]> gitweb.factorcode.org Git - factor.git/log
factor.git
8 years agotools.image-analyzer.*: wrap read objects in heap-node tuples with their addresses
Björn Lindqvist [Mon, 7 Dec 2015 03:20:43 +0000 (04:20 +0100)]
tools.image-analyzer.*: wrap read objects in heap-node tuples with their addresses

8 years agoDocs: more docs for vocabs in core/
Björn Lindqvist [Sat, 5 Dec 2015 15:34:46 +0000 (16:34 +0100)]
Docs: more docs for vocabs in core/

8 years agotools.test: print the callstack in test failures
Björn Lindqvist [Thu, 3 Dec 2015 11:00:40 +0000 (12:00 +0100)]
tools.test: print the callstack in test failures

So when mason reports a test failure it should include the whole
callstack. That should make it easier to catch the "No suitable
arithmetic method" random error that is hard to debug without a
callstack.

8 years agoVM: refactors the prepare_boot_image method
Björn Lindqvist [Fri, 4 Dec 2015 12:57:57 +0000 (13:57 +0100)]
VM: refactors the prepare_boot_image method

easier to read if the methods compile_all_words and
initialize_all_quotations which are only used once are "inlined"

8 years agobootstrap.*: fix for #1512, load docs after everything else
Björn Lindqvist [Fri, 4 Dec 2015 12:29:29 +0000 (13:29 +0100)]
bootstrap.*: fix for #1512, load docs after everything else

By making it so the help component is loaded almost last, all vocabs are loaded without docs and only then are they loaded. It should make it so you don't get cyclic dependency problems in doc files during bootstrapping.

8 years agobootstrap.stage2: more logging output and fix for #1444
Björn Lindqvist [Fri, 4 Dec 2015 12:16:21 +0000 (13:16 +0100)]
bootstrap.stage2: more logging output and fix for #1444

The error clearing needs to happen almost last so that an error is not
set in the image.

8 years agoFUEL: slightly improved font-locking for alien functions and aliases
Björn Lindqvist [Fri, 4 Dec 2015 11:43:01 +0000 (12:43 +0100)]
FUEL: slightly improved font-locking for alien functions and aliases

8 years agoui.backend.gtk: set the active? to 100 in on-map, should fix #1506
Björn Lindqvist [Thu, 3 Dec 2015 23:57:19 +0000 (00:57 +0100)]
ui.backend.gtk: set the active? to 100 in on-map, should fix #1506

what now happens is that configure events are skipped until the first
on-map signal is fired. seem to work fine in a variety of wm:s. if this
works well we can return active? to be plain boolean.

8 years agoBuild: fix semver_into so that it doesnt misparse versions like "2.3.4"
Björn Lindqvist [Wed, 2 Dec 2015 23:46:21 +0000 (00:46 +0100)]
Build: fix semver_into so that it doesnt misparse versions like "2.3.4"

8 years agosnake-game: refactor and restruture
Sankaranarayanan Viswanathan [Wed, 2 Dec 2015 08:06:48 +0000 (13:36 +0530)]
snake-game: refactor and restruture

8 years agoAdded extra:snake-game
Sankaranarayanan Viswanathan [Tue, 1 Dec 2015 18:03:15 +0000 (23:33 +0530)]
Added extra:snake-game

8 years agocompiler.cfg.intrinsics: fix doc and make tests work on 32bit
Björn Lindqvist [Tue, 1 Dec 2015 12:43:45 +0000 (13:43 +0100)]
compiler.cfg.intrinsics: fix doc and make tests work on 32bit

8 years agoui.gestures: fix gesture docs, #1509
Björn Lindqvist [Tue, 1 Dec 2015 10:01:34 +0000 (11:01 +0100)]
ui.gestures: fix gesture docs, #1509

8 years agoui.*: move some of the gl initing to set-up-window, should fix #1510
Björn Lindqvist [Tue, 1 Dec 2015 03:49:36 +0000 (04:49 +0100)]
ui.*: move some of the gl initing to set-up-window, should fix #1510

it's also more efficent to run the check-extensions "1.0"
require-gl-version code only once when the window is created, rather
than once for each draw

8 years agoBuild: simpler version# parsing logic (parsing gcc --version failed on my system)
Björn Lindqvist [Mon, 30 Nov 2015 22:23:46 +0000 (23:23 +0100)]
Build: simpler version# parsing logic (parsing gcc --version failed on my system)

8 years agocompiler.cfg.intrinsics.slots: refactor + docs & tests
Björn Lindqvist [Sat, 28 Nov 2015 00:55:12 +0000 (01:55 +0100)]
compiler.cfg.intrinsics.slots: refactor + docs & tests

8 years agoui.tools.listener: only wait up to 5 seconds for the listener to start
Björn Lindqvist [Fri, 27 Nov 2015 12:15:17 +0000 (13:15 +0100)]
ui.tools.listener: only wait up to 5 seconds for the listener to start

otherwise unit tests can wait forever if there is an error in listener-thread

8 years agoVM: can't go through the accessor here because it triggers the assert
Björn Lindqvist [Wed, 25 Nov 2015 17:12:55 +0000 (18:12 +0100)]
VM: can't go through the accessor here because it triggers the assert

8 years agoVM: refactor tagged.hpp so it's not dependent on the factor_vm class
Björn Lindqvist [Tue, 24 Nov 2015 15:00:50 +0000 (16:00 +0100)]
VM: refactor tagged.hpp so it's not dependent on the factor_vm class

this way untag<byte_array>(relocation) can be used in code_blocks.hpp

8 years agotools.deploy: small bump in sizes.
John Benediktsson [Mon, 23 Nov 2015 18:42:49 +0000 (10:42 -0800)]
tools.deploy: small bump in sizes.

8 years agocompiler.cfg.builder.blocks: help-lint fix
Björn Lindqvist [Mon, 23 Nov 2015 15:32:01 +0000 (16:32 +0100)]
compiler.cfg.builder.blocks: help-lint fix

8 years agoVM: fixes to use the TAG macro over tagged<object>(obj).type()
Björn Lindqvist [Mon, 23 Nov 2015 01:27:22 +0000 (02:27 +0100)]
VM: fixes to use the TAG macro over tagged<object>(obj).type()

8 years agoVM: refactoring to use the TAG macro directly over the tagged struct
Björn Lindqvist [Sun, 22 Nov 2015 22:07:18 +0000 (23:07 +0100)]
VM: refactoring to use the TAG macro directly over the tagged struct

eg. TAG(obj) is equivalent to tagged<object>(obj).type()

8 years agocompiler.cfg.*: more fixes to pass basic-block on the stack than in a
Björn Lindqvist [Sun, 22 Nov 2015 00:06:11 +0000 (01:06 +0100)]
compiler.cfg.*: more fixes to pass basic-block on the stack than in a
dynamic variable

8 years agocompiler.cfg.*: refactors words to take a block parameter instead of
Björn Lindqvist [Wed, 18 Nov 2015 23:53:46 +0000 (00:53 +0100)]
compiler.cfg.*: refactors words to take a block parameter instead of
using the basic-block dynamic variable

the idea is to make the code easier to follow by limiting the use of
dynamic variables

8 years agocompiler.tree.dead-code.simple: small refactoring and docs
Björn Lindqvist [Wed, 18 Nov 2015 11:35:13 +0000 (12:35 +0100)]
compiler.tree.dead-code.simple: small refactoring and docs

8 years agocompiler.tree.propagation.recursive: better counter-class
Björn Lindqvist [Tue, 17 Nov 2015 09:11:09 +0000 (10:11 +0100)]
compiler.tree.propagation.recursive: better counter-class

8 years agohtml.templates: use unless* instead of or.
John Benediktsson [Fri, 20 Nov 2015 17:15:19 +0000 (09:15 -0800)]
html.templates: use unless* instead of or.

8 years agourls: only create linked-hash if query is f.
John Benediktsson [Thu, 19 Nov 2015 16:50:30 +0000 (08:50 -0800)]
urls: only create linked-hash if query is f.

8 years agocompiler.tree.propagation.info: better code for wrap-interval
Björn Lindqvist [Mon, 16 Nov 2015 11:30:24 +0000 (12:30 +0100)]
compiler.tree.propagation.info: better code for wrap-interval

8 years agourls: fix for the test failure in the yahoo vocab
Björn Lindqvist [Mon, 16 Nov 2015 10:41:21 +0000 (11:41 +0100)]
urls: fix for the test failure in the yahoo vocab

?set-at can't be used in set-query-param because it creates a normal
hashtable and we need a linked-assoc

8 years agostack-checker.known-words: nicer stack-effect declaration code
Björn Lindqvist [Sun, 15 Nov 2015 10:10:03 +0000 (11:10 +0100)]
stack-checker.known-words: nicer stack-effect declaration code

I think it looks better to put the primitives in groups and declare them
that way than one big block.

8 years agourls.encoding: make query>assoc docs have checked example.
John Benediktsson [Wed, 4 Nov 2015 23:16:36 +0000 (15:16 -0800)]
urls.encoding: make query>assoc docs have checked example.

8 years agoyahoo: fix tests for LH{ } change.
John Benediktsson [Wed, 4 Nov 2015 21:35:40 +0000 (13:35 -0800)]
yahoo: fix tests for LH{ } change.

8 years agocgi: better content length parsing.
John Benediktsson [Wed, 4 Nov 2015 21:08:15 +0000 (13:08 -0800)]
cgi: better content length parsing.

8 years agocgi: fix for LH{ } changes.
John Benediktsson [Wed, 4 Nov 2015 21:06:23 +0000 (13:06 -0800)]
cgi: fix for LH{ } changes.

8 years agohttp.server.requests: fix LH{ } in tests.
John Benediktsson [Wed, 4 Nov 2015 21:01:39 +0000 (13:01 -0800)]
http.server.requests: fix LH{ } in tests.

8 years agourls: Fix unit tests for linked-hash change
Doug Coleman [Wed, 4 Nov 2015 18:44:28 +0000 (10:44 -0800)]
urls: Fix unit tests for linked-hash change

8 years agoFactor.app: new icons.
John Benediktsson [Wed, 4 Nov 2015 18:27:13 +0000 (10:27 -0800)]
Factor.app: new icons.

8 years agoxdg: adding a XDG Base Directory Specification.
John Benediktsson [Wed, 4 Nov 2015 18:24:27 +0000 (10:24 -0800)]
xdg: adding a XDG Base Directory Specification.

8 years agoNew icons.
John Benediktsson [Wed, 4 Nov 2015 18:18:21 +0000 (10:18 -0800)]
New icons.

8 years agolinked-assocs: move prettyprint stuff to its own vocab.
John Benediktsson [Wed, 4 Nov 2015 17:16:02 +0000 (09:16 -0800)]
linked-assocs: move prettyprint stuff to its own vocab.

8 years agourls.encoding: fix tests for linked-hash changes.
John Benediktsson [Wed, 4 Nov 2015 16:43:08 +0000 (08:43 -0800)]
urls.encoding: fix tests for linked-hash changes.

8 years agourls.encoding: preserve order of URL query parameters.
John Benediktsson [Wed, 4 Nov 2015 16:36:56 +0000 (08:36 -0800)]
urls.encoding: preserve order of URL query parameters.

8 years agolinked-assocs: give linked-hash "LH{ }" literal pprint.
John Benediktsson [Wed, 4 Nov 2015 16:36:24 +0000 (08:36 -0800)]
linked-assocs: give linked-hash "LH{ }" literal pprint.

8 years agoShow persistent border on right clicking presentation gadgets
Sankaranarayanan Viswanathan [Tue, 3 Nov 2015 03:24:36 +0000 (22:24 -0500)]
Show persistent border on right clicking presentation gadgets

8 years agoRename variables in some stack declarations for consistency.
Alexander Iljin [Tue, 29 Sep 2015 14:48:10 +0000 (17:48 +0300)]
Rename variables in some stack declarations for consistency.

It doesn't look great when a word is called "v/n", but the declared
parameters are "u" and "n". Better to have the parameters "v" and "n".
Also for consistency all input parameters are named "v" for "vector" (or
"u" and "v" in the alphabetical order), and the result vector is always
"w".

8 years agoRevert "hacker-news: no longer need to load urls.secure!"
John Benediktsson [Wed, 4 Nov 2015 15:45:49 +0000 (07:45 -0800)]
Revert "hacker-news: no longer need to load urls.secure!"

This reverts commit 48a0fa5428655992201964f0d3d0e37fab57252b.

8 years agoRevert "Auto load urls.secure when needed"
John Benediktsson [Wed, 4 Nov 2015 15:45:46 +0000 (07:45 -0800)]
Revert "Auto load urls.secure when needed"

This reverts commit bb7740091eac7e88c240f19a92d6e9111aa78533.

8 years agocompiler.cfg.linear-scan.allocation.state: fixes the the 64-bit-o-centric testcases
Björn Lindqvist [Wed, 4 Nov 2015 12:54:14 +0000 (13:54 +0100)]
compiler.cfg.linear-scan.allocation.state: fixes the the 64-bit-o-centric testcases

8 years agocompiler.cfg.linear-scan.allocation.state: help-lint fixes
Björn Lindqvist [Wed, 4 Nov 2015 12:38:59 +0000 (13:38 +0100)]
compiler.cfg.linear-scan.allocation.state: help-lint fixes

8 years agowebbrowser.*: let's open the file in a detached process
Björn Lindqvist [Wed, 4 Nov 2015 10:35:40 +0000 (11:35 +0100)]
webbrowser.*: let's open the file in a detached process

Seems more useful overall and matches how it already works on Windows.

8 years agocolor-picker: small refactoring
Björn Lindqvist [Wed, 4 Nov 2015 09:03:52 +0000 (10:03 +0100)]
color-picker: small refactoring

8 years agomath.hashcodes,python.syntax: bi@ and -> both?
Björn Lindqvist [Wed, 4 Nov 2015 08:25:32 +0000 (09:25 +0100)]
math.hashcodes,python.syntax: bi@ and -> both?

8 years agocompiler.cfg.*: refactoring which removes spill-area-size and
Björn Lindqvist [Mon, 2 Nov 2015 17:40:13 +0000 (18:40 +0100)]
compiler.cfg.*: refactoring which removes spill-area-size and
spill-area-align from cfg

This makes the code for building the stack frame simpler.

8 years agocompiler.cfg.linear-scan.*: make it so cfg is given as an argument to next-spill...
Björn Lindqvist [Sun, 1 Nov 2015 08:57:07 +0000 (09:57 +0100)]
compiler.cfg.linear-scan.*: make it so cfg is given as an argument to next-spill-slot

8 years agoio.files.info.unix.linux: Special characters in mount points are encoded as octal...
Doug Coleman [Tue, 3 Nov 2015 23:55:14 +0000 (15:55 -0800)]
io.files.info.unix.linux: Special characters in mount points are encoded as octal escape sequences.

8 years agovocabs.refresh.monitor: Add exception handling in monitor-loop so that if
Doug Coleman [Tue, 3 Nov 2015 05:36:10 +0000 (21:36 -0800)]
vocabs.refresh.monitor: Add exception handling in monitor-loop so that if
a path isn't a valid vocabulary name then we print an error. Should spam
the console on windows downloads but 1) only for .com not the .exe, 2) we
prefer people to build from source in which case the :Zone.Identifier ADS
will not be present, and 3) it will educate people about ADS hopefully.

If it's annoying we can turn it off.

8 years agoio.monitors.windows: ADS pathnames crashing the monitors threads is the
Doug Coleman [Tue, 3 Nov 2015 05:19:12 +0000 (21:19 -0800)]
io.monitors.windows: ADS pathnames crashing the monitors threads is the
real culprit. Back out change and print out the crash and error message instead of ignoring
errors!

8 years agoio.files.windows: ALIEN: -1 puts an expired alien in the image. -1 <alien>
Doug Coleman [Tue, 3 Nov 2015 05:18:49 +0000 (21:18 -0800)]
io.files.windows: ALIEN: -1 puts an expired alien in the image. -1 <alien>
it is then.

8 years agoio.monitors.windows: Strip the :Zone.Identifier from pathnames reported
Doug Coleman [Mon, 2 Nov 2015 03:58:05 +0000 (19:58 -0800)]
io.monitors.windows: Strip the :Zone.Identifier from pathnames reported
by the file change api.

8 years agowindows.kernel32, io.files.windows: Add support for listing alternate data
Doug Coleman [Mon, 2 Nov 2015 02:17:51 +0000 (18:17 -0800)]
windows.kernel32, io.files.windows: Add support for listing alternate data
streams, e.g. file paths like foo.txt:Zone.Identifier.

8 years agohacker-news: no longer need to load urls.secure!
John Benediktsson [Fri, 30 Oct 2015 04:29:33 +0000 (21:29 -0700)]
hacker-news: no longer need to load urls.secure!

8 years agoAuto load urls.secure when needed
Sankaranarayanan Viswanathan [Thu, 29 Oct 2015 01:13:51 +0000 (21:13 -0400)]
Auto load urls.secure when needed

8 years agoui.tools.listener: revert the commit that added UP/DOWN keybindings to
Björn Lindqvist [Fri, 30 Oct 2015 01:16:37 +0000 (02:16 +0100)]
ui.tools.listener: revert the commit that added UP/DOWN keybindings to
cycle history

8 years agoarrays: don't need to use slots.private.
John Benediktsson [Thu, 29 Oct 2015 16:05:12 +0000 (09:05 -0700)]
arrays: don't need to use slots.private.

8 years agoarrays: can use the length>> accessor instead
Björn Lindqvist [Thu, 29 Oct 2015 05:36:38 +0000 (06:36 +0100)]
arrays: can use the length>> accessor instead

8 years agoui.backend.cocoa: fix for the silly error I made in 7bc1009973d26579dba598ceee2f10444...
Björn Lindqvist [Thu, 29 Oct 2015 03:41:36 +0000 (04:41 +0100)]
ui.backend.cocoa: fix for the silly error I made in 7bc1009973d26579dba598ceee2f10444ecb6749

8 years agoarrays: make it so pair? doesn't call length, fixes #1496
Björn Lindqvist [Wed, 28 Oct 2015 23:48:18 +0000 (00:48 +0100)]
arrays: make it so pair? doesn't call length, fixes #1496

8 years agoFUEL: better font-locking of a bunch of syntax
Björn Lindqvist [Wed, 28 Oct 2015 23:44:08 +0000 (00:44 +0100)]
FUEL: better font-locking of a bunch of syntax

Like for PREDICATE:, GAME: and SINGLETONS:

8 years agoui.tools.listener: only recall if the lexer-error is aborted (#375)
Björn Lindqvist [Wed, 28 Oct 2015 15:31:13 +0000 (16:31 +0100)]
ui.tools.listener: only recall if the lexer-error is aborted (#375)

8 years agoui.*: make active? an integer 0-100
Björn Lindqvist [Tue, 27 Oct 2015 18:00:28 +0000 (19:00 +0100)]
ui.*: make active? an integer 0-100

A UI backend can delay setting active? to 100 until the windows actually
is visible. That way, some redundant redraws that happen when the window
isn't visible is supressed. It makes the Windows and GTK UI startup a
little faster.

8 years agoui.backend.*: the offscreen buffer hooks appear unused - let's delete!
Björn Lindqvist [Mon, 26 Oct 2015 02:47:49 +0000 (03:47 +0100)]
ui.backend.*: the offscreen buffer hooks appear unused - let's delete!

8 years agounicode.normalize: cleanup usings.
John Benediktsson [Tue, 27 Oct 2015 01:25:35 +0000 (18:25 -0700)]
unicode.normalize: cleanup usings.

8 years agoimages.loader.gdiplus: both BGRA and BGRX formats are supported
Björn Lindqvist [Sun, 25 Oct 2015 20:53:30 +0000 (21:53 +0100)]
images.loader.gdiplus: both BGRA and BGRX formats are supported

8 years agoslides: some useful keybindings added
Björn Lindqvist [Sun, 25 Oct 2015 15:33:53 +0000 (16:33 +0100)]
slides: some useful keybindings added

8 years agotalks: let's move all the talks vocabs to its own hierarchy
Björn Lindqvist [Sat, 24 Oct 2015 22:59:59 +0000 (00:59 +0200)]
talks: let's move all the talks vocabs to its own hierarchy

8 years agoslides: add a title to the slides windows, looks a little more polished that way
Björn Lindqvist [Sat, 24 Oct 2015 21:05:13 +0000 (23:05 +0200)]
slides: add a title to the slides windows, looks a little more polished that way

8 years agofonts: fix name of sans-serif font.
John Benediktsson [Sun, 25 Oct 2015 20:32:02 +0000 (13:32 -0700)]
fonts: fix name of sans-serif font.

8 years agohacker-news: require urls.secure for https urls.
John Benediktsson [Sun, 25 Oct 2015 20:31:16 +0000 (13:31 -0700)]
hacker-news: require urls.secure for https urls.

8 years agoslides: fix the vocab to use relative sizes
Björn Lindqvist [Sat, 24 Oct 2015 12:57:54 +0000 (14:57 +0200)]
slides: fix the vocab to use relative sizes

8 years agohelp.stylesheet: make sizes calculated relative to the default-font-size
Björn Lindqvist [Sat, 24 Oct 2015 12:44:30 +0000 (14:44 +0200)]
help.stylesheet: make sizes calculated relative to the default-font-size

This makes Factors ui much easier to adapt to hidpi screens. You change
the default-font-size and elements should grow or shrink in proportion.

8 years agofonts: turn hardcoded font names and sizes into constants
Björn Lindqvist [Sat, 24 Oct 2015 12:29:12 +0000 (14:29 +0200)]
fonts: turn hardcoded font names and sizes into constants

8 years agoui.tools.listener: recall the last command on parse errors
Björn Lindqvist [Fri, 23 Oct 2015 15:59:04 +0000 (17:59 +0200)]
ui.tools.listener: recall the last command on parse errors

8 years agoFix clang version pattern
Arkady Rost [Thu, 22 Oct 2015 08:06:00 +0000 (11:06 +0300)]
Fix clang version pattern

'Apple LLVM version 6.0 (clang-600.0.57) (based on LLVM 3.5svn)' was parsed incorrectly.

8 years agoui.tools.listener.tests: fix tests, manifest needs to be set and vocab used
Björn Lindqvist [Thu, 22 Oct 2015 21:33:46 +0000 (23:33 +0200)]
ui.tools.listener.tests: fix tests, manifest needs to be set and vocab used

8 years agowindows.uniscribe: add SSA_FALLBACK and SSA_TAB to the flags, should fix #860
Björn Lindqvist [Thu, 22 Oct 2015 21:14:55 +0000 (23:14 +0200)]
windows.uniscribe: add SSA_FALLBACK and SSA_TAB to the flags, should fix #860

8 years agoui.tools.listener: call interactor-finish before try-parse #375
Björn Lindqvist [Thu, 22 Oct 2015 08:42:10 +0000 (10:42 +0200)]
ui.tools.listener: call interactor-finish before try-parse #375

Also changes the parse error handling in try-parse. If a parse error
occurs it is wrapped in a small quotation to defer handling it until the
quotation is ran.

8 years agoui.tools.listener: tried to simplify stream-read-quot and handle-interactive
Björn Lindqvist [Thu, 22 Oct 2015 04:57:27 +0000 (06:57 +0200)]
ui.tools.listener: tried to simplify stream-read-quot and handle-interactive

8 years agoui.backend.gtk: on-key-press and on-key-release can be merged into one
Björn Lindqvist [Tue, 20 Oct 2015 15:13:40 +0000 (17:13 +0200)]
ui.backend.gtk: on-key-press and on-key-release can be merged into one
on-key-press/release handler

8 years agofile-picker.linux: more correct to use gint instead of int.
John Benediktsson [Wed, 21 Oct 2015 23:57:45 +0000 (16:57 -0700)]
file-picker.linux: more correct to use gint instead of int.

8 years agotools.ps.linux: recover from missing file in ps-cmdline, fixes #1120
Björn Lindqvist [Tue, 20 Oct 2015 14:55:48 +0000 (16:55 +0200)]
tools.ps.linux: recover from missing file in ps-cmdline, fixes #1120

8 years agoui.gadgets.incremental: fix so that the scrolled of gadgets are ungrafted
Björn Lindqvist [Sun, 18 Oct 2015 15:35:26 +0000 (17:35 +0200)]
ui.gadgets.incremental: fix so that the scrolled of gadgets are ungrafted

8 years agoui.gadgets.tests: clear the layout queue, so test dont interfere with
Björn Lindqvist [Sat, 17 Oct 2015 13:58:59 +0000 (15:58 +0200)]
ui.gadgets.tests: clear the layout queue, so test dont interfere with
other tests

8 years agoui.gadgets.tests: fix the test (layout-queue is a vector now)
Björn Lindqvist [Fri, 16 Oct 2015 20:28:31 +0000 (22:28 +0200)]
ui.gadgets.tests: fix the test (layout-queue is a vector now)

8 years agoui.gadgets.incremental: trivial scrolling, maybe solves #1488?
Björn Lindqvist [Fri, 16 Oct 2015 08:55:48 +0000 (10:55 +0200)]
ui.gadgets.incremental: trivial scrolling, maybe solves #1488?

8 years agoui.gadgets.buttons: memoize the roll button themes pens
Björn Lindqvist [Fri, 16 Oct 2015 07:20:29 +0000 (09:20 +0200)]
ui.gadgets.buttons: memoize the roll button themes pens

It reduces memory usage by about 10% for the #1488 test on my machine
and doesnt seem to have any bad side-effects.

8 years agoui: the layout-queue is more efficient as a vector than a dlist
Björn Lindqvist [Thu, 15 Oct 2015 16:34:49 +0000 (18:34 +0200)]
ui: the layout-queue is more efficient as a vector than a dlist

Improves memory usage in the #1488 example from about 20,4% to 17,4% on
my machine.

8 years agoui.gadgets: ensure that layout-later yields if the queue gets to large
Björn Lindqvist [Thu, 15 Oct 2015 14:29:49 +0000 (16:29 +0200)]
ui.gadgets: ensure that layout-later yields if the queue gets to large

This improves memory utilization a lot for #1488. Otherwise the queue
grows to two million entries before being processed.

8 years agodlists: new word dlist-length for getting the length
Björn Lindqvist [Thu, 15 Oct 2015 13:16:57 +0000 (15:16 +0200)]
dlists: new word dlist-length for getting the length