Commit Graph

93 Commits

Author SHA1 Message Date
Gustavo Massaccesi
bfc9eb8d62 Add ´not´ to the relevant predicates list
Previously all the predicates recognized only non-#f things, so ´not´ can be
added to the list of disjoint predicates. But many of the parts of the code
relied on the non-#f property and had to be modified.
2015-07-14 19:20:11 -03:00
Gustavo Massaccesi
dfc64053b7 Infer type from comparisons in test positions
In (if (eq? x <pred?-expr>) <tbranch> <fbranch>) infer that the type of
x is pred? in the tbranch.

Also, reduce (eq? x y) => #f when the types are different.
2015-07-14 19:19:50 -03:00
Gustavo Massaccesi
bc2cf531e3 Reduce variables with type null? and void? to null and #<void>
The optimizer reduces the variables with a known type to #t in a Boolean context.
But some predicates imply that the variable has a definite values, so they can be
reduced in a non-Boolean context too.

For example, in (lambda (x) (if (null? x) x 0))) reduce the last x ==> null.
2015-07-14 19:19:28 -03:00
Gustavo Massaccesi
58300857db Fix bug in type propagation to avoid the reduction of mutable variables
This fixes the bug twice:
* Don't reduce mutable variables with a type to #t in a Boolean context.
* Don't record the type of mutable variables when a predicate is
     checked in a test condition.
2015-07-14 19:19:05 -03:00
Gustavo Massaccesi
b7ae673ee0 Mark (values <expr>) as single valued
While reducing some ignored constructors, the optimizer may wrap the arguments
<expr> in (values <expr>) to ensure that it's a single value non-cm expression.
This avoids the unnecessary nesting of (values (values <expr>)).

Similarly, add the cases for begin and begin0 to single_valued_noncm_expression
2015-07-14 13:43:54 -03:00
Matthew Flatt
331b104345 JIT: inline ptr-ref and ptr-set!
Special treatment of `ptr-ref` and `ptr-set!` applies when the second
argument is one of a few primitive C types: `_int`, `_double`, etc.
2015-07-02 15:59:35 -06:00
Gustavo Massaccesi
72132ea3aa Reoptimize propagated constants
Reduces them to #t or #f when they are copied to a Boolean context
2015-06-29 23:44:47 -03:00
Matthew Flatt
6f6a792d06 fix memory-management problem with #:lock-name
Repairs a bug in 290020c597.
2015-06-26 07:44:43 +08:00
Gustavo Massaccesi
95bac91268 Propagate types form the body of let's forms and inlined functions
The optimizer was able to use the type information gained outside
the let's to reduce expressions inside the lets. For example, in
  (lambda (z) (car z) (let ([o (random)]) (pair? z)))
it reduces (pair? z) ==> #t.

This enable the propagation in the other direction so in
  (lambda (z) (let ([o (random)]) (car z)) (pair? z))
it reduces (pair? z) ==> #t too.
2015-06-23 16:52:40 -03:00
Gustavo Massaccesi
1a091f535e Intersect types gathered in if branches 2015-06-23 16:46:23 -03:00
Matthew Flatt
ea016bec96 allow definition in module to shadow a preceding import
This change is experimental and intended to reduce (but certainly not
eliminate) the problem of breaking existing code by adding exports to
a module.
2015-06-23 21:31:57 +08:00
Alexis King
c79f646545 Add stream* to complement stream 2015-06-20 21:43:27 +08:00
Sam Tobin-Hochstadt
c8ff0c8b18 Fix timeout for core tests. 2015-05-27 13:51:09 -04:00
Gustavo Massaccesi
542b960d12 Tests for 16ce8fd90d 2015-05-16 22:32:05 -03:00
Matthew Flatt
298feb1bb6 fix problem with expand and local-require
When `local-require` is used in a non-phase-0 position and it is
`expand`ed (as opposed to compiled directly), then the generated
`#%require` form had the wrong binding phase.

Merge to v6.2
2015-05-15 10:44:20 -06:00
Gustavo Massaccesi
6c2888937a Make (make-vector <number>) omittable
In many use cases the length of the vector is fixed and know,
so we are sure that make-vector will not raise an error and
we can recognize these expressions as omittable and drop
them when the result is ignored.
2015-05-14 16:36:35 -03:00
Gustavo Massaccesi
2be6eb9570 Mark the result of more procedures as vector?
The result of some procedures is a vector, but they are not omittable
because they may rise an error. With the recent changes of the
predicate reduction these cases are correctly handled.
2015-05-14 16:36:21 -03:00
Asumu Takikawa
c2cd44c928 Implement class sealing/unsealing
Adds a sealing and unsealing function to attach (or detach)
seals onto a class via impersonator properties. Since these
properties override, they do not accumulate wrappers.

Calling seal multiple times will still accumulate multiple seal
values inside the property.

A sealed class cannot be instantiated and a subclass may not
add class members that match any of the sealed names in its
sealed parent.

These functions are intended for use by TR's `sealing->/c`
contract, but are parameterized over checking functions and
could be used for other purposes.
2015-05-13 13:21:51 -04:00
Matthew Flatt
6f984d868c fix expander problem with set! 2015-05-09 18:35:25 -06:00
Stephen Chang
7d434d266e add free-id sets
merges github pull-request #815
2015-05-04 19:05:07 -04:00
Gustavo Massaccesi
4c10a9efac Check the type of the arguments of more procedures
The optimizer checks the type of the argument of some unary procedures and
uses the gathered information to replace them by the unsafe version, reduce
predicates and detect type errors. This extends the checks to more procedures
that have no unsafe version and procedures that have more than one argument.
2015-05-03 15:15:24 -03:00
Matthew Flatt
469763ca37 Adjust use of readtable argument in read/recursive
Use the given readtable more consistently to parse
delimiters in the top-level form. This change particularly
addresses problems with trying to restore the original
`(` when parsing a hash table, but allowing nested
forms to still use a different `(` mapping.
2015-04-15 13:01:58 -06:00
Eric Dobson
29b1ac70df Make for raise better error message if binding clauses are bad.
Closes PR 13272.
2015-04-10 13:56:42 -06:00
Matthew Flatt
23ec573e51 repair chaperone handling in current-command-line-arguments 2015-04-06 12:44:47 -05:00
Matthew Flatt
50ff92b784 defend against strage inferred-name property call with keyword args 2015-04-03 18:23:08 -05:00
Matthew Flatt
d22082f7e5 add #:retry option to _fun 2015-03-27 13:03:20 -06:00
Matthew Flatt
1a5e3e5953 _enum: allow negative integers
Corrects a problem with ad899173b9
2015-03-26 20:15:25 -06:00
Eric Dobson
ad899173b9 Add more error checking to _enum.
This also fixes some tests which were not running.
2015-03-26 11:33:20 -06:00
Eric Dobson
9fd9a42c61 Make enum work in non first order cases.
Closes PR 13323.
2015-03-26 11:33:20 -06:00
Matthew Flatt
ff344a6622 tempoarily remove test 2015-03-26 11:33:08 -06:00
Jay Kominek
3ad60aa67a fix integer-length overflow (PR14986) improve performance on integer-length of negative bignums 2015-03-26 11:20:40 -06:00
Matthew Flatt
2dd29f7e3d fix pessimism in optimizer reordering
When determing whether expressions can be reordered, a reference to a
module-defined variable was considered unreorderable when it is
known to have a value and no further mutation, but the value isn't
constant across all runs.
2015-03-26 09:15:13 -06:00
Sam Tobin-Hochstadt
785fb57dc1 Add test for "base" version. 2015-03-20 11:56:05 -04:00
Sam Tobin-Hochstadt
f73b4066a7 Add prop:object-name. 2015-03-18 09:55:27 -04:00
Eric Dobson
cb67ea3edf Raise error message if _fun is used incorrectly.
Closes PR 11323.
2015-03-18 04:19:57 -06:00
Eric Dobson
408d86f1c7 Make provide expansion respect disappeared uses on expanded syntax. 2015-03-17 19:35:26 -06:00
Gustavo Massaccesi
7981513b95 More redutions of predicates
The optimizer had some reductions of predicates applications, like (pair? X),
only when X was very simple and the type of X was obvious.
Use expr_implies_predicate and make_discarding_sequence to allow
the reduction of more complex expressions.

Also, the reduction of procedure? and fixnum? were special cases in
optimize_application2. Move the checks to expr_implies_predicate
to take advantage of the reductions in more general cases.
2015-03-17 19:28:37 -06:00
Matthew Flatt
332b380ca2 repair impersonator-porperty predicate and accessor
Repair for b923269569, helpfully reported again by Scott
2015-03-09 15:33:41 -06:00
Matthew Flatt
5749d4080c add tracking of require and provide subforms
Use `syntax-track-origin` and 'disappeared-use properties to
communicate `require` and `provide` form bindings to tools such as
Check Syntax.

Relevant to PR 13186
2015-03-09 15:28:08 -06:00
Eric Dobson
9c014b61e6 Make for/fold/derived check the values in the body in the last clause.
Closes PR 13285.
2015-03-09 12:48:49 -06:00
Eric Dobson
acd1fe7f8d Make define-cpointer-type declare its function with an inferred-name.
Closes PR 12296.
2015-03-09 12:45:35 -06:00
Matthew Flatt
b923269569 make impersonator properties sensitive to prop:impersonator-of
When a structure type has `prop:inpersonator-of`, follow it
when attemptng to access imperonator properties.

This change fixes a problem with `impersonate-procedure` as
reported by Scott Moore.
2015-03-08 19:27:11 -06:00
Matthew Flatt
092f6bb7e1 add object-or-false=? 2015-03-05 14:27:17 -07:00
Gustavo Massaccesi
de3fa9a855 Tests for correct use of reverse in procedure? reduction 2015-02-27 08:32:27 -07:00
Asumu Takikawa
5dc6be1a17 Allow chaperones on class methods
This allows libraries to add metadata to methods in the form
of chaperone properties which can be read by contracts on
methods.
2015-02-24 17:17:32 -05:00
Matthew Flatt
54fb57d6b9 fix extflonum-printing test when extflonums are disabled
The minimal suport for extflonums just preserves the text of the
input, so adjust the test to work for both.
2015-02-17 19:12:29 -07:00
Gustavo Massaccesi
4b8517b27c Recognize more procedures in scheme_optimize_apply_values
scheme_optimize_apply_values reduces (call-with-values gen proc)
to (#%apply-values proc gen) when recognizes proc as a procedure.
This extends the expressions that are recognized as procedures.
2015-02-16 10:06:52 -07:00
Gustavo Massaccesi
0c5944d64a Reduce (procedure? <inlineable>) => #t 2015-02-16 10:06:51 -07:00
Matthew Flatt
a8026824dd adjust optimizer to improve intra-module inlining
Instead of delaying the registration of some constants until a
group of expressions is re-optimized, add constant information as
it is discovered, which can expose some additional optimizations.

The old grouping was probably aimed at avoiding excessive code growth,
but I think that other and better controls are now in place. The
overall size of ".zo" files in an installation did not grow
significantly with this change.

Closes PR 14978
2015-02-16 10:01:16 -07:00
Matthew Flatt
d747f8f806 require a #:offset keyword before a field offset in define-cstruct
Also, allow `#:offset` specifications on individual fields, instead
of all or node.
2015-02-14 18:40:20 -07:00