]>
gitweb.factorcode.org Git - factor.git/log
Doug Coleman [Tue, 4 Aug 2015 23:27:41 +0000 (16:27 -0700)]
GNUMakefile: Save entire git id.
Doug Coleman [Tue, 4 Aug 2015 23:26:53 +0000 (16:26 -0700)]
factor.cmd: Save entire gid id.
Doug Coleman [Tue, 4 Aug 2015 22:17:07 +0000 (15:17 -0700)]
report: Add build number to mason reports.
Doug Coleman [Tue, 4 Aug 2015 22:07:01 +0000 (15:07 -0700)]
io.files: fix docs
Doug Coleman [Tue, 4 Aug 2015 22:05:38 +0000 (15:05 -0700)]
bootstrap.image.upload: Upload build images as well for posterity.
Doug Coleman [Tue, 4 Aug 2015 22:03:37 +0000 (15:03 -0700)]
io.files: Add change-file-lines and change-file-contents words.
Doug Coleman [Tue, 4 Aug 2015 21:40:40 +0000 (14:40 -0700)]
tools.image-analyzer.gc-info: Deferred words don't have any gc-info. Fix the unit test to account for this. Fixes #1394.
Doug Coleman [Tue, 4 Aug 2015 19:43:24 +0000 (12:43 -0700)]
contexts.hpp: Increase callstack reserve space to 16k for Mac64. Fixes issue #1419.
Doug Coleman [Tue, 4 Aug 2015 19:23:55 +0000 (12:23 -0700)]
opengl: Temporary (?) fix for the ui. Issue #1379.
Throwing gl errors is not helpful because the ui gets messed up anyway. Print the error and continue. Maybe we should add a section in the F3 error window for opengl errors instead of printing them in the global stdout.
- We could also use gl-error on every platform except for MacOSX 10.11
- We could fix the underlying gl drawing commands that cause this error.
Doug Coleman [Tue, 4 Aug 2015 00:38:13 +0000 (17:38 -0700)]
urls: allow + in protocols, e.g. git+https://
John Benediktsson [Tue, 4 Aug 2015 14:49:50 +0000 (07:49 -0700)]
vm: minor fixes.
Björn Lindqvist [Tue, 4 Aug 2015 12:52:35 +0000 (14:52 +0200)]
VM: fix silly compile error
Björn Lindqvist [Tue, 4 Aug 2015 12:51:48 +0000 (14:51 +0200)]
kernel.tests: address of fault is now in the 3rd element, so these tests
need to be updated
Björn Lindqvist [Tue, 4 Aug 2015 12:27:14 +0000 (14:27 +0200)]
VM: bump stack_reserved from 1kb to 4kb
The added test case fails with a doulbe fault, which appears to be
caused by a stack overflow in the code that tries to handle the stack
overflow. So bumping it to 4096 bytes should give the code enough stack
space to play with.
Björn Lindqvist [Mon, 3 Aug 2015 22:06:57 +0000 (00:06 +0200)]
VM: merge full_collector.hpp into full_collector.cpp
Björn Lindqvist [Mon, 3 Aug 2015 21:48:08 +0000 (23:48 +0200)]
VM: new method context::address_to_error
Björn Lindqvist [Mon, 3 Aug 2015 21:45:20 +0000 (23:45 +0200)]
VM: signal_error and fp_trap_error not needed, call general_error directly
Björn Lindqvist [Mon, 3 Aug 2015 21:06:02 +0000 (23:06 +0200)]
VM: a bunch of methods on factor_vm that can be converted into free
functions
I think that makes it easier to see whats going on than having
everything added to factor_vm
Björn Lindqvist [Mon, 3 Aug 2015 13:07:48 +0000 (15:07 +0200)]
VM: lets merge entry_points.hpp with vm.hpp, so you have one less header
file to worry about
Björn Lindqvist [Sat, 1 Aug 2015 15:30:20 +0000 (17:30 +0200)]
VM: you can insert the write_barrier() the first time you interate the
blocks, so you don't need to do it twice
Björn Lindqvist [Sat, 1 Aug 2015 14:47:04 +0000 (16:47 +0200)]
VM: new method visit_instruction_operands(), it replaces the instruction
operand iteration code in compaction.cpp and image.cpp
Björn Lindqvist [Fri, 31 Jul 2015 18:45:11 +0000 (20:45 +0200)]
VM: refactors slot_visitor by removing a lot of one-use methods
I think it's easier to see the code flow if small method bodies that are
only called once are inlined into visit_all_roots() instead
Björn Lindqvist [Fri, 31 Jul 2015 18:31:47 +0000 (20:31 +0200)]
VM: word_stack_frame_p() is not used and find_all_quotations() is
trivial so those methods can be removed
Björn Lindqvist [Thu, 30 Jul 2015 23:34:53 +0000 (01:34 +0200)]
VM: using compute_external_address everywhere instead of store_external_address
Björn Lindqvist [Thu, 30 Jul 2015 23:14:38 +0000 (01:14 +0200)]
VM: new method compute_external_address
Björn Lindqvist [Wed, 29 Jul 2015 21:01:42 +0000 (23:01 +0200)]
VM: remove the collect_compact_code_impl(), collect_compact_impl() can
be used instead
this might slow down collect_growing_heap(), but heap growth happens so
rarely that it shouldn't matter
Björn Lindqvist [Wed, 29 Jul 2015 15:35:34 +0000 (17:35 +0200)]
VM: init_code_heap is trivial and only used once, so let's remove it
Björn Lindqvist [Wed, 29 Jul 2015 14:53:04 +0000 (16:53 +0200)]
VM: object_compaction_updater -> lambda func
Björn Lindqvist [Wed, 29 Jul 2015 14:33:59 +0000 (16:33 +0200)]
VM: another opportunity to use a lambda function over a struct
Björn Lindqvist [Wed, 29 Jul 2015 12:25:34 +0000 (14:25 +0200)]
VM: use a function update_relocation to replace the
code_block_compaction_relocation_visitor struct
Doug Coleman [Mon, 3 Aug 2015 19:23:08 +0000 (12:23 -0700)]
json.reader: Fix json reader for empty files to parse as H{ }. Add path>json word.
Rename read-jsons
Doug Coleman [Sat, 1 Aug 2015 21:34:04 +0000 (14:34 -0700)]
tools.deploy: increase image size for linux32
Doug Coleman [Sat, 1 Aug 2015 21:29:12 +0000 (14:29 -0700)]
syntax docs: Fix octal example.
John Benediktsson [Sat, 1 Aug 2015 14:59:21 +0000 (07:59 -0700)]
syntax: fix example in syntax-floats.
John Benediktsson [Sat, 1 Aug 2015 04:57:37 +0000 (21:57 -0700)]
stack-checker: too tired to figure out which using is required but not used.
John Benediktsson [Sat, 1 Aug 2015 03:41:46 +0000 (20:41 -0700)]
stack-checker: using cleanup.
John Benediktsson [Fri, 31 Jul 2015 01:16:31 +0000 (18:16 -0700)]
math.parser: make float-parse:point a fixnum.
John Benediktsson [Fri, 31 Jul 2015 00:59:21 +0000 (17:59 -0700)]
math.parser: some inline, some fixnum+fast, some disable tests.
Jon Harper [Wed, 24 Jun 2015 21:14:24 +0000 (23:14 +0200)]
math.parser: support >bin and >oct for floats
This is for symmetry with "0o1p0", "0b1p0", bin> and hex> which
all already work
Jon Harper [Tue, 23 Jun 2015 22:00:25 +0000 (00:00 +0200)]
math.parser: add tests/docs for floats 0b 0x bin> float> since they work
Jon Harper [Wed, 24 Jun 2015 21:22:59 +0000 (23:22 +0200)]
docs: hex floats exponents became mandatory
Jon Harper [Tue, 23 Jun 2015 21:09:51 +0000 (23:09 +0200)]
math.parser, don't take infinite time to parse huge exponents
Jon Harper [Tue, 23 Jun 2015 21:08:57 +0000 (23:08 +0200)]
math.parser: don't lose precision in make-float-bin-exponent
John Benediktsson [Thu, 30 Jul 2015 17:35:51 +0000 (10:35 -0700)]
reddit: use unix-time>timestamp.
John Benediktsson [Thu, 30 Jul 2015 17:34:49 +0000 (10:34 -0700)]
calendar.format: don't need unix-time>relative-time.
unix-time>timestamp relative-time
John Benediktsson [Thu, 30 Jul 2015 16:41:58 +0000 (09:41 -0700)]
math.ratios: moving to core.
Doug Coleman [Thu, 30 Jul 2015 16:10:48 +0000 (09:10 -0700)]
hacker-news: fix typo with descendants. use unix-time>relative-time
Doug Coleman [Thu, 30 Jul 2015 16:10:10 +0000 (09:10 -0700)]
calendar.elapsed: Support relative times in the future with "3 hours hence" etc.
Add unix-time>relative-time word for convenience.
John Benediktsson [Thu, 30 Jul 2015 15:56:55 +0000 (08:56 -0700)]
calendar.elapsed: make generic for more convenience.
John Benediktsson [Thu, 30 Jul 2015 15:54:02 +0000 (08:54 -0700)]
reddit: use calendar.elapsed for relative-time.
John Benediktsson [Thu, 30 Jul 2015 15:46:44 +0000 (08:46 -0700)]
hacker-news: add relative timestamp on posts.
John Benediktsson [Thu, 30 Jul 2015 15:16:34 +0000 (08:16 -0700)]
reddit: stop using tuples and just use the JSON.
Nice-looking tuples are nice but reddit keeps adding fields to their objects
and breaking our from-slots. We could hide the error, or ignore the extra
fields but this is probably better for now.
John Benediktsson [Thu, 30 Jul 2015 15:10:18 +0000 (08:10 -0700)]
calendar.elapsed: adding some elapsed time words.
Doug Coleman [Thu, 30 Jul 2015 06:51:26 +0000 (23:51 -0700)]
io.files.windows: Need the SLOT: file from io.sockets.secure.openssl
Doug Coleman [Thu, 30 Jul 2015 04:16:16 +0000 (21:16 -0700)]
hacker-news: Fix using, fix bug printing HN company posts. Refactor a bit
Doug Coleman [Thu, 30 Jul 2015 04:05:20 +0000 (21:05 -0700)]
hacker-news: use new hackernews api on firebase.
Doug Coleman [Thu, 30 Jul 2015 04:04:31 +0000 (21:04 -0700)]
benchmark.fasta: rename random.
Björn Lindqvist [Wed, 29 Jul 2015 10:51:05 +0000 (12:51 +0200)]
compiler.cfg.utilities: make it so block>cfg initializes spill-area-align
John Benediktsson [Wed, 29 Jul 2015 03:21:12 +0000 (20:21 -0700)]
fix another doc using.
John Benediktsson [Wed, 29 Jul 2015 03:20:40 +0000 (20:20 -0700)]
fix doc using.
John Benediktsson [Wed, 29 Jul 2015 03:06:59 +0000 (20:06 -0700)]
fix some help-lint errors.
John Benediktsson [Wed, 29 Jul 2015 02:33:24 +0000 (19:33 -0700)]
io.backend.unix.macosx: change to << require >>.
John Benediktsson [Wed, 29 Jul 2015 02:30:20 +0000 (19:30 -0700)]
command-line: rename file to path for user-init-error which is kinda exactly like a source-file-error.
John Benediktsson [Wed, 29 Jul 2015 02:23:54 +0000 (19:23 -0700)]
unix: don't need << >>.
John Benediktsson [Wed, 29 Jul 2015 02:23:43 +0000 (19:23 -0700)]
math.floats.env: don't need << >>.
John Benediktsson [Wed, 29 Jul 2015 02:14:14 +0000 (19:14 -0700)]
ui: cleanup some usings.
John Benediktsson [Wed, 29 Jul 2015 02:14:01 +0000 (19:14 -0700)]
ui.images: don't need to << >>.
John Benediktsson [Wed, 29 Jul 2015 02:13:43 +0000 (19:13 -0700)]
ui.commands: don't need to \ .
John Benediktsson [Wed, 29 Jul 2015 02:13:29 +0000 (19:13 -0700)]
io.backend.unix.macosx: change USE: to require.
John Benediktsson [Wed, 29 Jul 2015 02:13:13 +0000 (19:13 -0700)]
alien.libraries: don't need << >>.
John Benediktsson [Wed, 29 Jul 2015 01:14:20 +0000 (18:14 -0700)]
disjoint-sets: some cleanup.
John Benediktsson [Wed, 29 Jul 2015 01:10:24 +0000 (18:10 -0700)]
json.reader: fix segfault when parsing non-JSON documents.
John Benediktsson [Wed, 29 Jul 2015 01:02:57 +0000 (18:02 -0700)]
json.reader: enforce json> takes a string.
Björn Lindqvist [Tue, 28 Jul 2015 22:44:06 +0000 (00:44 +0200)]
compiler.cfg.*: tests against #1308
Björn Lindqvist [Tue, 28 Jul 2015 21:18:01 +0000 (23:18 +0200)]
compiler.cfg.ssa.destruction.coalescing: refactor and maybe a fix for #1308
the eliminatable-copy? word is the fix. the previous code only checked
that the registers had the same register class, but you also need to
check that they dont have the same representation size. because a copy
from double-rep -> double-2-rep is not eliminatable
Björn Lindqvist [Tue, 28 Jul 2015 21:15:21 +0000 (23:15 +0200)]
compiler.cfg.*: some extra unit tests for the compiler
Björn Lindqvist [Tue, 28 Jul 2015 18:39:51 +0000 (20:39 +0200)]
compiler.cfg.ssa.destruction.coalescing: simpler code for setting up the
initial leader-map and class-element-map
Björn Lindqvist [Mon, 27 Jul 2015 21:44:59 +0000 (23:44 +0200)]
cpu.x86: define %alien-invoke and %alien-indirect using %alien-assembly
Björn Lindqvist [Mon, 27 Jul 2015 09:35:19 +0000 (11:35 +0200)]
Docs: various more compiler-related docs
Björn Lindqvist [Mon, 27 Jul 2015 09:26:48 +0000 (11:26 +0200)]
compiler.cfg.representations.selection: use a union instead of a generic
to see if an insn is peephole-optimizable + test
Björn Lindqvist [Mon, 27 Jul 2015 09:22:17 +0000 (11:22 +0200)]
compiler.tree.*: new docs
Björn Lindqvist [Mon, 27 Jul 2015 09:11:49 +0000 (11:11 +0200)]
compiler.cfg.*: even more docs
Björn Lindqvist [Sun, 26 Jul 2015 18:40:50 +0000 (20:40 +0200)]
compiler.tree.escape-analysis.*: new doc files
Björn Lindqvist [Sun, 26 Jul 2015 18:37:24 +0000 (20:37 +0200)]
compiler.tree.escape-analysis.allocations: stub doc file from comments
Björn Lindqvist [Sun, 26 Jul 2015 18:32:12 +0000 (20:32 +0200)]
compiler.cfg.*: a bunch of docs updates
Björn Lindqvist [Sun, 26 Jul 2015 16:14:59 +0000 (18:14 +0200)]
compiler.cfg.representations.selection: some initial docs from comments
Björn Lindqvist [Sun, 26 Jul 2015 15:56:29 +0000 (17:56 +0200)]
compiler.cfg.linear-scan.live-intervals: simpler code for finding the
sync points in the cfg
Björn Lindqvist [Thu, 23 Jul 2015 08:11:24 +0000 (10:11 +0200)]
compiler.cfg.instructions: ##read -> read-insn and ##write -> write-insn
Björn Lindqvist [Thu, 23 Jul 2015 08:08:10 +0000 (10:08 +0200)]
compiler.cfg.instructions: rename ##allocation to allocation-insn,
because it's nicer if all instruction unions have the -insn suffix and
the ## prefix is reserved for tuples
Doug Coleman [Mon, 27 Jul 2015 19:41:02 +0000 (12:41 -0700)]
math.statistics: fix docs
Doug Coleman [Mon, 27 Jul 2015 16:53:10 +0000 (09:53 -0700)]
core: rename parse-call( to parse-call-paren
Doug Coleman [Mon, 27 Jul 2015 16:52:31 +0000 (09:52 -0700)]
math.statistics: clean up some weird stack effects
John Benediktsson [Mon, 27 Jul 2015 15:42:42 +0000 (08:42 -0700)]
ui.gadgets.panes: don't use extract-keys, don't clone twice in specified-font.
John Benediktsson [Mon, 27 Jul 2015 15:41:01 +0000 (08:41 -0700)]
ui.gadgets.sliders: rename slider-pen-tuple to slider-pen.
John Benediktsson [Mon, 27 Jul 2015 15:40:30 +0000 (08:40 -0700)]
ui.gadgets.grids: rename grid-layout-tuple to grid-layout.
John Benediktsson [Mon, 27 Jul 2015 04:20:55 +0000 (21:20 -0700)]
bootstrap: cleanup usings.
Jon Harper [Sun, 26 Jul 2015 19:11:29 +0000 (21:11 +0200)]
math.integers, comment and simplify bignum/f
change the "while" that could only execute once to "when"
change the f/loop word name to "mantissa-and-guard" since it's what it
computes
change the check against 2^53 to be explicit
Jon Harper [Sun, 26 Jul 2015 16:07:10 +0000 (18:07 +0200)]
math.integers, bignum/f, improve performance.
This changes avoids looping many times if the denominator is a power of
2. After this change, the implementation matches the linked sbcl
algorithm. This was probably a mistake done when porting the algorithm.
Basically, as an optimization, all trailing zeros are removed from the
base2 representation of the denominator to have smaller bignums to
divide. But the previous factor implementation didn't take this into
account when making the initial guess of the shift of the numerator to
obtain a result in the range [2^54-1,2^53]. The loop would then correct
the initial guess by a factor of 2 at each iteration, so it would run as
many iteration as the denominator base2 power reduction, instead of only
a few times. For pathological cases, the speed up is huge (10^4):
1 1000 2^ bignum/f
Jon Harper [Sun, 26 Jul 2015 15:53:42 +0000 (17:53 +0200)]
math.integers, fix bignum/f for results in ]0x1.0p-1022,0x0.4p-1022]
Doug Coleman [Sun, 26 Jul 2015 19:24:23 +0000 (12:24 -0700)]
math.vectors.conversion: [foo] for words that make quots is ok, but [[foo]] is overboard. change to ([foo])