Vincent St-Amour
382a45ad6f
Fixed the optimizer so that it optimizes float comparisons.
2010-06-25 11:00:26 -04:00
Vincent St-Amour
d6008f9191
Typed Scheme now optimizes (exact->inexact <Integer>) to (->fl <Integer>).
2010-06-25 11:00:25 -04:00
Vincent St-Amour
b345d5f0f0
Typed Scheme now optimizes (fl)vector-length for all vectors and
...
vector-(ref,set!) for vectors of known length.
2010-06-25 11:00:24 -04:00
Vincent St-Amour
d47221c55d
Added better tracing to the optimizer.
2010-06-25 11:00:23 -04:00
Sam Tobin-Hochstadt
3251712ef7
More fixnum improvements.
2010-06-24 18:17:01 -04:00
Sam Tobin-Hochstadt
4b1c62c978
Add Fixnum type.
2010-06-24 18:05:55 -04:00
Sam Tobin-Hochstadt
a33f460b25
recreate selector table automatically.
2010-06-24 16:48:10 -04:00
Sam Tobin-Hochstadt
40ca16bc30
Types are not quotable.
2010-06-24 16:47:25 -04:00
Sam Tobin-Hochstadt
3c8952d63d
Use the correct field numbers in substructs.
...
Maintain table of struct accessors/mutators.
2010-06-24 14:53:38 -04:00
Sam Tobin-Hochstadt
3078807757
Better handling of unsafe identifier generation.
2010-06-24 11:05:55 -04:00
Vincent St-Amour
3c9e7e34db
Switched indexing to integers in the typechecker's special cases for
...
heterogenous vectors and unsafe struct operations.
2010-06-22 18:42:26 -04:00
Vincent St-Amour
7d1040283e
Added support for unsafe struct operations to Typed Scheme.
2010-06-22 18:42:25 -04:00
Vincent St-Amour
379d9a2125
Added support for unsafe operations on heterogenous vectors.
2010-06-22 18:29:57 -04:00
Vincent St-Amour
eea31102b0
Moved the types for some byte string operations to
...
base-env-indexing-abs.rkt.
2010-06-22 18:29:57 -04:00
Vincent St-Amour
192c1fa995
Fixed the types of flvector operations.
2010-06-22 18:29:56 -04:00
Vincent St-Amour
61545c4e75
Made make-vector's type more convenient.
2010-06-22 18:29:56 -04:00
Sam Tobin-Hochstadt
6e1954d79b
Add `struct:'
2010-06-22 16:29:35 -04:00
Sam Tobin-Hochstadt
bce2cedf38
Typecheck `map' expression, not just #'map.
2010-06-22 10:12:54 -04:00
Sam Tobin-Hochstadt
4df7d09ef1
Docs for List x ...
2010-06-21 18:30:30 -04:00
Sam Tobin-Hochstadt
0c7c722e16
support `map' over multiple ListDots with the same bound
...
- also support ListDots + Listof (map errors when not same length)
2010-06-21 18:30:30 -04:00
Sam Tobin-Hochstadt
b8777d20b2
allow optional specifcation of var pattern
2010-06-21 18:30:29 -04:00
Stevie Strickland
034b22d014
Generate dcon-dotted, and substitute for them.
...
- generated only in the case where we have two ListDots or two ValuesDots
- or when -> with the same bound, and fixed argument lengts are identical
- currently errors if there are any 'fixed' portions
2010-06-21 18:30:29 -04:00
Stevie Strickland
9c22701bd3
Take dcon-dotteds and create an appropriate i-subst/dotted entry.
2010-06-21 18:30:29 -04:00
Stevie Strickland
3f7ca52e37
Add fixed part to dcon-dotted.
2010-06-21 18:30:29 -04:00
Sam Tobin-Hochstadt
339add9f78
Implement substitutions as hash tables from vars to subst-rhs.
...
- add convenience function for generating substitutions
- give trivial substitution for unreferenced index variables
2010-06-21 18:30:28 -04:00
Sam Tobin-Hochstadt
44d46e4cd7
Refactor substitution.
...
- new file types/substitute
- use structs for substitutions
2010-06-21 18:30:28 -04:00
Sam Tobin-Hochstadt
8990459be6
provide more information in inference failures
2010-06-21 18:30:27 -04:00
Sam Tobin-Hochstadt
692dc025c0
fix fail! to use the correct arguments
2010-06-21 18:30:27 -04:00
Sam Tobin-Hochstadt
9f99c914f3
Fix move-to-dmap functions
2010-06-21 18:30:27 -04:00
Stevie Strickland
b5d4d54d45
The initial shot at extending cgen and friends to take both X and Y.
...
This doesn't fully work, in that more test cases are broken. However,
this is mostly due to questionable inference rules in the past, and
so fixing these errors should lead to a more correct inference algorithm.
(i.e. we need to handle things like (List X ... a) (List Y .. b),
where a and b are not the same bound. We'd started this work before,
but never actually gone through with it, since smashing regular and
dotted type variables into the same environment meant some things
magically worked when they possibly shouldn't have.)
2010-06-21 18:30:26 -04:00
Stevie Strickland
078b57c77d
Export the types provide macro.
2010-06-21 18:30:26 -04:00
Stevie Strickland
ec09139e0a
Fix up subst-gen to treat vars and idxs separately.
2010-06-21 18:30:26 -04:00
Sam Tobin-Hochstadt
d886331807
Various small assorted fixes.
2010-06-21 18:30:25 -04:00
Sam Tobin-Hochstadt
82e7d281cb
More progress on refactoring infer
2010-06-21 18:30:25 -04:00
Sam Tobin-Hochstadt
18f45c4138
Fix a bunch of uses of `infer' et al.
2010-06-21 18:30:25 -04:00
Sam Tobin-Hochstadt
99e499a503
Change inference to take into account index vs regular vars.
2010-06-21 18:30:24 -04:00
Sam Tobin-Hochstadt
e8a591d291
Fix free index computation when the body refers to the bound.
2010-06-21 18:30:24 -04:00
Sam Tobin-Hochstadt
662f982b72
Refactor free variables.
...
- free-idxs no longer holds free de Bruijn indexes
This information is not needed, and not stored anywhere
- free-idxs holds free *index* variables, in the sense of the TOPLAS submission
2010-06-21 18:30:24 -04:00
Sam Tobin-Hochstadt
fe9395fdbf
Add case to be fixed
2010-06-21 18:30:23 -04:00
Sam Tobin-Hochstadt
b787c7766c
Switch to using functional hash tables for free variables.
2010-06-21 18:30:23 -04:00
Sam Tobin-Hochstadt
a93aecbed7
Special case for (apply values ListDots)
2010-06-21 18:30:20 -04:00
Sam Tobin-Hochstadt
12384c6c3f
Change type variable environments from hash tables to sets.
2010-06-21 18:29:59 -04:00
Sam Tobin-Hochstadt
d570006db8
Add index-env.
...
- Remove Dotted and DottedBoth values from tvar-env
- Abstract env extension and lookup for tvar/index-env
- Abstract index inference
- Remove pointless parameterizations
2010-06-21 18:29:58 -04:00
Sam Tobin-Hochstadt
0fb1ac66bd
Properly compute Dotted variance for ListDots and ValuesDots.
2010-06-21 18:29:58 -04:00
Sam Tobin-Hochstadt
a9c34dae6a
Remove dotted-env.
2010-06-21 18:29:58 -04:00
Sam Tobin-Hochstadt
f7b59f78e9
Generalize ListDots to Listof in inference of loop var types.
2010-06-21 18:29:58 -04:00
Sam Tobin-Hochstadt
a2af89bafd
Inference for passing ListDots as Listof.
2010-06-21 18:29:57 -04:00
Sam Tobin-Hochstadt
fd5a662ccc
Subtyping between (List T ... a) and (Listof T[Any/a])
2010-06-21 18:29:57 -04:00
Sam Tobin-Hochstadt
4c3f279ab9
Eliminate expressions with dotted pre-types.
...
- Now looks for (List T ...) types.
- special handling of `map', `andmap', `ormap' when list arg is dotted
- remove tc-dots-unit
2010-06-21 18:29:57 -04:00
Sam Tobin-Hochstadt
4cbeb0b2f0
Change type of ... rest args to have (List T ...) types.
2010-06-21 18:29:56 -04:00