re2 library (which came out of code search) could
return the longest fixed string (if any) as part of its API, but I can't find
that now. In https://swtch.com/~rsc/regexp/regexp4.html it mentions building
trigrams (directly) from parsing the regex (with ANDs and ORs) which is what
some $ork code
bsanitizer/asan/asan_malloc_linux.cc:93 __interceptor_realloc
==8139==ABORTING
Nicholas Clark
systems return 31 bit values.
ANSI C only mandates 15 bit as a minimum.
If so, this third-party code is buggy.
But if that's true, why on Earth has no-one else hit this before?
Nicholas Clark
test binding to the default address
(which IIRC is all interfaces), and hence can the warning be fixed by
explicitly binding only to localhost?
Nicholas Clark
core/interp.c:3964
#4 0x7fb46ccf5bf4 in MVM_vm_run_file src/moar.c:304
#5 0x401a4f in main src/main.c:191
#6 0x7fb46c20bd1c in __libc_start_main (/lib64/libc.so.6+0x1ed1c)
SUMMARY: AddressSanitizer: double-free
../../.././libsanitizer/asan/asan_malloc_linux.cc:62 __interceptor_free
==9674==ABORTING
Nicholas Clark
ter this month's
release (to give a whole month to iron out any instabilities or bugs).
So it might be a little bit bad this month, but next month will be awesome.
Allegedly :-)
Nicholas Clark
nnot invoke this object (REPR: Null; VMNull)
in block at 129120.pl line 9
Line 9 is `print await @results;`
What do I need to do better, to get actual useful details?
Nicholas Clark
<$> const(IV 1) s
a <2> eq sK/2
b <|> and(other->c) vK/1
c <0> pushmark s
d <@> return K
e <;> nextstate(main 48 -e:1) v:%,{,2048
f <|> once(other->g)[$h:48,49] vK/1
g <$> const(IV 42) s
h <0> padsv[$h:48,49] sRM*/LVINTRO,STATE
i <2> sassign KS/2
goto j
o <0> padsv[$h:48,49] sRM*/STATE
j <;> nextstate(main 49 -e:1) v:%,{,2048
k <0> pushmark s
l <0> padsv[$h:48,49] l
m <@> say sK
n <1> leavesub[1 ref] K/REFC,1
-e syntax OK
Nicholas Clark
ain (/lib64/libc.so.6+0x1ed1c)
SUMMARY: AddressSanitizer: double-free
../../.././libsanitizer/asan/asan_malloc_linux.cc:93 __interceptor_realloc
==25229==ABORTING
Nicholas Clark
On Wed, Aug 03, 2016 at 03:15:49PM -0400, Brandon Allbery wrote:
> On Wed, Aug 3, 2016 at 3:12 PM, Nicholas Clark > wrote:
>
> > I don't know what file would be safer. Maybe mem?
>
>
> None of them. There's no guarantee that /proc exists (non-SVR4 commercial
&
On Wed, Aug 03, 2016 at 03:15:49PM -0400, Brandon Allbery wrote:
> On Wed, Aug 3, 2016 at 3:12 PM, Nicholas Clark > wrote:
>
> > I don't know what file would be safer. Maybe mem?
>
>
> None of them. There's no guarantee that /proc exists (non-SVR4 commercial
&
t part is just Garbage In, Garbage Out, and a distraction from the
true cause)
Nicholas Clark
/moar-san/bin/moar+0x401058)
AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: SEGV src/core/interp.c:2859 MVM_interp_run
==20272==ABORTING
Looks like "just" a direct NULL pointer dereference, not the fallout from
earlier more subtle undefined behaviour.
Nicholas Clark
.c:245: MVM_gc_root_gen2_add: Assertion `!(c->flags &
MVM_CF_FORWARDER_VALID)' failed.
Aborted
(I wasn't paying close attention, but I guess I built with full-on debugging
flags)
I hope that makes it more obvious to someone how to fix this.
Nicholas Clark
f6
Poisoned by user:f7
Contiguous container OOB:fc
ASan internal: fe
==10839==ABORTING
[hopefully with formatting fixed]
Nicholas Clark
liner changes from ASAN barfage to no output.
(which means that the SEGV is fixed).
That means it's now "tests needed" ?
Nicholas Clark
; Segmentation fault
>
>
> Code:
> my $x := (my $y := $x);
> say $x.^methods;
>
> Result:
> Segmentation fault
ASAN shows that all of these are "just" NULL pointer dereferences.
No (other) undefined behaviour at the C level.
Nicholas Clark
guous container OOB:fc
ASan internal: fe
==16988==ABORTING
Nicholas Clark
On Tue, Nov 03, 2015 at 05:31:40PM +, Nicholas Clark wrote:
> Not a NULL pointer dereference. A more serious bug than that.
Also present with MoarVM/NQP/Rakudo at
47bb37e Include \r\n synthetic in default separators.
7710de0 Fix thinko in \r\n -> grapheme prep.
385850b Stre
f9
Global init order: f6
Poisoned by user:f7
Contiguous container OOB:fc
ASan internal: fe
==13267==ABORTING
Not a NULL pointer dereference. A more serious bug than that.
Thanks for reporting the bug, and thanks for reducing it down to a really
small case.
Nicholas Clark
On Thu, Oct 29, 2015 at 07:41:06PM +, Nicholas Clark wrote:
> On Thu, Oct 29, 2015 at 09:22:30AM -0700, Pepe Schwarz wrote:
>
> > The same code on R-M segfaults:
> >
> > 16:15 m: A: for 1 { for 1 { last A }; CONTROL { when CX::Last { say
> > "last"
n __libc_start_main (/lib64/libc.so.6+0x1ed5c)
#5 0x401058 (/home/nicholas/Sandpit/moar-san/bin/moar+0x401058)
AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: SEGV src/strings/ops.h:30 MVM_string_graphs
==28758==ABORTING
Looks like it's "just" a NULL pointer dereference.
Nicholas Clark
ere's nothing wrong happening before a NULL pointer dereference.
Nicholas Clark
1058)
AddressSanitizer can not provide additional info.
SUMMARY: AddressSanitizer: SEGV src/gc/roots.c:171
MVM_gc_root_add_temps_to_worklist
==17783==ABORTING
Nicholas Clark
CopFILE(PL_curcop),
+strerror(EISDIR));
+
return rsfp;
}
I think that Rakudo should make the same check.
(And I think that "not a directory" is better than some attempt to whitelist
"file", er wait, "file or symlink", er wait "file, symlink or named pipe" etc)
Nicholas Clark
ably it should be memory comparison on the meat, carefully avoiding
the padding.
Nicholas Clark
est::Deep, etc.
It doesn't look to me like there's a way to do that (yet) in the Perl 6
ecosystem. Am I missing a way to mix in new test functions without having
to replace/redo the existing ones?
Nicholas Clark
in.c:189)
Is this something to do with P6Int having alternative storage representations?
If so, it implies that some more tests would be useful for values around the
switchover points.
(To be robust I'd suggest testing all ranges ±2 around values 256**n for
n 0 to 9)
Nicholas Clark
On Thu, Mar 05, 2015 at 08:41:51PM +, Nicholas Clark wrote:
> On Thu, Mar 05, 2015 at 07:57:12PM +0000, Nicholas Clark wrote:
>
> > Sadly PPC32 is a mess. With these patches NQP will build, and the nativecall
> > tests pass. However 2 other tests fail (arithmetic and pri
On Thu, Mar 05, 2015 at 07:57:12PM +, Nicholas Clark wrote:
> Sadly PPC32 is a mess. With these patches NQP will build, and the nativecall
> tests pass. However 2 other tests fail (arithmetic and printf problems), and
> the Rakudo build explodes with a syntax error backtrace.
[
s it's not even worth attempting to
figure out the Rakudo problem, as it might be caused by the NQP bugs.
Nicholas Clark
>From 14f8c73cce49fc789088d660f262eadb8e0c27c2 Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Thu, 5 Mar 2015 11:13:38 -0800
Subject: [PATCH] The nativecall test for
it's possible to can dump MoarVM config with the stage0
NQP - this means that we can do other stuff in a similar way.
This means that a lot more dogfood possible a lot earlier than is obvious.
Nicholas Clark
>From 1e3fbde3562d77ba8f50d6c62d28e61c87ad6e1f Mon Sep 17 00:00:00 2001
From: Nic
ating with default action of signal 11 (SIGSEGV):
dumping core
==23928== Access not within mapped region at address 0x10
==23928==at 0x4F57A5E: MVM_interp_run (interp.c:1226)
==23928==by 0x502FBA3: MVM_vm_run_file (moar.c:208)
==23928==by 0x40110D: main (main.c:191)
This is the same C backtrace as #120397, in OP(assign)
Nicholas Clark
gn to an
immutable value");
1232 }
1233 goto NEXT;
I can't convince gdb to "like" the various local libraries on this machine
(local updated compiler toolchain, but not debugger), so I can't verify which
pointer is NULL.
Nicholas Clark
On Tue, Sep 16, 2014 at 11:10:54AM +0200, Gerd Pokorra wrote:
> Hello,
>
> Am Dienstag, den 16.09.2014, 09:32 +0100 schrieb Nicholas Clark:
> > What are the current active ports of Rakudo?
> >
> > I see a list in http://rakudo.org/how-to-get-rakudo/
> >
> &g
kudo-star/README.html
Are there any OSes (or packaging systems) that are carrying Rakudo that I've
missed?
Nicholas Clark
Agree, for the reasons that Perl 5 made this switch.
In the short term, it avoids problems with "shadowing" vs needing to overwrite
the master copy (which upsets inferior package managers)
In the long term doing "site" first also provides a way for Star to phase out
shipping superceded modules.
Nicholas Clark
on freenode
and ask for suggestions on how to get involved.
Nicholas Clark
On Tue, May 20, 2014 at 07:28:06PM +0100, Nicholas Clark wrote:
> For the Pi (which is swapping heavily) the results are quite noticeable.
Peak swap is now 515928 (Mb)
Previously it was 524996
On an run on May 10th it was 458968
My book keeping isn't perfect, so I don't know f
On Sat, May 17, 2014 at 09:36:00AM +0100, Nicholas Clark wrote:
> On Fri, May 16, 2014 at 10:05:36PM +0100, Nicholas Clark wrote:
>
> > Both the 32 and 64 bit code paths tested on x86_64 (with some editing of
> > configs). Builds on real x86. The Raspberry Pi has passed all
On Fri, May 16, 2014 at 10:05:36PM +0100, Nicholas Clark wrote:
> Both the 32 and 64 bit code paths tested on x86_64 (with some editing of
> configs). Builds on real x86. The Raspberry Pi has passed all of NQPs
> tests, which means that it should work (without this it would SEGV on th
efine MVM_DIRECT_SC_IDX_SENTINEL 0xFF
#else
# define MVM_DIRECT_SC_IDX_SENTINEL ~0
#endif
but I'm not yet in a position to benchmark it. (The 32 bit platforms are
slow, and don't have as may cores at the x86_64 beasts)
Nicholas Clark
>From 85577a3dfcf9c95c3e59b41dc9dd6ad379fe217d Mon Sep 17
d.
I'm not too worried about that currently - most folks will be using stuff
from packages. eg, here's what apt gave me:
This is perl6 version 2014.03.01 built on parrot 6.0.0 revision 0
Nicholas Clark
>From a1ff1a2d5dd49ef25ca0e962fd026e694cbb26fc Mon Sep 17 00:00:00 2001
From: Nicholas
it's worth someone checking this on Windows (32 bit and 64
bit) and if it doesn't break anything that currently works, applying it.
Nicholas Clark
>From d57ceba5b4064846980b944943cea4d2f4b54bbc Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Sun, 4 May 2014 14:45:41 +0200
Subjec
ay, if I change the type of bytes to be MVMuint8 *, then the bug is
repeatable on x86_64 with the default char (ie signed).
So attached are two patches, one to remove the above buggy code, and the
second to change the prototype of MVM_string_utf8_decode to take MVMuint8 *,
as conceptually UTF-8 sequences a
e?
Or is it giving a truthful less-than-useful answer?
(Because what matters to MoarVM is the alignment of, and therefor padding
needed for, structure members)
Nicholas Clark
On Tue, Feb 18, 2014 at 12:02:50PM -0500, Andy Dougherty wrote:
> On Tue, Feb 18, 2014 at 03:11:01PM +0000, Nicholas Clark wrote:
> > Andy, is it OK to use your Perl 5 static inline probing code in MoarVM?
>
> Absolutely. Please feel free. Perl 5's Configure script is actu
On Tue, Feb 18, 2014 at 08:52:42PM +, Nicholas Clark wrote:
> Hence the first two patches (to Rakudo and Moar) that get a pointer to the
> MVMCollectable contained within every object. Which applies on its own,
> without adding any compiler warnings.
Fail. 2 copies of the Moar patc
On Tue, Feb 18, 2014 at 03:11:01PM +, Nicholas Clark wrote:
> Attached is the proposed patch, which implements compiler based probing for
> the MoarVM Configure.pl. Tested on OS X, FreeBSD, Linux, HP/UX and AIX.
> Not tested on Win32, so might not work there without tweaks.
Attac
list is going to be more work than this.
Nicholas Clark
>From dba2746602e470cd325907b0aaedb46970ee96a6 Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Tue, 18 Feb 2014 15:13:10 +0100
Subject: [PATCH 1/6] Add probing code to Configure.pm to learn how the
compiler does 'static inline
less code, and at least to my eye it's simpler too.
Also attached is another test for NQP for serialisation.
Feedback and other opinions welcome.
Nicholas Clark
>From 27820d14304fd94f47158e1de0228779e2bfea01 Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Tue, 11 Feb 2014 20:52:59
On Tue, Feb 11, 2014 at 04:48:29PM +, Nicholas Clark wrote:
> Attached are a first stab at some tests for native integer serialisation.
> They pass on the JVM, and on MoarVM (tested 64 bit and 32 bit Linux) with
> patching. Specifically the two commits from the branch maybe-varin
Oops, missed something
On Tue, Feb 11, 2014 at 04:48:29PM +, Nicholas Clark wrote:
> >From b20f8f328c6155019702cf44d27ed10a4251ac21 Mon Sep 17 00:00:00 2001
> From: Nicholas Clark
> Date: Tue, 11 Feb 2014 17:40:54 +0100
> Subject: [PATCH] Fix various errors in varint ser
)
(gen/parrot/stage2/NQP.nqp:3646)
The problem seems to be serialising an array. Is the correct trick to make
them work to refactor them to use an object with an array attribute?
I don't think that fixing the tests should hold up fixing MoarVM, if the
C code patch is acceptable.
Nicholas C
nline functions:
commit b575e123dfc6ab49639fec28871247b8fe8bdeab
Author: Nicholas Clark
Date: Wed Jan 29 21:41:26 2014 +0100
Encapsulate SC access through MVM_sc_{get,set}_{stable,obj}_sc()
Ideally these would be inline functions.
diff --git a/src/6model/sc.c b/src/6model/sc.c
index 3
ot;torture") and the setting compilation is faster.
Confirming how faster, and whether it's just the setting, is left as an
exercise for the reader :-)
Nicholas Clark
>From ace9d7a50510a85e8c7b63a17946a9e9bf027a84 Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Tue, 28 Jan
On Mon, Jan 27, 2014 at 09:43:49PM +, Nicholas Clark wrote:
> With this, core setting just produces 1 error from valgrind:
> I'm not sure how to find that one.
Writing everything plausible to /dev/null helped a lot, as it caused valgrind
to produce an earlier backtrace:
diff --
On Mon, Jan 27, 2014 at 09:43:49PM +, Nicholas Clark wrote:
> I don't think that there are that many places where this happens (but I
> haven't rigged the build to count them yet), so I don't think that it's
> worth adding to the C structures to remember
lues come from
==11990== ERROR SUMMARY: 1 errors from 1 contexts (suppressed: 9 from 9)
I'm not sure how to find that one.
Nicholas Clark
>From 9717c39e5e74cfe145e2210342e63f183478fa14 Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Mon, 27 Jan 2014 22:20:54 +0100
Subject: [PATCH] In P6
On Thu, Jan 09, 2014 at 10:32:37PM +0100, Jonathan Worthington wrote:
> Hi,
>
> On 1/9/2014 22:00, Nicholas Clark wrote:
> > I've been running with the appended for a few weeks now, and I think that
> > it's the right fix.
> Unfortunately, releasing the m
arguments with side effects
passed to "functions" that happens if the "functions" are actually macros
3) it makes it easier to debug in toolchains that don't have something like
gcc's -g3
(No, I'm not volunteering myself. I'm already busy looking f
On Sat, Dec 21, 2013 at 08:45:49PM +, Nicholas Clark wrote:
> ==11811== Thread #1: Attempt to re-lock a non-recursive lock I already hold
> ==11811==at 0x4C2BFC1: pthread_mutex_lock (hg_intercepts.c:484)
> ==11811==by 0x4FACB40: uv_mutex_lock (thread.c:69)
> ==11811==
1, as set by this code:
OP(scwbdisable):
GET_REG(cur_op, 0).i64 = ++tc->sc_wb_disable_depth;
cur_op += 2;
goto NEXT;
I don't know how to debug this further.
Nicholas Clark
lection?
ie, should that MVM_string_find_encoding(tc, enc_name) be yanked to a line
above and assigned to a local variable, so that any GC run has happened
before the arguments to MVM_string_decode() start to be put onto the CPU
stack for the C function call?
Nicholas Clark
>From b54de7c404e5bd01
oDo_All) {
MVM_gc_collect_free_stables(tc);
}
I mail it here in case it's useful to anyone else trying to figure out missing
write barriers.
2) Do most missing write barrier bugs look as crazy to figure out as this one?
Or was this one particularly special by corrupti
table might be in the nursery and hence
might move
unless, and only unless, the code can prove that it doesn't need to?
Should MoarVM API calls that can't trigger GC be explicitly documented as
such?
Nicholas Clark
>From 1b046a7a2832171bd2e50399410396365a6634dd Mon Sep 17
On Sun, Dec 22, 2013 at 08:42:05PM +, Nicholas Clark wrote:
> On Sun, Dec 22, 2013 at 01:31:35PM +0000, Nicholas Clark wrote:
> > I don't know how to chase this further. I have a string that is from
> > fromspace
> > that is being referenced by MVM_string_equal. I t
On Sun, Dec 22, 2013 at 01:31:35PM +, Nicholas Clark wrote:
> I don't know how to chase this further. I have a string that is from fromspace
> that is being referenced by MVM_string_equal. I think that it has come from a
> register. Output (with debugging):
Following suggestions
nt frame, hence not obvious how to iterate over them
and attempt to validate them by dereferencing.
Nicholas Clark
>From 8e9a66570cf2048be49fe0690364222ad921c53d Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Sun, 22 Dec 2013 05:04:23 +0100
Subject: [PATCH] Release tc->instance->mu
recursive mutex?
*) Or should MVM_sc_create be moving the mutex lock much more tightly around
the manipulations of tc->instance->sc_weakhash?
(But can MVM_HASH_BIND trigger a GC run?
Can MVM_string_flatten?)
nor do I know if similar systematic flaws exist in other mutex handling -
can the GC attempt to lock any of the other mutexes?
Nicholas Clark
mporarily root, and encapsulates the conditional check for
exists.
Anyway, for now, I'm applying the attached patch locally, and continuing to
test.
Nicholas Clark
>From e3b867213b855506682aa6a5956ac691a227479d Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Fri, 20 Dec 2013 17:47:43 +0100
On Tue, Dec 17, 2013 at 11:54:44PM +0100, Jonathan Worthington wrote:
> On 12/15/2013 21:50, Nicholas Clark wrote:
> > I don't know for sure what the right fix is.
> Hopefully, 191c383 is the right fix. Confirmation welcome. :-)
I can confirm that with it, the error goes away an
On Mon, Dec 16, 2013 at 07:58:06PM +0100, Tobias Leich wrote:
> I don't think they can move...
>
> value_obj is only used from its assignment to before an allocation. Same
> goes for base_obj.
[reply below because I think that it makes more sense]
> Am 15.12.2013 21:50, sc
On Sun, Dec 15, 2013 at 08:50:27PM +, Nicholas Clark wrote:
> On Sun, Dec 15, 2013 at 05:56:40PM +0000, Nicholas Clark wrote:
>
> > With this, all NQP tests pass I can build the Rakudo setting, with a nursery
> > size of 2048K.
> >
> > I have now written
On Sun, Dec 15, 2013 at 05:56:40PM +, Nicholas Clark wrote:
> With this, all NQP tests pass I can build the Rakudo setting, with a nursery
> size of 2048K.
>
> I have now written a script that will attempt to repeatedly recompile that
> far, each time dropping the nursery si
On Sun, Dec 15, 2013 at 04:19:19PM +, Nicholas Clark wrote:
> (Again) I don't know what the correct fix is.
For now I'm working round it with this:
commit d3b4bca8139a5875e6a91d14896c2167ce36e23d
Author: Nicholas Clark
Date: Sun Dec 15 18:31:29 2013 +0100
XXX Uncle
On Sun, Dec 15, 2013 at 02:30:14PM +, Nicholas Clark wrote:
> I hacked up (somewhat literally) the attached 3 patches, to try to help
> automate GC debugging for MoarVM. The intent is to be able to change how
Next mole to whack:
$ gdb --args /home/nicholas/Sandpit/moar-g/bin/moar nqp.
I'm offering it here on the assumption that it will be useful to others, who
will be able to fix the problems that it discovers. (Until the build is
clean)
Nicholas Clark
>From edbb71149b3cf91cdf6c7c0a406ed39737ac7fb5 Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Sun, 15 Dec
month, and git is attributing to months based on commit date.
So any work that was rebased and then merged will be attributed to the date
of rebasing. I think that this is partly why April is relatively barren in
Perl 5 land - because branch activity in April was rebased onto v5.19.0 in
May before
On Tue, Aug 27, 2013 at 09:18:17AM -0400, Parrot Raiser wrote:
> On 8/27/13, Nicholas Clark wrote:
> >
> > 2) Frankfurt has Apfelwein. It's often available in large jugs, from which
> >you refill a small glass, traditionally with a diamond pattern on it
> >
lly with a diamond pattern on it
Generally it's more "Old Rosie" than "Aspells". So it can be a bit fun
drinking it in large quantities.
Nicholas Clark
* In 2008 we had massive fun in Mainz (the far end of the line from the
airport), trying to find a machine tha
at idea on this, but "out of memory" on which step?
Compiling the setting?
and what platform - x86_64 (GNU) Linux?
How much RAM?
If you use -Xmx to set a larger heap size, does it work?
Nicholas Clark
using the branches rak-jvm-support and jvm-support that
will be the problem.
Also, there's an obsolete branch jvm-support in NQP. You might have that
locally, just to confuse things. If you run `git remote prune origin` that
should clean up. (Or git fetch --prune)
Nicholas Clark
On Sat, May 18, 2013 at 05:17:58PM +0200, Gerd Pokorra wrote:
> Hello,
>
> how did you fix the build of NQP for the JVM?
>
> I have the same problem that you reported in an other email.
As I said in the e-mail:
> Am Freitag, den 17.05.2013, 08:28 +0100 schrieb Nicholas C
, with -Xmx you can configure more memory for you JVM. To be
sure that you don't leak or waste memory. Take a heap dump and use
the Eclipse Memory Analyzer to analyze your memory consumption.
Does that sound familiar enough to someone to be tempting?
Nicholas Clark
/parrot/ops/nqp.ops | 37 +
2 files changed, 39 insertions(+), 0 deletions(-)
I don't know if that's useful, but I hope that it is.
Nicholas Clark
ns of Perl 5 checked out from git. We've also got smoke testers
running on various platforms people care about, which spots portability
bugs and regressions well before release (headache) time.)
This sounds like the sort of thing that new volunteers could usefully
contribute to the Rakudo effort, as it doesn't have a high barrier to entry,
and there's plenty of Perl 5 toolchain to steal.
Nicholas Clark
9"
OpenJDK Runtime Environment (rhel-2.3.9.1.el6_4-x86_64)
OpenJDK 64-Bit Server VM (build 23.7-b01, mixed mode)
$ javac -version
javac 1.7.0_19
$ uname -srvmpio
Linux 2.6.32-358.el6.x86_64 #1 SMP Fri Feb 22 00:31:26 UTC 2013 x86_64 x86_64
x86_64 GNU/Linux
Nicholas Clark
haven't
My unease comes, I think, from a Perl 5 perspective. undef/defined on an
array is dealing with something associated with the container, and an
implementation detail at that. undef/defined on a scalar is dealing with
something associated with the value.
If, as I think you are saying, Perl 6 has a much clearer idea of what is
value and what is container, and defined has a clearer idea of what it is
interacting with, then it's not going to get confusing for Perl 6.
Nicholas Clark
On Thu, Mar 21, 2013 at 04:50:40PM +, Nicholas Clark wrote:
Bother. That should have been more to perl6-language than perl6-compiler.
Should I resend it?
Nicholas Clark
ty data
(deprecation warnings turned off, script appended)
ie "undefined" in scalars is a property of the value, and always passed to
the assignee in assignment.
whereas for arrays, not so. The shape of that table is definitely an
implementation detail exposed.
Nicholas Clark
#!/usr/
On Mon, Feb 11, 2013 at 08:11:53AM -0600, Patrick R. Michaud wrote:
> On Mon, Feb 11, 2013 at 09:56:31AM +0000, Nicholas Clark wrote:
> > Here are a better set of patches for Rakudo. They don't duplicate the check
> > for $p < 0, and they avoid calling nqp::elems(), by
atches remain fine.
On Sun, Feb 10, 2013 at 09:19:22PM +0000, Nicholas Clark wrote:
> I've no idea if the coding style is appropriate. Whether the indentation is
> crazy. Whether one should be using Mu. etc
This still holds. Style-wise, it would be possible to avoid using ret
appropriate. Whether the indentation is
crazy. Whether one should be using Mu. etc
But it does work :-)
[Including the spectests. But why should you trust me on that? :-)]
Nicholas Clark
>From 2c73f49b51bf43c046e42780fefbb0d35720 Mon Sep 17 00:00:00 2001
From: Nicholas Clark
Date: Sun, 10
On Sun, Feb 03, 2013 at 10:35:12AM +, Nicholas Clark wrote:
> On Sat, Feb 02, 2013 at 06:32:10PM -0800, Matthew Wilson wrote:
> > Did you mean to use $z in the say output of the nqp and perl versions of
> > the microbenchmark, or did you mean to run it twice?
>
> I
29;
my $t0 = time;
my $z = fib($N);
my $t1 = time;
print "fib($N) = $z\n";
print "time= " . ($t1-$t0) . "\n";
$ ~/Sandpit/5162/bin/perl5.16.2 ~/test/fib.pl
fib(29) = 514229
time= 0.5503830909729
Nicholas Clark
e= 0.544455051422119
So the NQP JVM prototype is about 1.5 faster at running that code than Perl
5. That's without any sort of tuning of its code generator.
That's quite impressive.
Well done Jonathan. Please don't stop :-)
Nicholas Clark
On Wed, Jan 09, 2013 at 10:07:21AM +, Nicholas Clark wrote:
> I think that the right way to go is to write sprintf, or as much of it as is
> possible, in NQP. Specifically, I suspect that all the parsing of the % codes
> should be done in high level language, only kicking down to
) possible to optimise in ways that just can't be done via a monolithic
call. For an example, see this PyPy blog entry:
http://morepypy.blogspot.com/2011/08/pypy-is-faster-than-c-again-string.html
Nicholas Clark
On Thu, Nov 15, 2012 at 04:13:01PM +, Nicholas Clark wrote:
> Following a chat on IRC today:
>
> 09:28 < nwc10> I see accesses to all structure members except constraints in
>the rest of the code
> 09:29 < jnthn> Yeah. It could be moved into Rakudo
1 - 100 of 1342 matches
Mail list logo
