Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2015-04-23

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:31 vendethiel joined #moarvm
03:02 vendethiel joined #moarvm
04:32 vendethiel joined #moarvm
05:01 vendethiel joined #moarvm
05:09 dalek MoarVM: f12710c | paultcochrane++ | .travis.yml:
05:09 dalek MoarVM: Change Travis matrix option to --no-jit
05:09 dalek MoarVM:
05:09 dalek MoarVM: This is because JIT is on by default, thus only the JIT was being tested in
05:09 dalek MoarVM: all configurations.  It was also throwing the warning that --enable-jit was
05:09 dalek MoarVM: obsolete and mentioning the --no-jit option, which would make sense so as to
05:09 dalek MoarVM: test with and without JIT.
05:09 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/f12710c137
05:09 dalek MoarVM: b1fe8c1 | (Jimmy Zhuo)++ | .travis.yml:
05:09 dalek MoarVM: Merge pull request #205 from paultcochrane/pr/enable_jit_to_no_jit
05:09 dalek MoarVM:
05:09 dalek MoarVM: Change Travis matrix option to --no-jit
05:10 dalek MoarVM: 0afe1e0 | paultcochrane++ | src/6model/reprs/MVM (3 files):
05:10 dalek MoarVM: Initialise unset fields in MVMStorageSpec
05:10 dalek MoarVM:
05:10 dalek MoarVM: These fields were uninitialised and thus flagged as a potential problem by
05:10 dalek MoarVM: Coverity Scan.  The nqp tests pass and this seems correct in this situation,
05:10 dalek MoarVM: however a code review would be a good idea to be on the safe side.
05:10 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/0afe1e047c
05:10 dalek MoarVM: 42e66bc | (Jimmy Zhuo)++ | src/6model/reprs/MVM (3 files):
05:10 dalek MoarVM: Merge pull request #204 from paultcochrane/pr/set_spec_fields
05:10 dalek MoarVM:
05:10 dalek MoarVM: Initialise unset fields in MVMStorageSpec
05:10 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/42e66bc5cb
05:12 dalek MoarVM: 56dcbaa | (Francois Perrad)++ | / (2 files):
05:12 dalek MoarVM: enforce C99 compliance
05:12 dalek MoarVM:
05:12 dalek MoarVM: needs -D_GNU_SOURCE or -D_XOPEN_SOURCE=600
05:12 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/56dcbaa08c
05:13 dalek joined #moarvm
05:22 colomon joined #moarvm
05:26 vendethiel joined #moarvm
06:25 vendethiel joined #moarvm
06:28 jnthn On 0afe1e0, any code reading those fields when the box type is reference is wrong anyway; they're meaningless whatever values they have in that case.
06:29 jnthn So, wasn't a bug, but don't mind the patch.
06:29 jnthn 56dcbaa is curious in that it's C89 we're targetting.
07:10 zakharyas joined #moarvm
07:15 dalek MoarVM: 94199b9 | jnthn++ | src/core/bytecode.c:
07:15 dalek MoarVM: Toss commented out code.
07:15 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/94199b9464
07:23 dalek MoarVM: 052c210 | jnthn++ | docs/ChangeLog:
07:23 dalek MoarVM: Add 2015.04 ChangeLog entries.
07:23 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/052c210289
07:23 jnthn Busy month!
07:24 jnthn 157 commits
07:24 nwc10 I failed to make 2 (more) of mine work. (So far)
07:25 jnthn There's always a next month
07:25 nwc10 right now it's feeling like unless that's a month made of Sundays, they're not going to work.
07:27 FROGGS joined #moarvm
07:27 dalek MoarVM: 4ee2748 | jnthn++ | VERSION:
07:27 dalek MoarVM: Bump VERSION.
07:27 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/4ee2748fe6
07:27 dalek MoarVM: 3607c62 | jnthn++ | README.markdown:
07:27 dalek MoarVM: Tweak a README item.
07:27 dalek MoarVM:
07:27 dalek MoarVM: To better indicate what sort of things we can do, now we can do 'em.
07:27 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/3607c624dc
07:28 vendethiel joined #moarvm
07:28 jnthn I'll give folks a chance to review the README etc.
07:29 jnthn Anything from that aside, 3607c62 is what will be released in a couple of hours.
07:29 FROGGS jnthn++
07:30 jnthn masak: Moar release should be in place in time :)
07:31 FROGGS jnthn: the section "Building a Rakudo with MoarVM" mentions parrot
07:31 jnthn Ah
07:32 FROGGS s/parrot/jvm/?
07:32 FROGGS "Precise, generational, parallel, GC" -- is the last comma correct there?
07:33 dalek MoarVM: cdcbf43 | jnthn++ | README.markdown:
07:33 dalek MoarVM: Give Rakudo Configure.pl example that works.
07:33 dalek MoarVM:
07:33 dalek MoarVM: FROGGS++
07:33 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/cdcbf4353a
07:34 jnthn FROGGS: It's probably an over-reaction to http://cdn2.thegloss.com/wp-content/uploads/2011/09/jfk.jpg :)
07:35 FROGGS lol
07:38 dalek MoarVM: 6c7541e | jnthn++ | README.markdown:
07:38 dalek MoarVM: Grammar tweak; FROGGS++.
07:38 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/6c7541e192
07:42 nwc10 jnthn: "Implement a more compact varint serialization format" isn't quite correct
07:43 nwc10 I think it's usually *less* compact.
07:43 nwc10 (slightly)
07:43 nwc10 but it uses less CPU
07:43 nwc10 it will be more compat for values between 64 and 126
07:45 jnthn nwc10: Better wording? "more efficient" :)
07:47 nwc10 Yes, that seems bnetter
07:47 * jnthn cvhanges it
07:48 FROGGS you guys are snilly
07:49 dalek MoarVM: d56c8a6 | jnthn++ | docs/ChangeLog:
07:49 dalek MoarVM: Tweak varint ChangeLog entry; nwc10++.
07:49 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/d56c8a6924
07:55 * lizmat wonders how interesting http://www.liblfds.org would be for MoarVM ?
07:56 lizmat probably not, as it is already 2.5 years old  :-(
07:56 lizmat sorry for the noise
07:57 jnthn lizmat: Well, code doesn't rot with age :)
07:58 jnthn lizmat: It looks familiar, I'm sure I've looked over it before
07:58 jnthn lizmat: But it's quite restricted
07:58 jnthn e.g.
07:58 jnthn "hash (add-only)"
07:58 FROGGS wasn't there an intend to to something (replace) uthash?
07:59 jnthn At some point, yes.
07:59 FROGGS why ooc? is it inefficient?
08:00 jnthn It could be more memory efficient, but more notably it's missing sufficient randomization.
08:01 FROGGS ahh jnthn, before I forget... can you look at my question from yesterday? http://irclog.perlgeek.de/moarvm/2015-04-22#i_10484760
08:01 FROGGS ahh
08:01 FROGGS I always wonder why we want to have randomization at all
08:01 jnthn Check for matching SC - yes
08:01 jnthn Randomization - thwart DoS attacks
08:02 FROGGS where are these checks?
08:02 jnthn See write_obj_ref
08:02 jnthn Well, actually it's not "check for matching SC"
08:02 jnthn It's "check for no SC and put it into the current one if so"
08:02 FROGGS ahh, nice
08:02 FROGGS that is what I was looking for
08:03 FROGGS I was thinking the check happens earlier
08:03 jnthn I think "packed" is the thing what nwc10++ added recently so we can store things more compact. :)
08:03 jnthn No, happens during traversal
08:03 FROGGS I see
08:04 FROGGS veeeery nice... so I can continue now :o)
08:04 jnthn \o.
08:04 jnthn \o/ even
08:04 rurban joined #moarvm
08:15 nwc10 Given that MoarVM isn't using shared hash keys, I suspect that randomisation could be added by storing a small (eg) 32 bit salt value for each (and every hash)
08:16 nwc10 with the salt coming from a private drand48() or similar
08:16 nwc10 ie, not much effort needed
08:17 nwc10 because, as best I can tell/remember, the actual integer hash values are not exposed outside of the hash table code
08:33 masak jnthn: thank you :)
08:34 vendethiel joined #moarvm
08:48 Ven joined #moarvm
09:04 Ven joined #moarvm
09:44 jnthn http://www.moarvm.org/releases/MoarVM-2015.04.tar.gz
09:44 jnthn masak: ^^
09:45 jnthn Will do the website updates later.
09:58 vendethiel joined #moarvm
10:04 raiph joined #moarvm
11:10 vendethiel joined #moarvm
11:21 masak \o/
11:21 masak jnthn++
11:22 masak "Native reference related optimizations (goal: 2015.04)"
11:23 masak "Optimize away unrequired native reference takes and dereferences, including over inlining boundaries."
11:23 masak I'm out of touch. how much was this goal achieved?
11:24 FROGGS still todo
11:42 vendethiel joined #moarvm
11:52 * FROGGS thinks about adding MVM_nuke_idx_of_sc()
12:11 dalek joined #moarvm
12:19 dalek joined #moarvm
12:23 dalek joined #moarvm
12:35 BinGOs joined #moarvm
12:48 vendethiel joined #moarvm
12:49 Ven joined #moarvm
13:14 vendethiel joined #moarvm
13:43 vendethiel joined #moarvm
14:09 vendethiel joined #moarvm
14:14 jnthn Yeah, missed the native ref opts. The parameterics serialization got done, and NFG is basically there.
14:15 jnthn Not wired up everywhere, but the key algos are in place. Plus we got all /NFK?[D|C]/ too and an update to Unicode 7 :)
14:18 Ven joined #moarvm
14:49 FROGGS joined #moarvm
14:59 vendethiel joined #moarvm
15:13 TimToady joined #moarvm
15:33 vendethiel joined #moarvm
15:50 brrt joined #moarvm
15:50 brrt \o
15:51 jnthn o/ brrt
15:52 jnthn I see you're now up to 7 comments on the JIT proposal :)
15:53 brrt yes :-) and all positive
15:53 brrt i'd better start reading intel docs then :-)
15:55 brrt i think what you said about using the jit as a stepping-stone as it were for small-bigints, nativecall, etc.. that has a lot of potential
15:55 jnthn *nod*
15:56 jnthn Well, they also can produce expression trees.
15:57 vendethiel joined #moarvm
16:02 brrt right. expression trees should be much more ... malleable (i think)
16:15 brrt i'm off for an errand
16:27 vendethiel joined #moarvm
16:46 brrt joined #moarvm
16:59 vendethiel joined #moarvm
18:08 vendethiel joined #moarvm
18:53 FROGGS jnthn: do you want to answer? https://github.com/MoarVM/MoarVM/pull/202#issuecomment-95679004
19:11 brrt joined #moarvm
19:24 zakharyas joined #moarvm
21:42 Peter_R joined #moarvm
21:50 pyrimidine joined #moarvm
22:15 vendethiel joined #moarvm
22:28 Peter_R joined #moarvm
22:44 LLamaRider joined #moarvm
23:15 vendethiel joined #moarvm
23:41 vendethiel joined #moarvm
23:54 danaj joined #moarvm

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary