]> gitweb.factorcode.org Git - factor.git/log
factor.git
12 years agoSet up listener's manifest so that words defined in "scratchpad" take precedence...
Slava Pestov [Tue, 3 Jan 2012 05:18:52 +0000 (23:18 -0600)]
Set up listener's manifest so that words defined in "scratchpad" take precedence over other vocabularies and do not raise a restart. This matches the behavior of IN: and USE: in files, where the IN: takes precedence. Also remove unused clear-manifest word. Fixes #407

12 years agogdbm.ffi: fix library name on Linux
Slava Pestov [Tue, 3 Jan 2012 04:44:43 +0000 (22:44 -0600)]
gdbm.ffi: fix library name on  Linux

12 years agoextra:math.erato: Remove txt files; vocab was moved to basis.
Doug Coleman [Mon, 2 Jan 2012 20:14:03 +0000 (12:14 -0800)]
extra:math.erato: Remove txt files; vocab was moved to basis.

12 years agoRevert "ui.gadgets.books: needs help.markup.private."
John Benediktsson [Fri, 30 Dec 2011 05:38:05 +0000 (21:38 -0800)]
Revert "ui.gadgets.books: needs help.markup.private."

This reverts commit dc9aa68961e5d2c4e5762b09e1abebea70d5c97b.

12 years agounicode: Fix some $vocab-links in docs.
John Benediktsson [Fri, 30 Dec 2011 01:32:05 +0000 (17:32 -0800)]
unicode: Fix some $vocab-links in docs.

12 years agomath.vectors: fix some $examples in docs.
John Benediktsson [Fri, 30 Dec 2011 01:31:55 +0000 (17:31 -0800)]
math.vectors: fix some $examples in docs.

12 years agoui.gadgets.books: needs help.markup.private.
John Benediktsson [Fri, 30 Dec 2011 01:19:48 +0000 (17:19 -0800)]
ui.gadgets.books: needs help.markup.private.

12 years agoui: fix error in docs (USE: ui.gadgets.books).
John Benediktsson [Fri, 30 Dec 2011 00:33:16 +0000 (16:33 -0800)]
ui: fix error in docs (USE: ui.gadgets.books).

12 years agosplitting.monotonic: Don't throw away the last result. Fixes #462.
Doug Coleman [Tue, 27 Dec 2011 17:43:32 +0000 (11:43 -0600)]
splitting.monotonic: Don't throw away the last result. Fixes #462.

12 years agovm: actually code_block_pic is a fine frame type
Joe Groff [Thu, 22 Dec 2011 05:12:54 +0000 (21:12 -0800)]
vm: actually code_block_pic is a fine frame type

12 years agovm: also smoke-check callstack after pic update
Joe Groff [Thu, 22 Dec 2011 02:31:37 +0000 (18:31 -0800)]
vm: also smoke-check callstack after pic update

12 years agovm: flush after print_obj so it's usable from gdb
Joe Groff [Wed, 21 Dec 2011 21:25:28 +0000 (13:25 -0800)]
vm: flush after print_obj so it's usable from gdb

12 years agovm: fashion brigade
Joe Groff [Wed, 21 Dec 2011 21:25:05 +0000 (13:25 -0800)]
vm: fashion brigade

12 years agovm: walk stack on inline_cache_miss if PIC_DEBUG=1
Joe Groff [Wed, 21 Dec 2011 21:24:35 +0000 (13:24 -0800)]
vm: walk stack on inline_cache_miss if PIC_DEBUG=1
Reveals inline cache stack coherency bug behind #460

12 years agovm: load mega cache object before popping frame
Joe Groff [Tue, 20 Dec 2011 22:38:01 +0000 (14:38 -0800)]
vm: load mega cache object before popping frame
Fixes #457 harder

12 years agocpu.x86.32.bootstrap: oops, meant bootstrap-cell
Joe Groff [Tue, 20 Dec 2011 04:55:07 +0000 (20:55 -0800)]
cpu.x86.32.bootstrap: oops, meant bootstrap-cell

12 years agocpu.x86.32.bootstrap: update unwind-native-frames
Joe Groff [Tue, 20 Dec 2011 02:42:14 +0000 (18:42 -0800)]
cpu.x86.32.bootstrap: update unwind-native-frames
The arguments passed from C are in a different position now.

12 years agovm: unwind-native-frames is special
Joe Groff [Tue, 20 Dec 2011 00:54:25 +0000 (16:54 -0800)]
vm: unwind-native-frames is special
Fixes #457

12 years agocombinators.smart: Make map-reduce-outputs a macro. To be a combinator, we need a...
Doug Coleman [Mon, 19 Dec 2011 19:12:37 +0000 (11:12 -0800)]
combinators.smart: Make map-reduce-outputs a macro. To be a combinator, we need a full stack-checker rewrite.

12 years agomath.parser: hex> etc. shouldn't take radix prefix
Joe Groff [Sat, 17 Dec 2011 01:20:05 +0000 (17:20 -0800)]
math.parser: hex> etc. shouldn't take radix prefix
Fixes #453

12 years agofix link when extracting the feed address
Samuel Tardieu [Wed, 14 Dec 2011 22:18:26 +0000 (23:18 +0100)]
fix link when extracting the feed address

If the <link rel="alternate" href="..."/> was not the first
<link/> attribute under the <feed/> tag, the bad one was extracted
as the corresponding site URL.

12 years agoadd tag-named-with-attr to xml.traversal
Samuel Tardieu [Wed, 14 Dec 2011 22:05:57 +0000 (23:05 +0100)]
add tag-named-with-attr to xml.traversal

12 years agofastcgi: er, linux + macosx = unix
Joe Groff [Sat, 17 Dec 2011 00:06:46 +0000 (16:06 -0800)]
fastcgi: er, linux + macosx = unix

12 years agofastcgi: only load on unix until #454 is resolved
Joe Groff [Sat, 17 Dec 2011 00:02:53 +0000 (16:02 -0800)]
fastcgi: only load on unix until #454 is resolved

12 years agotools.deploy.shaker: strip globals harder
Joe Groff [Fri, 16 Dec 2011 23:56:59 +0000 (15:56 -0800)]
tools.deploy.shaker: strip globals harder
Separate the list of strippable globals in "stripped" and "cleared". Completely remove the former set, but only reset to 'f' the latter. Fixes #447.

12 years agothreads: update test
Joe Groff [Wed, 14 Dec 2011 21:45:53 +0000 (13:45 -0800)]
threads: update test

12 years agoRevert "pubsubhubbub support by samueltardieu". Fixes #455.
Joe Groff [Wed, 14 Dec 2011 21:13:18 +0000 (13:13 -0800)]
Revert "pubsubhubbub support by samueltardieu". Fixes #455.

This reverts commit e3483c9c0383dbc89888f58fa471b8777fe35ce1.

12 years agofix compiler.codegen tests
Joe Groff [Wed, 14 Dec 2011 20:51:12 +0000 (12:51 -0800)]
fix compiler.codegen tests

12 years agotools.deploy.shaker: clear MEMO: caches
Joe Groff [Wed, 14 Dec 2011 20:44:40 +0000 (12:44 -0800)]
tools.deploy.shaker: clear MEMO: caches

12 years agofix help-lint failures
Joe Groff [Wed, 14 Dec 2011 19:33:17 +0000 (11:33 -0800)]
fix help-lint failures

12 years agocpu.x86.64: define jit-load-vm as a no-op
Joe Groff [Wed, 14 Dec 2011 19:33:05 +0000 (11:33 -0800)]
cpu.x86.64: define jit-load-vm as a no-op
This was the signal handler prolog code can be shared completely between x86.32 and x86.64.

12 years agovm: cleanups
Joe Groff [Wed, 14 Dec 2011 17:52:43 +0000 (09:52 -0800)]
vm: cleanups

12 years agooops, handler still need to jit-load-vm on x86-32
Joe Groff [Wed, 14 Dec 2011 04:09:07 +0000 (20:09 -0800)]
oops, handler still need to jit-load-vm on x86-32

12 years agovm: don't scribble memory handling leaf signals
Joe Groff [Wed, 14 Dec 2011 03:36:05 +0000 (19:36 -0800)]
vm: don't scribble memory handling leaf signals

12 years agocpu.x86.*: clean up signal handler prolog code
Joe Groff [Wed, 14 Dec 2011 03:31:59 +0000 (19:31 -0800)]
cpu.x86.*: clean up signal handler prolog code
Also bump the signal handler frame sizes up again since it looks like we shrank them too much.

12 years agovm: factor out word_stack_frame_size
Joe Groff [Wed, 14 Dec 2011 01:04:35 +0000 (17:04 -0800)]
vm: factor out word_stack_frame_size

12 years agovm: get rid of now-trivial innermost_stack_frame
Joe Groff [Wed, 14 Dec 2011 00:21:53 +0000 (16:21 -0800)]
vm: get rid of now-trivial innermost_stack_frame

12 years agocpu.x86: reduce compiled frames by 2 cells too
Joe Groff [Tue, 13 Dec 2011 23:41:40 +0000 (15:41 -0800)]
cpu.x86: reduce compiled frames by 2 cells too

12 years agosimplify function prologs
Joe Groff [Tue, 13 Dec 2011 23:13:32 +0000 (15:13 -0800)]
simplify function prologs

12 years agovm: fix types after rebase
Joe Groff [Tue, 13 Dec 2011 20:27:39 +0000 (12:27 -0800)]
vm: fix types after rebase

12 years agovm: replace remaining stack_frame-based logic
Joe Groff [Wed, 7 Dec 2011 01:53:52 +0000 (17:53 -0800)]
vm: replace remaining stack_frame-based logic

12 years agovm: remove _reversed from callstack iterator names
Joe Groff [Tue, 6 Dec 2011 23:00:02 +0000 (15:00 -0800)]
vm: remove _reversed from callstack iterator names
Now that they're the only game in town we can give them the short names

12 years agovm: remove iterate_callstack_object
Joe Groff [Tue, 6 Dec 2011 22:57:09 +0000 (14:57 -0800)]
vm: remove iterate_callstack_object

12 years agovm: callstack_reversed for callstack_to_array
Joe Groff [Tue, 6 Dec 2011 22:51:41 +0000 (14:51 -0800)]
vm: callstack_reversed for callstack_to_array

12 years agoremove iterate_callstack now that it's unused
Joe Groff [Tue, 6 Dec 2011 17:41:50 +0000 (09:41 -0800)]
remove iterate_callstack now that it's unused

12 years agovm: callstack_reversed in call_frame_scrubber
Joe Groff [Tue, 6 Dec 2011 17:28:00 +0000 (09:28 -0800)]
vm: callstack_reversed in call_frame_scrubber

12 years agovm: stage code block map fixup properly for GC
Joe Groff [Tue, 6 Dec 2011 17:14:56 +0000 (09:14 -0800)]
vm: stage code block map fixup properly for GC
Don't update the map until the very last thing, and pass untranslated addresses to the iterator functors. Somewhat complicated by the fact that, for startup_fixup, the map is initialized with fixed-up values, so the fixup functor needs a flag indicating whether it operates with a fixed or unfixed code heap map.

12 years agovm: callstack_reversed in code_block_visitor
Joe Groff [Mon, 5 Dec 2011 22:18:42 +0000 (14:18 -0800)]
vm: callstack_reversed in code_block_visitor

12 years agovm: iterate_callstack_reversed in slot_visitor
Joe Groff [Mon, 5 Dec 2011 20:53:34 +0000 (12:53 -0800)]
vm: iterate_callstack_reversed in slot_visitor

12 years agovm: reversed callstack iteration
Joe Groff [Fri, 2 Dec 2011 17:11:34 +0000 (09:11 -0800)]
vm: reversed callstack iteration

12 years agoWIP verify_callstack function
Joe Groff [Thu, 1 Dec 2011 01:56:24 +0000 (17:56 -0800)]
WIP verify_callstack function
dumps callstack in reverse order so i can visually inspect that it matches the old forward walking logic

12 years agovm: method to calculate frame size for address
Joe Groff [Wed, 30 Nov 2011 20:57:35 +0000 (12:57 -0800)]
vm: method to calculate frame size for address
If we're before the prolog or or in a leaf procedure, the stack frame is really a leaf frame created by the signal handler, with a known fixed size, instead of the real stack frame (if any) with a different size.

12 years agovm: don't suffer null return addresses
Joe Groff [Wed, 30 Nov 2011 20:55:57 +0000 (12:55 -0800)]
vm: don't suffer null return addresses

12 years agocpu.*.bootstrap: return address for (set-context)
Joe Groff [Wed, 30 Nov 2011 20:52:06 +0000 (12:52 -0800)]
cpu.*.bootstrap: return address for (set-context)
Instead of pushing zero as the fake return address for a set-context call, push an address that points back into the primitive so that callstack walking can match the frame back to the subprimitive.

12 years agovm: frame size for signal handler code blocks
Joe Groff [Wed, 30 Nov 2011 20:39:21 +0000 (12:39 -0800)]
vm: frame size for signal handler code blocks

12 years agovm: initialize all_blocks set after initial fixup
Joe Groff [Wed, 30 Nov 2011 04:19:13 +0000 (20:19 -0800)]
vm: initialize all_blocks set after initial fixup

12 years agovm: make all_blocks a set of cells
Joe Groff [Wed, 30 Nov 2011 04:18:51 +0000 (20:18 -0800)]
vm: make all_blocks a set of cells
gcc was doing bad things assuming the pointer given to upper_bound was aligned

12 years agovm: verifier shouldn't check for leftovers
Joe Groff [Sat, 26 Nov 2011 03:44:43 +0000 (19:44 -0800)]
vm: verifier shouldn't check for leftovers
It doesn't work after things have been freed

12 years agovm: pack frame size into code block header bits
Joe Groff [Sat, 26 Nov 2011 03:23:07 +0000 (19:23 -0800)]
vm: pack frame size into code block header bits

12 years agocompiler.constants: remove unused constants
Joe Groff [Sat, 26 Nov 2011 02:58:40 +0000 (18:58 -0800)]
compiler.constants: remove unused constants

12 years agovm: store stack frame size in code blocks
Joe Groff [Sat, 26 Nov 2011 02:58:21 +0000 (18:58 -0800)]
vm: store stack frame size in code blocks
Change modify-code-heap primitive so it takes a sixth element in each array for the frame size.

12 years agomath.vectors: add higher-order stack effect to vif
Joe Groff [Wed, 14 Dec 2011 03:46:44 +0000 (19:46 -0800)]
math.vectors: add higher-order stack effect to vif

12 years agovm: erase code blocks from all_blocks during sweep
Joe Groff [Tue, 13 Dec 2011 20:28:09 +0000 (12:28 -0800)]
vm: erase code blocks from all_blocks during sweep
Fixes #437

12 years agovm: reinitialize all_blocks from scratch after gc
Joe Groff [Tue, 13 Dec 2011 03:05:48 +0000 (19:05 -0800)]
vm: reinitialize all_blocks from scratch after gc
Faster and more likely to be correct than trying to fixup the existing set

12 years agovm: only verify all_blocks set if it's invalid
Joe Groff [Tue, 13 Dec 2011 03:04:56 +0000 (19:04 -0800)]
vm: only verify all_blocks set if it's invalid
Verifying on every sample is way too slow to be practical when debugging bootstrap.

12 years agovm: add compact-gc command to factorbug
Joe Groff [Tue, 13 Dec 2011 03:03:42 +0000 (19:03 -0800)]
vm: add compact-gc command to factorbug

12 years agocpu.x86: enable POPCNT with a flag
Joe Groff [Tue, 13 Dec 2011 00:43:49 +0000 (16:43 -0800)]
cpu.x86: enable POPCNT with a flag
If bootstrapped with '-enable-popcnt' and the host CPU supports it, enable the POPCNT-based fixnum-bit-count intrinsic. Fixes #436.

12 years agotools.deploy.shaker: more timid global stripping
Joe Groff [Tue, 13 Dec 2011 00:08:25 +0000 (16:08 -0800)]
tools.deploy.shaker: more timid global stripping
Strip globals by replacing their value with "f". The list of globals to strip was written with the assumption that the stripped globals would get restored like normal hashtable keys when accessed. This isn't valid anymore with the globals optimization, since creating a new box breaks the foldable invariant between a global name and its associated box. Need to go through the list of globals to strip and find the ones that actually can be entirely removed as opposed to just reset to f. Fixes #445. Fixes #446.

12 years agoadded B: (break for parsing words) and appropriate documentation
rien [Sun, 11 Dec 2011 01:59:22 +0000 (20:59 -0500)]
added B: (break for parsing words) and appropriate documentation

12 years agofixed effect based on inner behavior and outer context
rien [Sun, 11 Dec 2011 00:47:34 +0000 (19:47 -0500)]
fixed effect based on inner behavior and outer context

12 years agomath.vectors.simd.intrinsics: fix deploy bug
Joe Groff [Wed, 7 Dec 2011 20:46:52 +0000 (12:46 -0800)]
math.vectors.simd.intrinsics: fix deploy bug
If not called with a known-at-compile-time SIMD type, the intrinsic definitions were leaving behind runtime quotations that referenced the c-array words after c-type information had been stripped, causing deployed applications that used SIMD to fail.

12 years agohelp-lint fixes for recent changes
Doug Coleman [Wed, 7 Dec 2011 20:26:35 +0000 (12:26 -0800)]
help-lint fixes for recent changes

12 years agocommand-line: don't pull in tools.errors
Joe Groff [Wed, 7 Dec 2011 18:52:58 +0000 (10:52 -0800)]
command-line: don't pull in tools.errors
tools.errors brings the debugger with it, which bloats deploy image size

12 years agoregexp.parser: fix failing test
Joe Groff [Wed, 7 Dec 2011 05:11:58 +0000 (21:11 -0800)]
regexp.parser: fix failing test

12 years agocompiler: fix failing test
Joe Groff [Wed, 7 Dec 2011 05:11:44 +0000 (21:11 -0800)]
compiler: fix failing test

12 years agolazy.lists: lfrom-by documentation fix
Samuel Tardieu [Tue, 6 Dec 2011 18:09:22 +0000 (19:09 +0100)]
lazy.lists: lfrom-by documentation fix

Also add a unit test for lfrom-by.

12 years agoui.backend.gtk: don't die if the icon doesn't exist, partially fixes #369
Philipp Brüschweiler [Wed, 16 Nov 2011 11:39:05 +0000 (12:39 +0100)]
ui.backend.gtk: don't die if the icon doesn't exist, partially fixes #369

12 years agocompiler: The non-optimizing compiler is too slow with smart combinators in the compi...
Doug Coleman [Mon, 5 Dec 2011 22:26:14 +0000 (14:26 -0800)]
compiler: The non-optimizing compiler is too slow with smart combinators in the compiler. Speed up bootstrap by not using them.

12 years agohelp.html: change back to the old naming scheme because we have too many links to...
Slava Pestov [Sun, 4 Dec 2011 21:46:13 +0000 (13:46 -0800)]
help.html: change back to the old naming scheme because we have too many links to existing doc articles all over the net

12 years agomemoize: Don't use identity hashtables for macro expansion; it uses all the memory...
Doug Coleman [Thu, 1 Dec 2011 18:14:16 +0000 (10:14 -0800)]
memoize: Don't use identity hashtables for macro expansion; it uses all the memory. Fixes #439.

12 years agobootstrap/stage1: Turn off parser-quiet? flag sooner so the files loaded in the USING...
Doug Coleman [Thu, 1 Dec 2011 07:14:14 +0000 (23:14 -0800)]
bootstrap/stage1: Turn off parser-quiet? flag sooner so the files loaded in the USING: list in stage2 print a loading message.

12 years agobasis/*: Pare down some using lists
Doug Coleman [Thu, 1 Dec 2011 07:09:49 +0000 (23:09 -0800)]
basis/*: Pare down some using lists

12 years agotools.profiler.sampling: Don't use calendar because it pulls in ratios
Doug Coleman [Thu, 1 Dec 2011 06:55:42 +0000 (22:55 -0800)]
tools.profiler.sampling: Don't use calendar because it pulls in ratios

12 years agomemoize: Use identity hashtables for macros
Doug Coleman [Thu, 1 Dec 2011 06:49:34 +0000 (22:49 -0800)]
memoize: Use identity hashtables for macros

12 years agocombinators.smart: Tell the stack checker that the return values from inputs/ouputs...
Doug Coleman [Thu, 1 Dec 2011 06:47:20 +0000 (22:47 -0800)]
combinators.smart: Tell the stack checker that the return values from inputs/ouputs are input-parameters
Fixes #435

12 years agocore/basis: trim down using lists
Doug Coleman [Thu, 1 Dec 2011 00:02:37 +0000 (16:02 -0800)]
core/basis: trim down using lists

12 years agofurnace.auth.providers.couchdb: fix manifest
Joe Groff [Thu, 1 Dec 2011 01:58:29 +0000 (17:58 -0800)]
furnace.auth.providers.couchdb: fix manifest

12 years agocombinators.smart: Use an identity memoize on inputs/outputs word to speed up the...
Doug Coleman [Wed, 30 Nov 2011 22:32:25 +0000 (14:32 -0800)]
combinators.smart: Use an identity memoize on inputs/outputs word to speed up the non-optimizing compiler

12 years agomemoize: implement identity memoize, trim using lists so bootstrap continues to work
Doug Coleman [Wed, 30 Nov 2011 22:28:40 +0000 (14:28 -0800)]
memoize: implement identity memoize, trim using lists so bootstrap continues to work

12 years agoio.encodings.iana: remove values
Joe Groff [Tue, 29 Nov 2011 21:22:23 +0000 (13:22 -0800)]
io.encodings.iana: remove values

12 years agoxml.entities: remove USING: values
Joe Groff [Tue, 29 Nov 2011 20:21:08 +0000 (12:21 -0800)]
xml.entities: remove USING: values

12 years agovm: fix msvc warning
Joe Groff [Tue, 29 Nov 2011 20:19:17 +0000 (12:19 -0800)]
vm: fix msvc warning

12 years agoio.sockets: remove unneeded USING: values
Joe Groff [Tue, 29 Nov 2011 17:46:51 +0000 (09:46 -0800)]
io.sockets: remove unneeded USING: values

12 years agoremove values vocab
Joe Groff [Tue, 29 Nov 2011 05:26:06 +0000 (21:26 -0800)]
remove values vocab

12 years agocombinators.smart: Fix infer-known; should be 100% now.
Doug Coleman [Tue, 29 Nov 2011 04:54:17 +0000 (20:54 -0800)]
combinators.smart: Fix infer-known; should be 100% now.
It's stack-checker bugs all the way down.

12 years agocombinators.smart: Implement reduce-outputs using compose, fix some stack effects...
Doug Coleman [Tue, 29 Nov 2011 00:18:51 +0000 (16:18 -0800)]
combinators.smart: Implement reduce-outputs using compose, fix some stack effects, add using

12 years agoio.directories.unix: remove redundant using
Joe Groff [Tue, 29 Nov 2011 04:36:46 +0000 (20:36 -0800)]
io.directories.unix: remove redundant using

12 years agoio: implement stream protocols on f
Joe Groff [Tue, 29 Nov 2011 04:36:34 +0000 (20:36 -0800)]
io: implement stream protocols on f
Fixes deployed applications that write to output-stream

12 years agoio.files.info: fix circularity
Joe Groff [Tue, 29 Nov 2011 03:37:01 +0000 (19:37 -0800)]
io.files.info: fix circularity
Remove the undocumented and redundant "copy-file-and-info" word, since "copy-file" does the same thing, and the circular dependency on "io.directories" was causing some deploys to fail

12 years agofastcgi: fix load and compile errors
Joe Groff [Tue, 29 Nov 2011 02:56:00 +0000 (18:56 -0800)]
fastcgi: fix load and compile errors