]> gitweb.factorcode.org Git - factor.git/log
factor.git
12 years agovariables: fix for recent initial-value change.
John Benediktsson [Mon, 14 Nov 2011 18:25:28 +0000 (10:25 -0800)]
variables: fix for recent initial-value change.

12 years agoui.tools.listener: removing unnecessary spawn.
John Benediktsson [Mon, 14 Nov 2011 18:16:15 +0000 (10:16 -0800)]
ui.tools.listener: removing unnecessary spawn.

12 years agoimages.viewer.prettyprint: mark not loaded/tested
Joe Groff [Mon, 14 Nov 2011 07:09:18 +0000 (23:09 -0800)]
images.viewer.prettyprint: mark not loaded/tested
Prevent mason from loading it and complaining that it can't write gadgets to a text file.

12 years agowebbrowser.linux: switching to xdg-open.
John Benediktsson [Mon, 14 Nov 2011 00:49:25 +0000 (16:49 -0800)]
webbrowser.linux: switching to xdg-open.

12 years agowebbrowser: changing open-url to use open-file everywhere.
John Benediktsson [Mon, 14 Nov 2011 00:31:52 +0000 (16:31 -0800)]
webbrowser: changing open-url to use open-file everywhere.

12 years agomath.vectors.simd: don't try distance on int types
Joe Groff [Mon, 14 Nov 2011 00:09:30 +0000 (16:09 -0800)]
math.vectors.simd: don't try distance on int types
"v- norm" doesn't work right for uint-4s because of underflow, but who's going to call "distance" on a uint vector?

12 years agomath.vectors.simd: unencrypt check-optimizer
Joe Groff [Sun, 13 Nov 2011 23:55:05 +0000 (15:55 -0800)]
math.vectors.simd: unencrypt check-optimizer
And make it report real information about what values were tested and how they failed

12 years agocode beautification
Joe Groff [Sun, 13 Nov 2011 23:45:03 +0000 (15:45 -0800)]
code beautification

12 years agomath.vectors.simd: add vcount to the tests
Joe Groff [Sun, 13 Nov 2011 08:02:40 +0000 (00:02 -0800)]
math.vectors.simd: add vcount to the tests
Also add a base case to (approx=) for integer returns. vcount returns an integer even for float vector inputs, but the test machinery isn't smart enough to figure that out and uses (approx=) on the outputs anyway.

12 years agomath.vectors.simd.intrinsics: fix double-2 vcount
Joe Groff [Sun, 13 Nov 2011 07:19:49 +0000 (23:19 -0800)]
math.vectors.simd.intrinsics: fix double-2 vcount
Turns out we use MOVMSKPS for double vectors too. Gotta save that extra instruction byte!

12 years agocompiler.tree.propagation: missed using
Joe Groff [Sun, 13 Nov 2011 03:29:17 +0000 (19:29 -0800)]
compiler.tree.propagation: missed using

12 years agocompiler: more peephole tests for shl/sar
Joe Groff [Sun, 13 Nov 2011 03:29:01 +0000 (19:29 -0800)]
compiler: more peephole tests for shl/sar

12 years agocompiler: tighten (simd-select) output class
Joe Groff [Sun, 13 Nov 2011 01:00:39 +0000 (17:00 -0800)]
compiler: tighten (simd-select) output class
Ints and smaller (or shorts and smaller on 32-bit) can be assumed to be fixnums. Fixes #393

12 years agomath.vectors.simd: make vcount work for all ints
Joe Groff [Sun, 13 Nov 2011 00:53:07 +0000 (16:53 -0800)]
math.vectors.simd: make vcount work for all ints

12 years agomath.vectors.simd.intrinsics: better fake vgetmask
Joe Groff [Sat, 12 Nov 2011 23:37:25 +0000 (15:37 -0800)]
math.vectors.simd.intrinsics: better fake vgetmask
More accurate simulacrum of PMOVMSKB/MOVMSKPS/MOVMSKPD for non-intrinsic definition

12 years agomath.bitwise: don't count the same bits over&over
Joe Groff [Sat, 12 Nov 2011 07:36:21 +0000 (23:36 -0800)]
math.bitwise: don't count the same bits over&over

12 years agomath.bitwise: hotrod nonintrinsic fixnum-bit-count
Joe Groff [Sat, 12 Nov 2011 06:49:09 +0000 (22:49 -0800)]
math.bitwise: hotrod nonintrinsic fixnum-bit-count

12 years agocompiler: add intrinsic for PMOVMSKB/MOVMSKP[SD]
Joe Groff [Sat, 12 Nov 2011 06:47:54 +0000 (22:47 -0800)]
compiler: add intrinsic for PMOVMSKB/MOVMSKP[SD]
Combined with a fast bit-count this will let us rice byte-counting.

12 years agonew module images.viewer.prettyprint
Joe Groff [Sun, 13 Nov 2011 20:10:23 +0000 (12:10 -0800)]
new module images.viewer.prettyprint
Pretty-prints image objects as images.

12 years agovm: fashion police
Joe Groff [Sun, 13 Nov 2011 06:39:28 +0000 (22:39 -0800)]
vm: fashion police

12 years agovm: set stdin_loop cancellation state
Joe Groff [Sat, 12 Nov 2011 22:49:20 +0000 (14:49 -0800)]
vm: set stdin_loop cancellation state
Make it asynchronous so it doesn't wait till the next syscall to cancel.

12 years agovm: close_console before exiting anywhere
Joe Groff [Sat, 12 Nov 2011 22:47:32 +0000 (14:47 -0800)]
vm: close_console before exiting anywhere
Add a factor_vm::exit function that does close_console() before ::exit-ing

12 years agovm: close the console when exiting
Philipp Brüschweiler [Sat, 12 Nov 2011 21:06:47 +0000 (22:06 +0100)]
vm: close the console when exiting

This prevents an endless loop caused by the nvidia drivers on linux.
See http://www.nvnews.net/vbulletin/showthread.php?t=164619

12 years agoclasses.struct.vectored: update for recent slots change
Slava Pestov [Sun, 13 Nov 2011 06:41:08 +0000 (22:41 -0800)]
classes.struct.vectored: update for recent slots change

12 years agovariants: update for recent slots change
Slava Pestov [Sun, 13 Nov 2011 06:40:55 +0000 (22:40 -0800)]
variants: update for recent slots change

12 years agoNow that #foo and ##foo are symbols we can remove a bunch of \
Slava Pestov [Sun, 13 Nov 2011 06:04:26 +0000 (22:04 -0800)]
Now that #foo and ##foo are symbols we can remove a bunch of \

12 years agovm: typo in os-windows.hpp
Joe Groff [Sun, 13 Nov 2011 03:23:44 +0000 (19:23 -0800)]
vm: typo in os-windows.hpp

12 years agoui.operations: fix stack names.
John Benediktsson [Sun, 13 Nov 2011 02:48:03 +0000 (18:48 -0800)]
ui.operations: fix stack names.

12 years agoio.directories: fix delete-file on windows, add unit test
Doug Coleman [Sun, 13 Nov 2011 02:27:33 +0000 (18:27 -0800)]
io.directories: fix delete-file on windows, add unit test

12 years agoassocs: 5-8% performance improvement to assoc>map. Fixes #69.
John Benediktsson [Sun, 13 Nov 2011 00:56:22 +0000 (16:56 -0800)]
assocs: 5-8% performance improvement to assoc>map.  Fixes #69.

12 years agoFix bootstrap
Slava Pestov [Sun, 13 Nov 2011 00:43:16 +0000 (16:43 -0800)]
Fix bootstrap

12 years agowebbrowser: open url's in your webbrowser.
John Benediktsson [Sun, 13 Nov 2011 00:18:52 +0000 (16:18 -0800)]
webbrowser: open url's in your webbrowser.

12 years agoformatting: improve summary.txt.
John Benediktsson [Sun, 13 Nov 2011 00:18:39 +0000 (16:18 -0800)]
formatting: improve summary.txt.

12 years agohashtables: small cleanup
Slava Pestov [Sat, 12 Nov 2011 23:35:19 +0000 (15:35 -0800)]
hashtables: small cleanup

12 years agoPrevent predicate engine words from showing up in usage. results
Slava Pestov [Sat, 12 Nov 2011 22:50:52 +0000 (14:50 -0800)]
Prevent predicate engine words from showing up in usage. results

12 years agoFix ambiguity between 'initial: f' and no initial value in a slot spec. Fixes #382
Slava Pestov [Sat, 12 Nov 2011 22:48:00 +0000 (14:48 -0800)]
Fix ambiguity between 'initial: f' and no initial value in a slot spec. Fixes #382

12 years agoeditors: wait 300ms for the run-process to possibly fail so errors are reported if...
Doug Coleman [Sat, 12 Nov 2011 23:13:34 +0000 (15:13 -0800)]
editors: wait 300ms for the run-process to possibly fail so errors are reported if the editor is missing or broken

12 years agoeditors.jedit: return "jedit" when looking for jedit path
Doug Coleman [Sat, 12 Nov 2011 22:36:35 +0000 (14:36 -0800)]
editors.jedit: return "jedit" when looking for jedit path

12 years agofactor.sh: Remove the redundant list of commands
Doug Coleman [Sat, 12 Nov 2011 22:05:21 +0000 (14:05 -0800)]
factor.sh: Remove the redundant list of commands

12 years agocpu.*.bootstrap: clear faulting flag directly
Joe Groff [Sat, 12 Nov 2011 20:57:24 +0000 (12:57 -0800)]
cpu.*.bootstrap: clear faulting flag directly
No reason to go through a safepoint in unwind-native-frames really.

12 years agovm: more defense against multi-faulting
Joe Groff [Fri, 11 Nov 2011 21:29:46 +0000 (13:29 -0800)]
vm: more defense against multi-faulting
* Clear faulting_p from a safepoint rather than inside general_error, because jumping into unwind-native-frames could blow up.
* Handle multiple faults from fatal_error by breakpointing. Is there anything else we can safely do at that point?
* Verify memory protection faults in the top half of the signal handlers because signal dispatch could fault. Treat memory faults during gc or fep as fatal errors.
* Add a function factor_vm::abort() that restores the default SIGABRT handler and ::abort()s. Use it from fatal_error() so we get useful context from gdb and so the user gets feedback from the system crash reporter that Factor blew up and didn't just disappear.
* In factorbug(), don't proceed with .s .r .c if it would be unsafe to do so.
* Don't pile on signals if we've already called fatal_error().

12 years agocompiler.cfg: Rename <##foo> to ##foo, in the low-level IR
Doug Coleman [Sat, 12 Nov 2011 03:48:38 +0000 (19:48 -0800)]
compiler.cfg: Rename <##foo> to ##foo, in the low-level IR

12 years agobenchmark.udp-echo0: Fix the memory leak.
Doug Coleman [Sat, 12 Nov 2011 01:58:45 +0000 (17:58 -0800)]
benchmark.udp-echo0: Fix the memory leak.

12 years agoio.sockets.windows: plug receive-from memory leak
Joe Groff [Sat, 12 Nov 2011 01:56:04 +0000 (17:56 -0800)]
io.sockets.windows: plug receive-from memory leak
Should fix #387.

12 years agomath.parser: undo some inlining to shrink size
Joe Groff [Fri, 11 Nov 2011 23:49:53 +0000 (15:49 -0800)]
math.parser: undo some inlining to shrink size
For #318.

12 years agovm: "edges" command for factorbug
Joe Groff [Fri, 11 Nov 2011 18:12:36 +0000 (10:12 -0800)]
vm: "edges" command for factorbug

12 years agotools.deploy: label small-enough? tests
Joe Groff [Fri, 11 Nov 2011 03:34:36 +0000 (19:34 -0800)]
tools.deploy: label small-enough? tests
Put a no-op "vocab name" drop in the small-enough? tests so that it's immediately apparent from build farm errors which vocabs are deploying over the limit.

12 years agotools.profiler.sampling: better test quots
Joe Groff [Fri, 11 Nov 2011 03:28:51 +0000 (19:28 -0800)]
tools.profiler.sampling: better test quots
Bump up the iota lengths so the tests don't sporadically fail with no data.

12 years agohashtables: fix docs for new-key@, and note that quadratic probing is used now.
John Benediktsson [Fri, 11 Nov 2011 00:20:13 +0000 (16:20 -0800)]
hashtables: fix docs for new-key@, and note that quadratic probing is used now.

12 years agobootstrap.primitives: remove counting "profiling"
Joe Groff [Fri, 11 Nov 2011 00:04:42 +0000 (16:04 -0800)]
bootstrap.primitives: remove counting "profiling"

12 years agovm: strip out call-counting profiler
Joe Groff [Fri, 11 Nov 2011 00:00:47 +0000 (16:00 -0800)]
vm: strip out call-counting profiler
This makes the separate "code" and "entry_point" fields in word and quotation redundant, so also remove them to reclaim an additional cell per word and quotation object, which should help with #318.

12 years agoinit signals or console after stage2 init
Joe Groff [Thu, 10 Nov 2011 23:14:03 +0000 (15:14 -0800)]
init signals or console after stage2 init
A signal during early bootstrap is certainly an error, and the environment isn't there yet for Factor's signal handlers to work, so it makes no sense to install Factor's signal handlers beforehand. They'll just crash.

12 years agovm: don't SIGQUIT on unix or Ctrl-Break on windows
Joe Groff [Thu, 10 Nov 2011 21:46:44 +0000 (13:46 -0800)]
vm: don't SIGQUIT on unix or Ctrl-Break on windows
This gives us an escape hatch for when things go way south and ^C can't be handled. Also unmask SIGQUIT and SIGTERM from the stdin_loop thread so the thread doesn't keep the process alive after those signals.

12 years agohashtables: allow re-using deleted tombstones. Fixes #381.
John Benediktsson [Thu, 10 Nov 2011 20:37:14 +0000 (12:37 -0800)]
hashtables: allow re-using deleted tombstones. Fixes #381.

12 years agosequences.parser: fix load error
Joe Groff [Thu, 10 Nov 2011 20:32:52 +0000 (12:32 -0800)]
sequences.parser: fix load error

12 years agokill tools.profiler.counting and ui.tools.profiler
Joe Groff [Thu, 10 Nov 2011 20:32:43 +0000 (12:32 -0800)]
kill tools.profiler.counting and ui.tools.profiler
Load tools.profiler.sampling from bootstrap/tools instead.

12 years agocombinators: fix docs for case.
John Benediktsson [Thu, 10 Nov 2011 18:00:24 +0000 (10:00 -0800)]
combinators: fix docs for case.

12 years agonew module typed.namespaces: get/set + type check
Joe Groff [Thu, 10 Nov 2011 18:46:56 +0000 (10:46 -0800)]
new module typed.namespaces: get/set + type check

12 years agofactor.sh: print more docs in usage()
Doug Coleman [Thu, 10 Nov 2011 06:56:59 +0000 (22:56 -0800)]
factor.sh: print more docs in usage()

12 years agoio.ports: clean up
Slava Pestov [Thu, 10 Nov 2011 05:39:21 +0000 (00:39 -0500)]
io.ports: clean up

12 years agoio.ports: make read-step inline
Joe Groff [Thu, 10 Nov 2011 03:59:46 +0000 (19:59 -0800)]
io.ports: make read-step inline

12 years agoslice boa -> <slice-unsafe>
Joe Groff [Thu, 10 Nov 2011 03:22:09 +0000 (19:22 -0800)]
slice boa -> <slice-unsafe>
If you didn't define it, don't boa it

12 years agoio: clean up some mess
Joe Groff [Thu, 10 Nov 2011 03:11:04 +0000 (19:11 -0800)]
io: clean up some mess

12 years agoio: fix each-block-slice typo
Joe Groff [Thu, 10 Nov 2011 02:04:46 +0000 (18:04 -0800)]
io: fix each-block-slice typo

12 years agoio: more hot-rodding for #376
Joe Groff [Thu, 10 Nov 2011 01:21:44 +0000 (17:21 -0800)]
io: more hot-rodding for #376
* Change the interface of read-into to return a slice and an eof boolean separately so the compiler can optimize the slice.
* Add an each-stream-block-slice combinator that behaves like each-block but reuses a preallocated buffer for every iteration.
* Pull some strings in the stream-read-into implementation to further improve type propagation and bounds check elimination.

12 years agoio.ports: fix bootstrap
Joe Groff [Wed, 9 Nov 2011 23:38:44 +0000 (15:38 -0800)]
io.ports: fix bootstrap
Apparently can't use TYPED: or have HINTS: in the wrong place.

12 years agoio.backend.unix: hints on M\ fd refill
Joe Groff [Wed, 9 Nov 2011 23:38:03 +0000 (15:38 -0800)]
io.backend.unix: hints on M\ fd refill
Trims some low-order dispatch overhead for #376.

12 years agosequences: improve docs for map-sum.
John Benediktsson [Wed, 9 Nov 2011 22:58:37 +0000 (14:58 -0800)]
sequences: improve docs for map-sum.

12 years agoio.files: add file-reader, file-writer mixins
Joe Groff [Wed, 9 Nov 2011 22:43:39 +0000 (14:43 -0800)]
io.files: add file-reader, file-writer mixins
This lets us do some trickery so that the stream-element-type propagates from <file-reader>/<file-writer>. For #376.

12 years agoio:read: don't resize output buffer if unnecessary
Joe Groff [Wed, 9 Nov 2011 22:29:12 +0000 (14:29 -0800)]
io:read: don't resize output buffer if unnecessary
Even with a short-circuit test in the VM primitives, resize still has an overly high runtime cost in tight loops. For #376.

12 years agoio.ports: make read-step TYPED:
Joe Groff [Wed, 9 Nov 2011 22:13:41 +0000 (14:13 -0800)]
io.ports: make read-step TYPED:
Eliminates more dispatch. For #376.

12 years agoio.buffers, io.ports: trim some dispatch
Joe Groff [Wed, 9 Nov 2011 21:31:02 +0000 (13:31 -0800)]
io.buffers, io.ports: trim some dispatch
For #376. Still some work to do.

12 years agotools.profiler.sampling: preliminary docs
Joe Groff [Wed, 9 Nov 2011 19:34:31 +0000 (11:34 -0800)]
tools.profiler.sampling: preliminary docs

12 years agovm: hand SIGALRM to signal pipe if not sampling
Joe Groff [Wed, 9 Nov 2011 07:37:22 +0000 (23:37 -0800)]
vm: hand SIGALRM to signal pipe if not sampling

12 years agodeques: fix help-lint
Joe Groff [Wed, 9 Nov 2011 07:27:33 +0000 (23:27 -0800)]
deques: fix help-lint

12 years agovm: fix build on windows
Joe Groff [Wed, 9 Nov 2011 07:17:51 +0000 (23:17 -0800)]
vm: fix build on windows

12 years agounix.signals: documentation and metadata
Joe Groff [Wed, 9 Nov 2011 07:12:10 +0000 (23:12 -0800)]
unix.signals: documentation and metadata

12 years agovm: make signal pipe nonblocking
Joe Groff [Wed, 9 Nov 2011 06:07:22 +0000 (22:07 -0800)]
vm: make signal pipe nonblocking
On the off chance the Factor end of the signal handler mechanism isn't listening, we don't want to tie up the VM because the pipe's buffer got full.

12 years agotools.profiler.sampling: top-down-max-depth report
Joe Groff [Tue, 8 Nov 2011 22:35:25 +0000 (14:35 -0800)]
tools.profiler.sampling: top-down-max-depth report
Fixes #366

12 years agotools.profiler.sampling: change some names around
Joe Groff [Tue, 8 Nov 2011 22:11:53 +0000 (14:11 -0800)]
tools.profiler.sampling: change some names around
Rename the versions of top-down etc. that take sample data to "top-down*" (from "(top-down)") so they don't look so scary to use. Change "get-raw-profile-data" to "most-recent-profile-data".

12 years agovm: don't send foreign samples to outer space
Joe Groff [Tue, 8 Nov 2011 21:59:12 +0000 (13:59 -0800)]
vm: don't send foreign samples to outer space

12 years agodeques: adding ?peek-front and ?peek-back.
John Benediktsson [Tue, 8 Nov 2011 21:20:56 +0000 (13:20 -0800)]
deques: adding ?peek-front and ?peek-back.

12 years agoio.backend: start signal thread after io-backend
Joe Groff [Tue, 8 Nov 2011 19:54:54 +0000 (11:54 -0800)]
io.backend: start signal thread after io-backend
The c-io-backend can't handle input-ports and raises an error during bootstrap.

12 years agovm: don't pass this to safepoint constructor
Joe Groff [Tue, 8 Nov 2011 19:06:00 +0000 (11:06 -0800)]
vm: don't pass this to safepoint constructor
It's bad juju and MSVC calls us out on it.

12 years agounix.process: take raise from unix.ffi
Joe Groff [Tue, 8 Nov 2011 18:42:51 +0000 (10:42 -0800)]
unix.process: take raise from unix.ffi
And fix introduced ambiguity between unix.ffi:kill and unix.process:kill

12 years agonew module unix.signals: app-level signal handlers
Joe Groff [Tue, 8 Nov 2011 18:15:44 +0000 (10:15 -0800)]
new module unix.signals: app-level signal handlers
example:

IN: scratchpad USING: unix.ffi unix.signals ;
Loading resource:basis/unix/signals/signals.factor
IN: scratchpad [ V{ "Me not that kind of orc!" "Me busy, leave me alone!" "Work work" "Zug zug" } pop print flush ] SIGINFO add-signal-handler
IN: scratchpad load: 0.60  cmd: factor 41541 running 0.28u 0.16s
Zug zug
load: 0.71  cmd: factor 41541 running 0.28u 0.16s
Work work
load: 0.71  cmd: factor 41541 running 0.28u 0.16s
Me busy, leave me alone!
load: 0.73  cmd: factor 41541 running 0.28u 0.16s
Me not that kind of orc!

12 years agovm: more commentary on lock_console() stuff
Joe Groff [Tue, 8 Nov 2011 16:52:52 +0000 (08:52 -0800)]
vm: more commentary on lock_console() stuff

12 years agoio.backend.unix: startup hook for signal-pipe
Joe Groff [Tue, 8 Nov 2011 07:07:00 +0000 (23:07 -0800)]
io.backend.unix: startup hook for signal-pipe

12 years agovm: don't mask SIGTTIN from stdin_loop
Joe Groff [Tue, 8 Nov 2011 06:19:18 +0000 (22:19 -0800)]
vm: don't mask SIGTTIN from stdin_loop
Fixes #360

12 years agoio.backend.unix: init-signal-pipe function
Joe Groff [Tue, 8 Nov 2011 05:14:41 +0000 (21:14 -0800)]
io.backend.unix: init-signal-pipe function
Currently just reads off signal numbers and drops them.

12 years agovm: interrupt stdin_loop when entering fep
Joe Groff [Tue, 8 Nov 2011 05:12:11 +0000 (21:12 -0800)]
vm: interrupt stdin_loop when entering fep
The stdin_loop thread will keep trying to consume input unless we stop it by sending it a signal. Use SIGUSR2 to stop the read syscall and a mutex to hold up the loop while the fep is active.

12 years agovm: write signals to pipe immediately
Joe Groff [Tue, 8 Nov 2011 02:53:07 +0000 (18:53 -0800)]
vm: write signals to pipe immediately
The whole point is to wake up the mx without waiting for the next safepoint derp

12 years agovm: move bignum.cpp includes to master.hpp
Joe Groff [Tue, 8 Nov 2011 01:44:21 +0000 (17:44 -0800)]
vm: move bignum.cpp includes to master.hpp
That way they get PCHed

12 years agovm: enqueue signals for FEP signals too
Joe Groff [Mon, 7 Nov 2011 22:05:31 +0000 (14:05 -0800)]
vm: enqueue signals for FEP signals too
Although SIGINT still can't interrupt the current Factor thread in a sane way, this will at least wake up the run loop when waiting for input and fix #348 when implemented at the application level.

12 years agoGNUmakefile: use PCH so compiler goes faster
Joe Groff [Mon, 7 Nov 2011 21:50:52 +0000 (13:50 -0800)]
GNUmakefile: use PCH so compiler goes faster
Also add more accurate dependencies on headers so "make" works right when headers change

12 years agovm: write async signals to pipe for multiplexer
Joe Groff [Mon, 7 Nov 2011 20:51:49 +0000 (12:51 -0800)]
vm: write async signals to pipe for multiplexer
also factor out safepoint logic into its own file

12 years agoio.launcher.windows: update tests for prompt
Joe Groff [Tue, 8 Nov 2011 08:09:25 +0000 (00:09 -0800)]
io.launcher.windows: update tests for prompt

12 years agovm: switch atomic::add to atomic::fetch_add
Joe Groff [Tue, 8 Nov 2011 01:58:53 +0000 (17:58 -0800)]
vm: switch atomic::add to atomic::fetch_add
Win32 only provides fetch-and-add, not add-and-fetch, and we don't really care. Rename the functions so the order of operations is clear.

12 years agolists: make lmap behave row-polymorphically
Joe Groff [Mon, 7 Nov 2011 18:37:13 +0000 (10:37 -0800)]
lists: make lmap behave row-polymorphically
Fix contributed by @bremac. Fixes #355.

12 years agocompiler.cfg: Change low-level IR constructors from ##foo to <##foo>
Doug Coleman [Mon, 7 Nov 2011 07:00:03 +0000 (23:00 -0800)]
compiler.cfg: Change low-level IR constructors from ##foo to <##foo>