issue #91 , which was actually fixed back in commit f02f9cbf29
. Closes #91 .
2010-07-26 15:52:59 -07:00
Graydon Hoare
3d69407b51
Fix numerous non-linearities and O(sizeof(crate)) issues in typestate system's dataflow algorithm. No longer substantial in profile.
2010-07-26 15:44:29 -07:00
Patrick Walton
4b97b4e79d
Move the test suite to the "as" form for casts. XFAIL a few tests for LLVM.
2010-07-26 15:22:21 -07:00
Graydon Hoare
ec5efd2577
Note the typestate dataflow timing. The culprit is revealed.
2010-07-26 12:31:43 -07:00
Graydon Hoare
c4897f849c
Add a cumulative timer count to be sure we're getting everything.
2010-07-26 12:31:08 -07:00
Graydon Hoare
1b41d7be1a
Change ctxt_crate_main to an Ast.name.
2010-07-26 12:30:43 -07:00
Graydon Hoare
025b1e4133
Do some more iflog-guarding.
2010-07-26 12:30:02 -07:00
Roy Frostig
085cd2ee7d
vec_alloc takes four arguments these days, not three.
2010-07-25 22:21:07 -07:00
Roy Frostig
beca62e2e3
Attempt 2. This time actually reading my own diffs before committing.
2010-07-25 22:07:21 -07:00
Roy Frostig
b9f82d85f9
Attempt number 1 at extinguishing the windows tree remotely (apologies to those hacking on Sunday evening PST).
2010-07-25 22:05:39 -07:00
Roy Frostig
5b6e714d05
Expose an RNG (the one used by our runtime) to Rust via std.
2010-07-25 21:45:09 -07:00
Roy Frostig
7ef9e82f51
Don't write to NULL after calling C natives returning void.
2010-07-25 21:27:28 -07:00
Graydon Hoare
3ce18d88c5
Another quick fix from jmuizelaar for that "some ocaml versions dont't notice wrong wildcard count in patterns" bug.
2010-07-23 15:38:34 -07:00
Graydon Hoare
44e2dc2789
Improve mutability checking. Closes #118 .
2010-07-23 15:29:17 -07:00
Graydon Hoare
8bd8413906
Add test for writing-through-uninit bug (reported on IRC by jrmuizel), plus fix in typestate system.
2010-07-23 13:52:46 -07:00
Graydon Hoare
62b69503d5
Stop using project_lval_ty_from_slot for lval_ty; cover residual un-caught cases in type.ml.
2010-07-23 13:51:56 -07:00
Graydon Hoare
ac228a59ce
Widen write mutability check to cover all writing stmts.
2010-07-23 13:51:17 -07:00
Graydon Hoare
fe82978980
Modify testcase to match new syntax and un-XFAIL mutable-vec-drop.rs.
2010-07-23 12:32:12 -07:00
Graydon Hoare
b5e46ac2a0
Two more testcases for overwriting already-live box-like fields in structures, without leaking.
2010-07-23 12:22:31 -07:00
Graydon Hoare
6668595ebf
Include all lval-writing statements in stmt_is_init calculation, not just "copy-like". Un-XFAIL generic-tag-alt.rs
2010-07-23 12:21:06 -07:00
Graydon Hoare
7d44ee7c3c
Add pretty-printing for alt-tag statements.
2010-07-23 12:20:20 -07:00
Graydon Hoare
4d4fa99b31
Rename STMT_init_* to STMT_new_*; former name was confusing.
2010-07-23 11:37:38 -07:00
Roy Frostig
1730d2e037
Notify copy glue of dst-initialization and fix _vec.alloc issues in lib and runtime. Closes #109 .
2010-07-22 17:47:32 -07:00
Graydon Hoare
ede42cf931
A certain incomplete quantity of wrestling with "INIT" statements that don't actually initialize. Should probably rename them to MAKE. Anyway, WIP, but two steps forward (and one back). More later.
2010-07-22 17:05:45 -07:00
Graydon Hoare
c6af9dda69
Un-XFAIL stuff that started working when pcwalton's new typechecker landed.
2010-07-22 15:40:04 -07:00
Graydon Hoare
7ff2094496
Back out too-platform-fussy bits in preempt-test work. I hate this test.
2010-07-22 15:39:41 -07:00
Graydon Hoare
1cb4a57b7b
Re-classify some err / bug cases as unimpl. Ideally rustboot should never produce a backtrace.
2010-07-22 15:25:29 -07:00
Graydon Hoare
0f220ecae9
Beat up on the preempt test a bit more, as it keeps hanging under valgrind.
2010-07-22 15:05:35 -07:00
Roy Frostig
3708865906
Re-introduce bits of vec-lib test that blocked on now-fixed issue #108 .
2010-07-22 12:49:11 -07:00