Camelia, the Perl 6 bug

IRC log for #parrot, 2009-05-03

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:06 Whiteknight cotto: ping?
00:09 AndyA joined #parrot
00:13 cotto Whiteknight, pong?
00:14 Whiteknight cotto: Is darbelo blogging about his project anywhere?
00:14 cotto No, but that'd be a good idea.
00:29 eternaleye joined #parrot
00:33 Whiteknight okay, I just want to make sure I post lots of links to him, if he has one
00:34 cotto I'll recommend it next time he's around.
00:36 MariachiElf Is he working on something the lay person would understand?
00:42 cotto an arbitrary precision PMC using decNumber
00:42 MariachiElf that's awesome
00:43 MariachiElf I was just thinking the other day that I'd expect they had one of those given the definition of int and float
00:43 dukeleto joined #parrot
00:44 kid51 joined #parrot
00:44 tetragon_ joined #parrot
01:14 Whiteknight MariachiElf: We actually have an implementation now called "BigInt", but darbelo is working on a better one
01:14 Whiteknight at least, we all hope it will be better
01:14 MariachiElf Better is better. :)
01:17 cotto It'll be decimal-based instead of binary.
01:17 MariachiElf I'm not sure the distinction....
01:18 cotto bcd?
01:18 purl i think bcd is basically "the hex digit this represents is actually the number" or Binary Coded Decimal or (e.g., 0x1234 == 1234) or Buoyancy Control Device - that blow-up vest thingy scuba divers weigh so you can adjust to neutral buoyancy underwater.
01:19 cotto binary-coded decimal is the one you want
01:19 MariachiElf So 4 bits per digit?
01:19 cotto a little bit less, but basically yes
01:19 * MariachiElf races out to patent the use of a blow up vest thingy to implement math in a computer
01:20 MariachiElf Sounds like a "string" with a 4-bit character set
01:20 MariachiElf interesting
01:20 cotto decnumber?
01:21 cotto decnumber is http://speleotrove.com/decimal/
01:21 cotto decnumber?
01:21 purl decnumber is http://speleotrove.com/decimal/
01:33 cognominal joined #parrot
01:41 bacek hi again
01:41 purl oh, you're back!
01:42 bacek why don't use libgmp for bignums?
01:42 cotto decNumber is bcd-like, so rounding is less confusing.
01:44 cotto we do have libgmp-based pmcs, though
01:55 bacek ah, ok
01:56 * bacek realised that he trying to do way too many stuff in Parrot.
01:57 bacek And all this stuff depends on bytecode generating from Parrot...
01:57 bacek *sigh*
02:00 cotto afk
02:06 Coke Infinoid: help yourself.
02:13 Theory joined #parrot
02:18 Ademan joined #parrot
02:35 janus joined #parrot
03:24 dalek joined #parrot
03:24 Infinoid Coke: Thanks.  Next commit you make to partcl should say coke++, not wcoleda++.
03:25 Infinoid See https://trac.parrot.org/parrot/changeset/38449 for the rest of the magic behind that.
03:26 jsut joined #parrot
03:29 cotto shiny
03:44 ilia joined #parrot
04:25 dduncan left #parrot
04:50 darbelo joined #parrot
05:09 dukeleto joined #parrot
05:25 darbelo cotto, ping
05:33 cotto darbelo, pong
05:36 darbelo I'm having build issues with the dynpmc. Dou you have a minute to confirm them?
05:36 cotto sure
05:37 cotto are your changes in svn?
05:38 darbelo Not yet. But svn has the problem too. It began when I added the ATTRs.
05:39 cotto oh boy! story time.
05:39 cotto continue
05:41 darbelo I didn't notice them before, becouse I was doing make install to the wrong place, took me a while to notice that.
05:41 cotto what's the problem?
05:42 purl i heard the problem was not hardware, perros's first implementation was nginx+memcached as front end server
05:42 cotto no, the problem is <reply>
05:42 purl okay, cotto.
05:42 darbelo Anyway. Right now, pmc2c isn't generating code for the METHODS or VTABLES.
05:43 darbelo running the pir on eaxmples/ gives "Method 'version' not found for invocant of class 'DecNum'"
05:43 cotto It's somewhat picky about formatting.  Can you commit or nopaste your pmc code?
05:43 cotto It being pmc2c
05:43 cotto .
05:45 dalek parrot: r38450 | cotto++ | branches/tt528_vtinit/lib/P​arrot/Pmc2c/PMCEmitter.pm:
05:45 dalek parrot: [pmc2c] add code to generate update_vtable for variant vtables, plus get_vtable
05:45 dalek parrot: Parrot builds, but with some test failures.
05:45 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38450/
05:47 Theory joined #parrot
05:50 darbelo Failing code is in as r11.
05:52 cotto darbelo, the problem is that you have some #includes right before the ATTRs.  If you put them with the others at the top of the file, you can get the next error. ;)
05:55 cotto lmk if you get stuck fixing any other errors.  I'll be around for at least an hour.
05:55 darbelo ok. I can't belive I never thought of moving the #includes arround.
05:55 cotto also, it'd be good to start blogging about your progress.
05:56 cotto once a week would be great
05:57 darbelo I don't have a blog. Any preferences as to where I sould start one?
05:58 cotto Either parrot.org or parrotblog.org would make sense.  I'm not sure if we're planning on hosting blogs on parrot.org, but istr something about merging it and parrotblog.
05:59 cotto I'll ask at #ps.
06:04 Infinoid parrotblog and parrotcode content is to be merged into parrot.org, and then we'll set up DNS forwards and retire those sites
06:05 Infinoid That was the plan according to allison on 2009-02-17, and I haven't heard any changes of plan since
06:13 darbelo cotto, is there any way to make src/pmc/pmc_decnum.h include decQuad.h ?
06:13 cotto hmmm
06:14 cotto let me play with that for a minute
06:14 darbelo no problem.
06:17 cotto darbelo, I don't think so.  If you're trying to include a decNumber struct, you can wrap it in a struct similar to what src/pmc/bigint.pmc does.
06:17 cotto although it's inelegant
06:18 cotto It might work to prefix the types in the ATTR definition with 'struct', making them forward declarations.
06:24 dalek parrot: r38451 | chromatic++ | trunk/docs/book/ch02_getting_started.pod:
06:24 dalek parrot: [book] Edited chapter 2 for style and accuracy.  There's an editorial note
06:24 dalek parrot: which needs addressing from an author-type person.
06:24 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38451/
06:27 cotto darbelo, any luck?
06:29 darbelo prefixing with struct didn't fly.
06:29 darbelo trying to wrap wit a struct now.
06:30 cotto If you can't find anything better, feel free to commit and let me play.
06:40 dalek parrot: r38452 | cotto++ | trunk/src/pmc/default.pmc:
06:40 dalek parrot: [pmc] use Parrot_str_equal to improve readability
06:40 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38452/
07:03 darbelo made ATTRS void *, lots of warnings but it works.
07:05 cotto eew
07:06 cotto I'll look at it after you commit and see if something less hacky is available.
07:07 cotto but go ahead and commit whenever you're ready
07:07 darbelo It just went in. I'll look for something less hacky too.
07:09 darbelo I have the feeling "ATTR union  decQuad *number;" should work, but it doesn't.
07:22 nopaste "cotto" at 96.26.202.243 pasted "using structs in ATTR definitions" (15 lines) at http://nopaste.snit.ch/16467
07:23 cotto msg darbelo http://nopaste.snit.ch/16467
07:23 purl Message for darbelo stored.
07:37 iblechbot joined #parrot
08:30 bacek O HI
08:32 bacek cotto: how is going with vtables refactoring?
08:37 cotto gah
08:37 cotto I'm so close
08:37 cotto I'll most likely have something working tomorrow, then something committable Monday.
08:38 cotto or Christmas
08:38 purl Christmas is all jolly goofy shit. weee! or o/~ we wish you a spendy christmas o/~ x3; o/~ and a 'spensive new year o/~ or not yet or next year or when Perl6 is released.
08:40 cotto It'll be nice when I can start ripping out code.
08:40 cotto That file is ugly.
08:42 bacek indeed.
08:43 bacek I'll put pmc_pct on hold waiting for your branch to merge. It's waste of time trying to reimplement current behaviour.
08:43 cotto hey.  I may have it.
08:43 cotto I wish I didn't have to wait for all the PMCs to be regenerated, although there's not really a way around it.
08:44 cotto bacek, you can work on ATTRs.  I think that's orthogonal to the vtable stuff.
08:44 bacek And METHODs as well.
08:45 bacek Problem is - I don't like current ATTRs. And don't understand METHODs :)
08:47 cotto nope.  didn't work.  I'm off to bed.
08:47 cotto night
08:47 bacek good night
08:49 cotto or not
08:53 bacek :)
08:53 bacek Ship it!
08:54 bacek I'll stay awake for couple more hours and can fix some stuff
08:59 cotto it builds...
08:59 bacek that's good
09:00 cotto btw, the code currently manually builds a default vtable.  It's just temporary.
09:01 cotto Once it's working I'll factor that out into an appropriate function.
09:02 bacek "manually builds default vtable"?
09:04 cotto it allocs a new vtable and fills in all the default entries, instead of copying from an existing location
09:04 cotto for some reason, interp->vtables[0] is null, so I can't use it.
09:05 cotto I tried to modify the default PMC, but its code is very... special.
09:05 cotto feel free to take a shot
09:05 bacek I'll try
09:06 cotto one failing test, but it's a pmc2c test.
09:06 cotto I win!
09:06 bacek cool!
09:07 bacek (pmc2c treats VTABLE* update_vtable as VTABLE method... What a shame)
09:08 cotto it's committed, and I'm asleep.  night
09:08 dalek parrot: r38453 | cotto++ | branches/tt528_vtinit/lib/P​arrot/Pmc2c/PMCEmitter.pm:
09:08 dalek parrot: [pmc2c] switch class_init to using runtime-built vtables - all tests pass, but there's much cleanup work to be done
09:09 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38453/
09:09 bacek cotto: night. I'll try not to break your stuff :)
09:24 HG` joined #parrot
10:07 Eevee joined #parrot
10:34 uniejo joined #parrot
10:47 cognominal joined #parrot
11:13 rdice joined #parrot
12:25 HG` joined #parrot
12:25 dalek parrot: r38454 | bacek++ | branches/tt528_vtinit/lib/Parrot/Pmc2c (2 files):
12:25 dalek parrot: Small cleanups:
12:25 dalek parrot: - Factor out get_vtable_func and update_vtable_func for generating
12:25 dalek parrot:   get_vtable and update_vtable functions.
12:25 dalek parrot: - Override default_get_vtable.
12:25 dalek parrot: - Fix dynpmc's generating.
12:25 dalek parrot: - Small other fixes that I forgot.
12:25 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38454/
12:28 bacek hmm... Probably they are not so small.
12:45 AndyA joined #parrot
12:49 pfig joined #parrot
13:38 kid51 joined #parrot
13:39 Ron joined #parrot
13:41 Limbic_Region joined #parrot
14:09 ilia joined #parrot
14:18 tetragon joined #parrot
14:19 amoc joined #parrot
14:23 uniejo joined #parrot
14:35 uniejo joined #parrot
14:56 LylePerl joined #parrot
15:07 davidfetter joined #parrot
15:13 Andy joined #parrot
15:35 ilia joined #parrot
16:01 bsdz joined #parrot
16:06 bsdz joined #parrot
16:07 uniejo joined #parrot
16:56 dalek parrot: r38455 | petdance++ | trunk/config/gen/makefiles/root.in:
16:56 dalek parrot: getting back to work on my splint target:
16:56 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38455/
17:06 HG` joined #parrot
18:08 Theory joined #parrot
18:23 dalek rakudo: rcddb162 | ((Moritz Lenz))++ | t/spectest.data:
18:23 dalek rakudo: t/spectest.data: passing integration test, #icu marker for uc.t
18:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​ddb162a685464e4abe99255ee6bad92a59af68f
18:24 dalek rakudo: rf68de36 | ((Moritz Lenz))++ | docs/ChangeLog:
18:24 dalek rakudo: [docs] typo (vim-o?) in ChangeLog
18:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/f​68de3629d211692ad9e98a6fe388b8939654702
18:24 dalek rakudo: re486ca5 | ((Moritz Lenz))++ | tools/ (2 files):
18:24 dalek rakudo: [tools] fix reading of spectest.data
18:24 dalek rakudo: Don't read the icu markers as part of the file name
18:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/e​486ca5eae08c20147f33570e1f228fc3cc59245
18:24 dalek rakudo: rd5a8940 | ((Moritz Lenz))++ | t/spectest.data:
18:24 dalek rakudo: track moved test files (S12-role -> S14-roles)
18:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​5a8940dbc68959685310f2392f166cbdf88be9b
18:31 Theory joined #parrot
18:52 Casan joined #parrot
19:13 ilia joined #parrot
19:28 Woody4286 joined #parrot
19:30 AndyA joined #parrot
19:35 kid51 joined #parrot
19:55 AndyA joined #parrot
19:58 darbelo joined #parrot
20:08 AndyA joined #parrot
20:31 dalek xml: redf93a5 | ((Francois Perrad))++ |  (5 files):
20:31 dalek xml: have a nice OO interface
20:31 dalek xml: (global variables are hidden)
20:31 dalek xml: review: http://github.com/fperrad/xml/commit/ed​f93a570d6c9d8fc698fbbae71e219f9d0fa67d
20:48 bsdz joined #parrot
21:37 bsdz joined #parrot
21:49 tetragon joined #parrot
21:58 darbelo left #parrot
22:13 darbelo joined #parrot
22:36 Whiteknight joined #parrot
23:03 ilia joined #parrot
23:13 PacoLinux joined #parrot
23:16 cotto apparently 'patch -p0 >x' is different from 'patch -p0 <x'
23:32 TonyC joined #parrot
23:41 Whiteknight cotto: Yeah, I hate patch because I always get the syntax wrong
23:56 darbelo cotto, tried your changes. Works well, give or take a warning.

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

Parrot | source cross referenced