2561 Commits

Author SHA1 Message Date
Michael Sullivan
95680474e2 Use lambdas in the freevars pass. 2011-08-04 16:18:12 -07:00
Patrick Walton
ae3312002a rustc: Use memmove in copy_ty. 45% LLVM codegen speed improvement. 2011-08-04 16:14:28 -07:00
Tim Chevalier
e0985c1060 Handle alt on a _|_ - typed value
Return the result of the discriminant from trans_alt,
rather than nil, in the _|_ case. This was breaking the
enclosed test case (alt-bot-2) when optimization was disabled.

Closes #769
2011-08-04 16:07:26 -07:00
Tim Chevalier
5f03ca48d8 Add a fast path in ty::occurs_check_fails
Use type_contains_vars in occurs_check_fails to avoid doing
any work most of the time. This fixes a performance regression.
(No one else noticed yet that typechecking just got 4x slower, right?
Well, now it isn't anymore. :-})
2011-08-04 15:56:40 -07:00
Tim Chevalier
2baaeab784 Implement the occurs check
In the writeback phase, the typechecker now checks that it isn't
replacing a type variable T with a type that contains T. It
also does an occurs check in do_autoderef in order to avoid
getting into an infinite chain of derefs.

I'm a bit worried that there are more places where the occurs
check needs to happen where I'm not doing it now, though.

Closes #768
2011-08-04 15:30:09 -07:00
Lindsey Kuper
d7d4b4fc38 trans_args no longer needs llobj argument. 2011-08-04 11:34:06 -07:00
Patrick Walton
2a3235b58c rustc: Actually emit shapes 2011-08-04 11:25:46 -07:00
Patrick Walton
4061ca2bbd rustc: Generate shapes 2011-08-04 10:46:10 -07:00
Lindsey Kuper
1b1d8e7b91 Killing an obsolete comment. 2011-08-03 20:03:05 -07:00
Lindsey Kuper
6df7c041bc Pointer-ifying llenv. Step 1 of 2 steps to object system sanity. 2011-08-03 19:50:19 -07:00
Lindsey Kuper
29ec2503db Put comments on trans contexts back where they belong. :( 2011-08-03 19:49:29 -07:00
Michael Sullivan
5d5a3ca52d Fix trans_put to properly return (). Closes #773. 2011-08-03 18:10:54 -07:00
Michael Sullivan
5ea8d7f467 Make the pretty printer put trailing newlines at the end of files. 2011-08-03 17:52:25 -07:00
Michael Sullivan
e33de59e47 Clean up how we deal with dynamic size function arguments/returns. 2011-08-03 14:24:41 -07:00
Michael Sullivan
7bcde7270d Have trans_bind_thunk handle polymorphic but statically sized return values.
Closes #775.
2011-08-03 14:24:41 -07:00
Tim Chevalier
001df3f0ca Revert "Revert "Handle conditionals on _|_ - typed values correctly""
This reverts commit ea81c03960264bf590cd99ed2b662243e3db7a7c.

Changed the case in trans_if where the conditional is _|_ - typed
but the block is terminated to return the result of the cond,
instead of nil.

This passes "make check" with optimization disabled as well as
enabled.
2011-08-03 13:07:41 -07:00
Paul Stansifer
513276e595 Add #concat_idents[] and #ident_to_str[] 2011-08-03 12:42:35 -07:00
Michael Sullivan
c9ae548dae Reject programs that do a put outside of iterators.
Closes #774.
2011-08-03 11:43:50 -07:00
Michael Sullivan
8871462575 Do a bunch more typechecking for iters and for each loops.
Closes #771.
Closes #772.
Closes #796.
2011-08-03 11:07:53 -07:00
Michael Sullivan
3254fb9806 Code cleanup in check_expr. 2011-08-03 11:06:58 -07:00
Michael Sullivan
fcc458d4a0 Fix typechecking when spawning something type inferred.
Of course, we still don't *translate* it.

Closes #757.
2011-08-03 11:06:58 -07:00
Brian Anderson
e4da7c5bee parse_crate_from_source_str takes a parse_sess, not codemap
This was causing problems when reading from stdin for subsequent passes that
needed to generate node ids.
2011-08-03 10:55:59 -07:00
Brian Anderson
97a8784c98 Fix pretty-printer to read from files again
I accidentally made the pretty-printer always read from stdin
2011-08-03 10:55:59 -07:00
Brian Anderson
0b7a94a94c Pretty-print kinds of type params 2011-08-03 10:55:59 -07:00
Brian Anderson
843767a841 Disambiguate unop statements in pretty-printer. Closes #674 2011-08-03 10:55:59 -07:00
Brian Anderson
ea81c03960 Revert "Handle conditionals on _|_ - typed values correctly"
This reverts commit 13f8b3f2a67a4e3a7a26b0e238e1b6ce9e1f3573.

run-pass/if-ret.rs does not translate correctly when unoptimized. Issue #797
2011-08-03 10:48:52 -07:00
Marijn Haverbeke
d08c0f0ec1 Make ast::pat_bindings an iterator
And use it to get rid of some repetetive code
2011-08-03 10:26:41 +02:00
Tim Chevalier
948f8090ae Handle _|_ - typed discriminants in alts correctly
Stop me, won't you, if you've heard this one before?

Closes #794
2011-08-02 19:02:38 -07:00
Tim Chevalier
13f8b3f2a6 Handle conditionals on _|_ - typed values correctly
Closes #776
2011-08-02 18:34:44 -07:00
Tim Chevalier
7c34550931 Make _|_ type binopable
But don't actually generate code that does the operation. That means
hoisting the check I added in my last commit from trans_compare
up into trans_eager_binop (don't generate any code if one operand
has type _|_ ).

Closes #777
2011-08-02 18:06:46 -07:00
Brian Anderson
2911156820 Pretty-print fn constraints more correctish 2011-08-02 17:49:11 -07:00
Brian Anderson
3eef9993af Don't pp extra lines after block open when preserving whitespace. Closes #759 2011-08-02 17:49:11 -07:00
Tim Chevalier
731797d075 In trans, don't assume both sides of a binop have the same type
This was at least partially responsible for Issue 777.

    The only solution I can think of is for trans to just not generate
    code for a comparison if one or both sides has type _|_. Since
    that means evaluating that subexpression diverges, it should be ok
    to never do the comparison. Actually generating code for the
    comparison would trip an LLVM assertion failure.
2011-08-02 17:36:41 -07:00
Lindsey Kuper
430a28bb49 Refactor: a backwarding vtable can only have one kind of method. 2011-08-02 16:59:49 -07:00
Lindsey Kuper
48467c4faa Comment tweaks and re-flows. 2011-08-02 16:53:31 -07:00
Michael Sullivan
f371482593 Consolidate environment building/loading between closure types. 2011-08-02 16:30:56 -07:00
Michael Sullivan
4cf4e17e1d Track the node_id of the function in trans_common::fn_ctxt. 2011-08-02 16:30:56 -07:00
Michael Sullivan
f8b0d3d7d2 Use GEPi a bunch. 2011-08-02 16:26:00 -07:00
Paul Stansifer
ab4764520c Allow patterns of the form `[a, b, c ...] to be matched and transcribed. 2011-08-02 14:46:02 -07:00
Tim Chevalier
40c1b864c6 Use or-patterns in trans::trans_compare, eliminating a FIXME 2011-08-02 13:17:59 -07:00
Brian Anderson
35e9e02066 Use ioivec::read_whole_file_str in a few places 2011-08-02 10:39:14 -07:00
Brian Anderson
840a09c86e Add a --no-trans command line option
Runs all passes before translation. This will be used for pretty-printing
tests to verify that the results are still probably valid Rust code.
2011-08-02 10:39:12 -07:00
Brian Anderson
5f4b7e1ba7 Compiler accepts input from stdin when source file is called "-" 2011-08-02 10:39:09 -07:00
Marijn Haverbeke
2d5b651f49 Assign collection element ty to loop local tvar when checking loops
This makes the type declarationg for the loop variable optional in
most cases.

Closes #790
2011-08-02 15:09:29 +02:00
Marijn Haverbeke
78a0d380cc Do not try to save block result when the block is an iter body
Closes #791
2011-08-02 14:28:32 +02:00
Marijn Haverbeke
6c9b90d06a Be a little more clever about picking columns to match on in trans_alt
This should result in slightly more efficient matching of 'complex'
patterns with multiple discriminants in them.
2011-08-02 12:57:27 +02:00
Marijn Haverbeke
f8fa574864 Copy locals created by destructuring on the content of a box
This is required so that assigning to these locals doesn't clobber
the content of the box.

(A possible optimization would be to only do this copying for
locals that actually are assigned to.)
2011-08-02 12:09:15 +02:00
Marijn Haverbeke
043d95a2bd Move ppaux::ty_to_str to new record syntax 2011-08-02 12:09:15 +02:00
Marijn Haverbeke
8d8ff16dfe Improve handling of bottom type in alt arms 2011-08-02 10:24:06 +02:00
Tim Chevalier
5cf5f5024d Handle bang functions correctly in typestate
The logic for how the "returns" constraint was handled was always
dodgy, for reasons explained in the comments I added to
auxiliary::fn_info in this commit. Fixed it by adding distinct
"returns" and "diverges" constraints for each function, which
are both handled positively (that is: for a ! function, the
"diverges" constraint must be true on every exit path; for
any other function, the "returns" constraint must be true
on every exit path).

Closes #779
2011-08-01 20:58:16 -07:00