suzanne.soy/racket
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
22,782 Commits 17 Branches 54 Tags 278 MiB
dcda2b60d3
Commit Graph

3361 Commits

Author SHA1 Message Date
Matthew Flatt
98bc4067c3 make bytecode compiler less picky about exports 
Closes PR 12134
 2011-08-28 11:21:15 -06:00
Matthew Flatt
4f83f7f279 allow phase 1+ references to not-yet-defined bindings 
For example,

 (define-for-syntax (f x) (g x))
 (define-for-syntax (g y) y)

is now allowed. The unbound-variable check for phase 1
and up is delayed until after the module body is partially expanded.
 2011-08-28 10:27:31 -06:00
Kevin Tew
5d27959c57 Place sockets and file port fixes 2011-08-26 14:25:18 -06:00
Kevin Tew
801e5dbe0e Fix serious clang warnings 2011-08-26 14:04:28 -06:00
Kevin Tew
6ee0c0298a Fix warnings in place.c 2011-08-26 11:46:57 -06:00
Matthew Flatt
290fe066b6 add optional argument to weak-box-value' and ephemeron-value' 2011-08-26 06:24:34 -06:00
Matthew Flatt
ecc058de4a change `raco exe' to disable collection lookup by default 2011-08-25 09:15:30 -06:00
Eli Barzilay
7bb8c6c4b3 New Racket version 5.1.3.6. 2011-08-25 03:50:21 -04:00
Matthew Flatt
ec29a2d85c add `use-collection-link-paths' 2011-08-24 20:08:19 -06:00
Matthew Flatt
84e3ab2a20 add support for collection roots to the collection links file 2011-08-24 20:08:19 -06:00
Kevin Tew
5a30aa89eb Allow TCP sockets and file ports to be communicated across places 2011-08-24 14:33:54 -06:00
Matthew Flatt
a7855e20a8 add `raco link' 
includes a rewrite the "Module Basics" section of the Guide
 2011-08-24 09:07:14 -06:00
Kevin Tew
8782f4c445 send ffi pointers across places 2011-08-23 13:33:06 -06:00
Eli Barzilay
41ae2b57b6 New Racket version 5.1.3.3. 2011-08-21 03:50:12 -04:00
Matthew Flatt
345b06838d FFI: add C arrays and unions 2011-08-20 10:24:32 -06:00
Eli Barzilay
2fbfe341b9 New Racket version 5.1.3.2. 2011-08-20 03:50:14 -04:00
Matthew Flatt
dc1a97cec7 another subprocess fix like d7a996ee0b 2011-08-19 16:56:10 -06:00
Matthew Flatt
d7a996ee0b fix for subprocess management (when places are enabled) 2011-08-19 13:57:28 -06:00
Matthew Flatt
6073ef5ff5 fix `place-break' on a terminated place 2011-08-19 12:52:19 -06:00
Matthew Flatt
0c7b1925a5 change bad code to an assertion that's never needed 2011-08-19 12:52:08 -06:00
Matthew Flatt
d185c2a0df fix optimizer bug 
Also fix missing copy-propagation opportunity
 2011-08-19 07:35:39 -06:00
Matthew Flatt
7eb2042bd9 fix compiler handling of top-/module-level constants 
The JIT and bytecode compiler disagreed on the definition of
"constant". Now there are two levels: "constant" means constant across
all instantiations, and "fixed" means constant for a given instantation.
The JIT uses this distinction to generate direct-primitive calls
or not. (Without the distinction, a direct jump to `reverse' could
be wrong, because `racket/base' might get instantiated with the
JIT disabled or not.)

Also, fixed a bug in the JIT's `vector-set!' code in the case that
the target vector is a top-/module-level reference that is ready,
fixed, or constant.
 2011-08-19 06:32:44 -06:00
Matthew Flatt
1d6c3aa4ea avoid pointless digits when printing single-precision floats 
Closes PR 12118
 2011-08-18 13:21:38 -06:00
Matthew Flatt
69929b7499 configure: make --disable-jit imply --disable-futures 2011-08-18 12:41:36 -06:00
Matthew Flatt
8291fef9de fix optimizer problems with mutability checking, inlining 
The inlining adjustment can sometimes trigger different
inlining than before. (One common benchmark wins; one loses.)
 2011-08-18 12:32:01 -06:00
Matthew Flatt
8eefaba187 fix `regexp-match' performance for short matches on long strings 
More specifically, for a string of length N and a match that
only looks at the first M characters, the complexity of
`regexp-match' is now O(M) instead of O(N). This allows
`regexp-split' to be O(N) for a string instead of O(N^2).

Also, fixed a bug in non-greedy matching that could affect
both long strings and input ports.
 2011-08-18 09:26:09 -06:00
Matthew Flatt
591e08fb10 fix bug in identifier resolution 
Commit 311d55b5cf fixed a shallow bug that masked a deeper
bug in the interaction of local bindings and module-level
bindings. This one fixes the deeper problem, which is that
the recursive resolution that ignores module bindings should
start from the beginning of the wraps, not the wrap after
a module renaming.

Closes PR 12116
 2011-08-16 20:24:08 -06:00
Matthew Flatt
8f27112c83 slightly safer CPP macros 2011-08-16 20:24:00 -06:00
Matthew Flatt
311d55b5cf fix bug that affects `free-identifier=?' 
This bug is in the "amazing that it lurked for years" category,
as well as the "stupid use of C preprocessor" category.
 2011-08-16 10:42:56 -06:00
Vincent St-Amour
02ba6e5d32 Remove trailing whitespace. 2011-08-15 12:19:53 -04:00
Matthew Flatt
8989e810fb fix over-eager bytecode transformation 2011-08-15 09:36:38 -06:00
Matthew Flatt
40197835c9 repair `dynamic-require-for-syntax' 
That primitive is probably a bad idea, but maybe it's useful
until we have separaely loadable parts of a module.
 2011-08-15 07:06:40 -06:00
Eli Barzilay
db1aa1479a New Racket version 5.1.3.1. 2011-08-15 03:50:14 -04:00
Eli Barzilay
05dc8a596d Post-release version for the v5.1.3 release 2011-08-14 08:43:02 -04:00
Matthew Flatt
ca92376381 detect __sync_bool_compare_and_swap() via `configure' 2011-08-12 17:03:28 -06:00
Matthew Flatt
1c04cf1b02 make `port-try-file-lock?' work for Solaris 
Shared locking now allowed only on input port, and exclusive
locking is allowed only on output ports, which allows an implementation
via fcntl(...,F_SETLK,...).
 2011-08-11 13:11:26 -06:00
Eli Barzilay
b27eacd4d0 New Racket version 5.1.2.4. 2011-08-10 03:50:14 -04:00
Matthew Flatt
f646511ca7 first small step toward cross-module optimization 
Mainly propagates constantness for the benefit of keyword
applications, but also propagates simple constants
 2011-08-09 16:33:17 -06:00
Matthew Flatt
5352d670c4 generalize #%variable-reference' and add variable-reference-constant?' 
Use the new functions to make the expansion of keyword applications
to known procedure work  with mutation.
 2011-08-09 16:33:17 -06:00
Matthew Flatt
39a96dd699 add `prop:liberal-define-context' 2011-08-09 16:33:16 -06:00
Matthew Flatt
ff024068d9 fix syntax-taint problem in expander 2011-08-09 16:33:16 -06:00
Matthew Flatt
376a673a0c use internal `equal?'-based table in atomic mode 
for kill safety
 2011-08-09 13:10:41 -06:00
Kevin Tew
22ac1f8c54 Places - check for inter-place pointers 2011-08-08 10:58:22 -06:00
Matthew Flatt
2d4e14dca8 fix a kill-safety problem in the module name resolver 2011-08-06 08:25:36 -06:00
Matthew Flatt
84b9cf6b90 fix non-place-friendly parts of finalizer support 2011-08-04 06:57:32 -06:00
Matthew Flatt
3b8d00029e fix `namespace-attach-module{-declaration}' problem 
Closes PR 12048
 2011-08-02 17:06:12 -06:00
Matthew Flatt
9a5bae277a set self modidx on primitive modules 
which fixes a crash related to `module->namespace'
 2011-08-01 15:16:04 -06:00
Matthew Flatt
ec27cbfd90 fix problems related to futures and is-list flags on pairs 
Although a future thread used an atomic compare-and-swap to
set "is a list" or "not a list" flag on pairs via the
JIT-implemented `list?', the hashing function in the runtime
thread did not; as a result, it might be possible to lose
a hash code due to cache inconsistency (although I'm not
sure it's actually possible, and I couldn't trigger a problem
with a test). Most of the changes are related to using
an atomic compare-and-swap when setting a hash code, as
well as clean-ups to related code. Processor-count tests
avoid using atomic compare-and-swap on uniprocessors, which
might not support the relevant machine instructions.

As significantly, the compare-and-swap operation for the
JIT-implemented `list?' did not actually set flags on
a pair that has a hash code. This could lead to `list?'
tests that were not constant time (but only if the relevant
pair's `eq?' hash code had been used previously).
 2011-07-31 04:43:24 -06:00
Matthew Flatt
e6b4d547c9 fix problem with initialization of tag name array 
The bug to lead to a crash from `(dump-memory-stats)'

Mrege to 5.1.2
 2011-07-27 15:26:48 +01:00
Matthew Flatt
7123996b30 change `eval-jit-enabled' initialization 
to reflect whether JIT compilation is supported by the build
(i.e., set it to #f initially when the JIT is not supported)
 2011-07-27 07:55:05 +01:00
James Swaine
783ee2cf88 Add convenience macros for raising exceptions in C functions which may run on future threads 2011-07-26 16:20:35 -05:00
Matthew Flatt
73b16c8578 don't constant-fold an fx operation if the result is non-fx 
In fact, the result must be an fx on all platforms, since
compilation should be platform-independent.
 2011-07-25 11:30:23 -04:00
Matthew Flatt
cab8e5e0e2 JIT tweaks for `unsafe-{s,u}16vector-set!' 
by making better use of the code that's in place for other
similar operations
 2011-07-25 11:30:22 -04:00
Matthew Flatt
c9d4e0fb8c win32: fix `copy-file' handling of file-exists error 
The specific error reported by CopyFileW doesn't seem
to be documented. It's unclear whether Racket's old test
for ERROR_EXISTS_ALREADY was the wrong choice (as opposed
to ERROR_FILE_EXISTS) or whether some Windows versions
use it; we test for both for now.

Also, improve error reporting when an errno or
GetLastError() value is available.

Closes PR 12074

Merge to 5.1.2
 2011-07-23 22:15:18 -04:00
Matthew Flatt
530bb1b9ba code-inspector fix for top-level code from bytecode 
Merge to 5.1.2
 2011-07-20 14:25:24 -06:00
Matthew Flatt
0d2b08f053 fix validation of top-level define-{syntaxes,values-for-syntax} 
(Only appears in bytecode for non-module code.)

Merge to 5.1.2
 2011-07-20 14:08:30 -06:00
Matthew Flatt
5b8a892fbb fix bug in .zo writing 
The bug showed up in the "racket/embed.rktl" test.

Merge to 5.1.2
 2011-07-20 13:57:52 -06:00
Matthew Flatt
30174b3c81 another try at Mac OS X 10.4 x86 libraries 
Merge to 5.1.2
 2011-07-20 13:26:48 -06:00
Matthew Flatt
d9ae1d048d fix optimizer bug related to `case-lambda' at module level 
The bug triggered a crash on ARM, and probably doesn't
affect other platforms, but I'm not competely sure.

Merge to 5.1.2
 2011-07-19 15:27:50 -06:00
Matthew Flatt
5efe7001d6 fix "block cache" layer to handle allocation failure 2011-07-17 20:08:26 -06:00
Matthew Flatt
92671ab3ea fix source name of built-in modules 
Closes PR 12051

 Merge to 5.1.2
 2011-07-16 21:42:57 -06:00
Matthew Flatt
c4b820e5c5 make `namespace->module' fail on kernel modules for the right reason 2011-07-16 21:36:41 -06:00
Matthew Flatt
701c9666d6 fix printing of namespace with places enabled 
This commit goes with 62acb298bd.
 2011-07-16 21:17:10 -06:00
Matthew Flatt
3f0914080b fix SGC 
Merge to 5.1.2
 2011-07-16 14:30:09 -06:00
Matthew Flatt
9ca07e7d65 minor code clarification 2011-07-16 14:30:09 -06:00
Matthew Flatt
4307bcace5 fix taint behavior of some syntax operations 
`syntax-local-get-shadower' and
 `syntax-make-delta-introducer' both taint their
 results when a given syntax object is tainted
 2011-07-16 08:03:54 -06:00
Kevin Tew
bf4a8ad5c4 create hash codes for place shared objects 2011-07-15 13:06:07 -06:00
Matthew Flatt
62acb298bd places: fix printing of symbol resolved module paths 2011-07-14 21:32:48 -06:00
Matthew Flatt
1c4722eaee remove obsolete reference to '#%mred-kernel 2011-07-14 21:32:47 -06:00
James Swaine
32a3828a2e Made continuation-mark-set-first future-safe. Fixed a rarely occurring bug with lightweight continuation capture for futures. 2011-07-14 19:30:52 -06:00
Matthew Flatt
4d03ffb57d another JIT tweak 2011-07-14 17:07:18 -06:00
Matthew Flatt
825548f5dc tweak JIT to improve and slightly generalize register tracking 2011-07-14 17:07:18 -06:00
Matthew Flatt
87a6b850a7 more info on 10.4 support 2011-07-14 17:07:18 -06:00
Matthew Flatt
0860e62bfa rebuild Mac x86 libraries to work with 10.4 
Merge to 5.1.2
 2011-07-13 10:45:19 -06:00
Vincent St-Amour
09b6616bfa Fix jitting of real->double-flonum. 
Merge to 5.1.2.
 2011-07-11 14:38:02 -04:00
Vincent St-Amour
7dfe1f636f Revert "Add real->double-flonum to the JIT." 
This reverts commit 2afff3d210.

This commit caused real->double-flonum to have a different behavior
when jitted as opposed to interpreted, and caused real->single-flonum
to break in some cases.

Merge to 5.1.2.
 2011-07-11 13:54:06 -04:00
Matthew Flatt
09eab9c3eb fix JIT problem that can break futures 
A recent (weeks-old) JIT change set one of a function's code
 pointers to NULL to indicate that JIT-compilation of the
 function is in progress, but that breaks futures. Set the
 code pointer to a different not-yet-ready function, instead.

 Merge to 5.1.2

 Closes PR 12037
 2011-07-11 07:01:41 -06:00
Eli Barzilay
ac139c5dcb New Racket version 5.1.2.3. 2011-07-11 03:50:16 -04:00
Matthew Flatt
cc6f3f69ab add `port-set-next-location!'; make prompt read handler use it 
Closes PR 12035
 2011-07-10 08:36:47 -06:00
Matthew Flatt
e765231dad fix hash-ref' bugs on immutable eq?'-based tables 
There were two:
   * new: after finding a hash code, the key wasn't
          always checked to be `eq?' to the desired key
   * old: the hash code wan't downshifted by 2, so
          changes in the low two bits (like when a pair
          is determined to start a list) could break
          lookup

 Merge to 5.1.2
 2011-07-09 20:12:59 -06:00
Matthew Flatt
5452a16479 fix arity check on `current-module-name-resolver' 2011-07-09 06:16:11 -06:00
Matthew Flatt
9e694ce5cb share modidx for module imported into multiple phases 2011-07-09 06:16:10 -06:00
Matthew Flatt
50bd06af9a fix dependency 2011-07-09 06:16:10 -06:00
Matthew Flatt
d49a1e1fa1 JIT-inline `path?' 2011-07-09 06:16:10 -06:00
Eli Barzilay
1ba2d85004 New Racket version 5.1.2.2. 2011-07-09 03:50:09 -04:00
Matthew Flatt
b98e1b189a change semantic of internal definitions 
and `letrec-syntaxes+values' --- allowing `let' in place
 of `letrec', which in turn lets the compiler optimize
 away location allocation
 2011-07-08 14:00:41 -06:00
Matthew Flatt
4392ab7636 make weak taint table actually weak 2011-07-08 06:28:43 -06:00
Matthew Flatt
0e95d49ddb allow more expression reordering, especially unsafe ops 2011-07-08 06:23:29 -06:00
Matthew Flatt
65b1a569a0 add `unsafe-list-{ref,set}' 
Use the new functions in `syntax-case'; the benefit is small, and
 it's mostly useful as hint to the optimizer that the operation
 can be dropped if the result isn't used
 2011-07-08 06:23:29 -06:00
Matthew Flatt
848bba80a3 optimizer repair; `unsafe-c{a,d}r' can be dropped 2011-07-08 06:22:40 -06:00
Eli Barzilay
4edbf125a9 New Racket version 5.1.2.1. 2011-07-08 03:50:19 -04:00
Ryan Culpepper
ed7495ce2a Post-release version for the v5.1.2 release 2011-07-07 23:37:55 -06:00
Kevin Tew
8128603b9b Fix non-places build 2011-07-07 17:14:01 -06:00
Kevin Tew
8ed62b715b Fix place-channel place-object race condition 2011-07-07 17:14:00 -06:00
Vincent St-Amour
3b33ed8c81 Abstract out magic numbers in the JIT. 2011-07-07 18:13:50 -04:00
Kevin Tew
68d49e9f59 Add assertions to place-channel tests 2011-07-07 12:38:20 -06:00
Vincent St-Amour
2afff3d210 Add real->double-flonum to the JIT. 2011-07-07 12:04:24 -04:00
Matthew Flatt
9ba6beacf5 fix non-futures, non-places build 2011-07-07 07:25:48 -06:00
Matthew Flatt
62a17da060 JIT-inline list-ref' and list-tail' 2011-07-07 07:25:48 -06:00
Matthew Flatt
a77a2fb3e1 JIT-inline all of the `cXr's 2011-07-07 07:25:48 -06:00