Parrot 2.0.0 "Inevitable" released! | parrot.org | Priorities: deprecated core PMCs and VTABLE entries | Roadmap: icanhaz.com/parrotroadmap | Latest modified TT's: icanhaz.com/parrotbugs
Set by moderator on 26 January 2010.
00:04 abqar joined 00:10 Limbic_Region joined 00:16 plobsing joined
Whiteknight is a bill from 1928 even a federal reserve note? 00:35
chromatic It's a Treasury note anyhow. 00:44
00:52 cconstantine joined 01:07 eternaleye joined 01:13 lichtkind joined
lichtkind ping 01:13
i just want to show a text about parrot if there is any thing uncorrect 01:15
cotto_work link?
purl or "Link is ... like ... this pointy eared goblin that walks around in midi-music land with a letter opener attacking circles and things and wooing princesses but not bannon, you know?" or preaction is Error.
cotto_work or nopaste?
purl nopaste is at nopaste.snit.ch/ (ask TonyC for new channels) or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/ or paste.scsys.co.uk (for #catalyst, #dbix-class, #moose and others) or gist.github.com/ or paste or gtfo or tools/dev/nopaste.pl or trac.parrot.org/parrot/browser/tru...nopaste.pl
lichtkind solarion: november-wiki.org/view/Parrot 01:16
01:16 cconstantine joined
lichtkind i mean cotto_work 01:16
cotto_work boggles for a moment and goes to check the page 01:17
lichtkind cotto_work: thanks
cotto_work: why boggle? 01:18
plobsing lichtkind: I've never heard of the "Parrto Compiler Tools" before 01:19
cotto_work wondering how solarion and cotto_work got confused. At any rate, do I bug masak to get a login?
lichtkind shiit
cotto_work: i misttyped instead co so and pressed tab so computer found that name
plobsing: thanks fixed now 01:20
any other mistakes? 01:22
cotto_work how do I get a login there? 01:26
lichtkind cotto_work: ask masak or viklund 01:27
cotto_work neither of whom are on atm. ok
thnaks
or thanks
lichtkind cotto_work: there is a channel on \t irc://irc.freenode.org/november-wiki
but the conent is currently written by me even if the initial texts where by masak himself 01:28
good night 01:30
cotto_work night 01:31
01:34 kid51 joined
dukeleto theory: make installcheck fails for me on a brand new database: gist.github.com/289369 01:39
theory dukeleto: make installcheck PGUSER=postgres PGDATABASE=test 01:40
dukeleto: Be sure to read the README. :-)
You can do the same thing for make test 01:41
dukeleto theory: i am looking at the readme
theory good man
dukeleto theory: that command does not appear 01:42
theory * `$PGDATABASE`
* `$PGHOST`
* `$PGPORT`
* `$PGUSER`
PGHOST and PGPORT are useless though, come to think of it.
dukeleto theory: i can read. YOUR COMMAND NEVER EXACTLY APPEARS
theory dukeleto: Well, no, because I don't know what username your going to use or what database you might have created. 01:43
dukeleto i created a db called 'leto' so I wouldn't have to set PGDATABASE
theory It could be added as an example, though.
yeah, cool
dukeleto it was confusing for me, so I am sure it will be to others
sudo make installcheck PGUSER=postgres PGDATABASE=test still fails for the same reason 01:44
using postmaster on Unix socket, default port)
============== dropping database "pl_regression" ==============
psql: FATAL: role "postgres" does not exist
command failed: "/opt/local/lib/postgresql83/bin/psql" -X -c "DROP DATABASE IF EXISTS \\"pl_regression\\"" "postgres"
theory you don't need the `sudo`
oh, you have no postres user
so
dukeleto pg_regress: could not open file "./regression.out" for writing: Permission denied
so, yes, I *do* need the sudo
theory make installcheck PGUSER=leto
well, if you don't own the current directory 01:45
I noticed your first error said "no user named root", so you didn't run it as leto.
dukeleto (kadath)(~/git/plparrot master )$ ls -lad .
drwxr-xr-x 30 leto staff 1020 Jan 28 17:37 .
i am the user leto
theory hrm
dukeleto so I am going to have to differ with you
chromatic You *are* the brute squad.
theory I don't know why pg_regress would bitch, then.
but you also don't have a "postgres" user 01:46
dukeleto all i did was install postgres 8.3.8 from darwin ports. it didn't do that
sudo make installcheck PGUSER=leto runs but fails 01:47
and failing is what it should do
with the current code
so we need more instructions about the names of databases and usernames
theory dukeleto: Sure. Go ahead and put them in however you'd like. 01:48
But clearly darwin parts has a non-standard install, as "postgres" is nearly always the name of the default super user for PostgreSQL
no idea what darwinports decided to call it.
dukeleto probably ipad or something else useless 01:49
i am headed to the OSB town hall soon
what are you up to tonight?
i will do my best to drop by the hackathon afterwards
theory date night with my woman
dukeleto sounds good. much more fun ;) 01:50
theory see ya later
kid51 Is darwinports still active? I thought it had been superseded by macports. 01:51
01:52 LaVolta joined 01:53 japhb joined
Whiteknight bacek_at_work: ping 02:36
purl msg bacek I think I know the problem in gc_encapsulate. List_chunk* is a GCable object, but in the branch you're allocating it with Parrot_gc_allocate_fixed_size_storage. List_chunk needs to be traced and collected like a PMC or STRING 02:39
purl Message for bacek stored.
chromatic Oh, List_chunk. How you are not my friend. 02:41
Whiteknight Where the hell is it even used for? 02:42
"used at"?
my grammar-foo is weak tonight
chromatic I run into it every now and then. 02:43
Whiteknight It's not used for any of the array PMCs, so what other purpose does it serve?
..and why couldn't it be replaced by ResizablePMCArray? 02:44
chromatic Could very well be. 02:45
Whiteknight not a task for this branch, but definitely something I would like to look into 02:50
chromatic I'm happy to see us get rid of old code that we barely use. 02:52
Whiteknight okay, it looks like this code is only used in the Array PMC 02:55
and that PMC is deprecated
...and I think it can disappear nowish. 02:56
Whiteknight does a small happy-dance without getting up off the couch
maybe I'll start a branch for the purpose tomorrow 02:57
...though that doesn't help the gc_encapsulate branch quite yet 02:58
Whiteknight goes to bed now. Goodnight 03:00
plobsing reads backscroll 03:35
ah! Array PMC is useful! 03:37
dalek parrot: 39f589f | dukeleto++ | README.plparrot:
Update README for 'make test'
 03:38
plobsing msg Whiteknight Array PMC beats RPA hands down running 'array_unshift_test.pir' on TT #1399 for values less than ~20000 (before Array starts segfaulting) 03:40
purl Message for whiteknight stored.
plobsing msg Whiteknight a suitable replacement for Array should have a similar order of performance 03:46
purl Message for whiteknight stored.
plobsing msg Whiteknight Array behaves ~O(n) on that, RPA is more like O(n**2) (in my crude benchmarks) 03:47
purl Message for whiteknight stored.
dalek parrot: 0a73291 | dukeleto++ | .gitignore:
Add regression files to .gitignore
 03:49
chromatic plobsing, let's fix RPA instead. 04:23
plobsing chromatic: we can do that by using struct List under the hood! ;-) 04:26
actually, I'd settle for not having to copy n elements for every shift/unshift 04:29
chromatic Right. 04:31
I remember a patch to Perl 5 several years ago that fixed similar behavior.
perlmonks.org/index.pl?node_id=17890 04:32
perlmonks.org/index.pl?node_id=42912 04:33
dalek parrot: f960f20 | dukeleto++ | (2 files):
Fix 'make test' and add a _PG_init
 05:54
05:55 jsut joined, eternaleye joined
dalek parrot: 0f2cc95 | dukeleto++ | src/handler/plparrot.c:
Be friendly to pre-8.2 pg's, in case it matters
 06:00
07:37 cotto joined 07:57 fperrad joined 08:09 iblechbot joined 08:29 particle1 joined 09:11 jan joined 10:36 leto joined 11:09 jan joined 11:17 jan joined 11:29 pdcawley_ joined 11:38 LaVolta joined 12:00 bepi joined 12:01 bepi left 12:10 cconstantine joined 12:36 payload joined 12:48 cotto joined, bluescreen joined 12:53 payload joined 13:15 iblechbot joined 13:20 whiteknight joined 13:22 cotto joined
whiteknight good morning #parrot 13:24
13:24 cognominal joined
whiteknight purl msg plobsing: that's a motivation to improve performance of ResizablePMCArray, not to keep Array. If RPA added "start of buffer" and "start of data" pointers, we could get similarly good performance. Would only need a way eventually to recycle the unused allocated storage at the beginning of the buffer after a while 13:26
purl Message for plobsing stored.
whiteknight purl msg plobsing: Array is also mostly unused in most production code, so even if it has better performance in some benchmarks it is unused. Part of this could be a naming problem. If it were called "SparsePMCArray", maybe that might garner more attention. But, it would still need major cleanup. 13:34
purl Message for plobsing stored.
13:57 ruoso joined
dalek rrot: r43635 | whiteknight++ | branches/kill_array_pmc:
creating a branch to remove deprecated Array PMC
 13:59
14:19 mikehh joined 14:22 payload joined 14:29 mikehh joined, lichtkind joined 14:35 plobsing joined 14:41 davidfetter joined
lichtkind november-wiki.org/edit/Parrot any complaines? 14:48
15:01 hudnix joined
dalek rrot: r43636 | whiteknight++ | branches/kill_array_pmc (10 files):
remove array.pmc, list.c and list.h. Cleanup all resulting code problems (not many). Miniparrot compiles but fails with message 'ResizablePMCArray: Cant shift from an empty array!
 15:04
15:08 bubaflub joined
tewk are there docs for creating a closure? 15:19
maybe docs/compiler_faq.pod:140:Then 15:21
dalek rrot: r43637 | whiteknight++ | branches/kill_array_pmc/src/pmc_freeze.c:
fix last problem in pmc_freeze.c. Build completes without errors
15:37 Psyche^ joined 15:40 cognominal joined 15:50 mikehh joined
dalek rrot: r43638 | whiteknight++ | branches/kill_array_pmc/t/pmc (2 files):
fix failure on t/pmc/exception-old.t that caused test to hang. While I was at it, converted the test from Perl/PASM to PIR and moved to exception.t
 15:55
cotto_work whiteknight++ 16:04
whiteknight this deprecation is straight-forward. Really a testament to how unused Array was
I'm having to dig into some very obscure, old PASM tests to find references to it. 16:05
cotto_work looks like trac.parrot.org/parrot/ticket/1399 can be closed when you merge that branch 16:06
whiteknight and #1270 16:08
dalek rrot: r43639 | whiteknight++ | branches/kill_array_pmc (3 files):
delete t/pmc/array.t, because the pmc is gone. Update a test in resizablepmcarray.t to use FixedStringArray instead of Array. The test was for the splice opcode between multiple array types, so the updated test should satisfy the letter of the test. More tests might be needed if we want a comprehensive set of tests for splicing together all types of array PMCs
 16:11
rrot: r43640 | whiteknight++ | branches/kill_array_pmc (2 files):
the decodetime op now returns an HLL-overridable FixedIntegerArray PMC instead of an Array. Update tests in t/op/time.t to reflect that (with TODO note about how we should test the HLL-override portion
rrot: r43641 | whiteknight++ | branches/kill_array_pmc/t (4 files):
update a few more tests. All as easy as converting 'Array' to 'resizablePMCArray'. A more thoughtful reading could maybe pick types slightly better suited for each case
 16:28
16:53 theory joined 16:54 fperrad joined
dalek rrot: r43642 | fperrad++ | trunk/config/gen/makefiles/root.in:
[build] install_config.o is needed when using pbc_to_exe --install
 17:01
17:18 theory joined 17:33 cghene joined 17:34 plobsing joined 17:49 jan joined 18:55 chromatic joined
dalek tracwiki: v14 | cotto++ | CottoTasklist 19:14
tracwiki: trac.parrot.org/parrot/wiki/CottoTa...ction=diff
cotto_work What code is responsible for creating a sub constant? 19:19
probably related to the Sub PMC's init 19:21
cotto_work breaks out the yak razor 19:28
darbelo cotto_work: Storing of constant subs (in the bytecode) is broken IIRC. plobsing had ideas about that. 19:29
it was half-related to the freeze-thaw mess. 19:30
There was a ticket about that that I can't find right now.
cotto_work I see that there's an extra possibly-unused string const after each sub that contains its name. 19:31
hexeditinh the bytecode to change that constant doesn't cause any test failures.
s/inh/ing/
19:36 iblechbot joined, ash_ joined
darbelo TT #1359 isn't the one I'm thinking about. But is related. 19:38
cotto_work and the rabbit hole gets deeper 19:44
19:45 plobsing joined
cotto_work speak of the plobsing 19:46
plobsing eh? 19:47
cotto_work darbelo just pointed at one of your tickets 19:48
darbelo TT #1359, cotto_work was shaving a related yak.
cotto_work which one? we have many 19:49
plobsing yeah, that's on my TODO list, unfortunately the list keeps getting longer 19:50
darbelo cotto_work: Subs in bytecode. It's mentioned in the ticket.
cotto_work dropping that would make pbc_merge substantially simpler
plobsing dropping how? 19:51
cotto_work Am I misreading or is the proposal to have a sub's opcodes live with it in pbc? 19:54
*part of the proposal
plobsing cotto_work: very much so. I want the freezer/thawer to manage mapping bytecode storage for the subs
right now Eval PMC stores its bytecode as a string in the image. AFAIC, thats a fail. 19:55
chromatic Yuck.
darbelo Indeed.
chromatic I wonder if that hides otherwise difficult packfile initialization bugs though. 19:56
:init and :load are right tricky to implement effectively.
plobsing I'm not very knowledgable about bytecode, just freeze/thaw, but how we do it now seems wrong (and leads to bugs - see TT #1142) 19:58
whiteknight I'm getting lots of failures in the new kill_array_pmc branch related to freeze/thaw, and It's a mess picking through it all 20:03
I would really like to see a comprehensive refactor of that system
darbelo whiteknight: Ehm, we did one.
whiteknight darbelo: think bigger, more comprehensive 20:04
there are still some fundamental problems with the design
darbelo There's some more work that was blocking on passing a deprecation boundary.
cotto_work So it's clobbering time? 20:07
darbelo Yep.
pmc_freeze_with_pmcs is the first step forward. 20:08
cotto_work What's the plan? 20:09
purl well, the plan is to use fastcgi
cotto_work go play in traffic
purl wanders off to dent some cars.
cotto_work What's the overall plan?
purl the overall plan is to abstract POE::Kernel's back-end modules (Loop::*, Resource::*, etc.) into a native C library. Thin XS shims will expose the libraries with Perl interfaces that match the existing Perl modules.
cotto_work kicks purl with a botsnack
darbelo Get rid of image_io, replacing it with a PMC. 20:10
plobsing plan or 'shiny ideal I'll probably never get to'?
cotto_work yeah. That one.
purl yeah. that one is probably horrible
cotto_work kicks purl
purl kicks cotto_work
cotto_work You should start a wiki page.
plobsing will do 20:11
darbelo plobsing: Has any work gone into pmc_freeze_with_pmcs? 20:14
I0m thinking we should erase and rebranch to take advantage of one_make and the other landed branches. 20:15
cotto_work looks like not 20:16
trac.parrot.org/parrot/log/branches..._with_pmcs
plobsing darbelo: not really
cotto_work you can sync with trunk
darbelo I don't trust svn that much. 20:17
cotto_work I don't like it, but I haven't had a bad time syncing.
darbelo I'll give it a go 20:19
dalek tracwiki: v150 | plobsing++ | WikiStart 20:20
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
plobsing blast the branch away 20:21
chromatic Or find someone with git-fu to rebase for you and avoid that throwaway. 20:22
darbelo Damm are svn merges slow. 20:31
Not that this machine is screamingly fast either, but still. 20:32
cotto_work yeah
plobsing I've put my ideas up at: "sigs" => ResizablePMCArray (size:0) [ 20:34
]
oops
ideas are at: trac.parrot.org/parrot/wiki/FreezeThawTasklist 20:35
note that I'm currently more interested in NCI, so I may not act on them for a while. 20:36
cotto_work That's fine.
dalek tracwiki: v1 | plobsing++ | FreezeThawTasklist
tracwiki: trac.parrot.org/parrot/wiki/FreezeT...ction=diff
cotto_work Thanks.
darbelo PMCs that use ATTRs other that INTVAL, FLOATVAL and PMC * can't be subclassed from PIR right? 20:39
Oh and STRING *. 20:40
whiteknight darbelo: basically, yes 20:42
we should come up with a way to make them extendable by autoboxing other types in an UnmanagedStruct PMC
or a CPointer PMC,
20:43 Andy joined, mikehh joined
darbelo no, Q < 30 20:55
Eh. Wrong window.
20:55 brooksbp joined 20:56 cotto_work joined 21:02 cotto_w0rk joined
darbelo Honestly, svn. You are the most compelling argument I've seen to use other VCSs. 21:03
bubaflub whiteknight: i saw on your blog your post about Strings and NFG 21:04
whiteknight: i'd love to be able to help out
whiteknight bubaflub: do you happen to be a student?
bubaflub whiteknight: yep. dukeleto was my mentor for last years' GSOC
darbelo thinks that NFG would make an excelent GSoC project
whiteknight bubaflub: awesome! Any help in the area of strings is welcome 21:05
and if you can get Google to pay you to do it, all the better
bubaflub whiteknight: cool. i'd love to get a leg up now. besides reading PDD28, is there anything else i should start with? 21:06
whiteknight bubaflub: I'll see if I can dig up some resources. Ther has been plenty of discussion about this in the past 21:07
you might want to chat with Allison too about it
bubaflub sounds good.
purl sounds good. is probably there a good way for me to find out when branches are merged, other than read every svn commit?
bubaflub purl, forget sounds good
purl bubaflub: I forgot sounds good
darbelo Coke: ping 21:08
dalek tracwiki: v4 | plobsing++ | NCITasklist 21:09
tracwiki: trac.parrot.org/parrot/wiki/NCITask...ction=diff
tracwiki: v151 | plobsing++ | WikiStart
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
chromatic seen lathos 21:10
purl lathos was last seen on purl 340 days, 5 hours, 50 minutes and 30 seconds ago, saying: <private message> [Feb 23 15:16:16 2009]
whiteknight bubaflub: I'm signing out for a while. We'll chat later
bubaflub whiteknight: cool. talk to you later.
dukeleto nfg? 21:11
purl nfg is No Fscking Good! or grapheme normalization form (see docs/pdds/draft/pdd28_character_sets.pod)
bubaflub dukeleto: the later one 21:14
purl i heard the later one was sleeker
21:17 joeri joined
dalek rrot: r43643 | darbelo++ | failed to fetch changeset:
Sync branch with trunk.
 21:21
21:23 mikehh joined
cotto_work If TT #1359 got implemented, would pbc have no separate bytecode segment? 21:31
If that's the idea, it certainly has some charm. 21:34
(i.e. all the bytecode would be stored in the subs' data) 21:35
darbelo cotto_work: pretty much, yeah. 21:36
The :init :load :eatkittens magic would still be there though. 21:37
cotto_work yeah
darbelo So it's not *completely* straightforward, but it's simpler than what we do now. 21:38
plobsing hmmm 21:40
darbelo I'm not sure how it interacts with the nebulous some-day-in-the-future "mmap() the bytecode" thing, however.
plobsing I was thinking about it the other way. provide a way to index into the bytecode segment from the freeze/thaw code
I can see advantages to each 21:41
darbelo plobsing: Do you have a better name than ImageIO for the image_io replacement PMC? 21:42
dalek tracwiki: v123 | fperrad++ | Languages
tracwiki: add link <span class="icon">�</span>www.factorcode.org/</a>
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
plobsing lol no. AFAIK the format it writes doesn't really have a name ATM 21:43
FPMC = Frozen PMC. but if we allow multiple types of serializers, that becomes ambiguous
maybe something alluding to its use in PBC? 21:44
darbelo freezer.pmc ?
well freezer_thawer.pmc
plobsing pbc_binary_freezer? maybe? 21:46
I don't like reserving freezer because I would like to have a text serialization at some point
darbelo I'll start building the bikeshed. We can paint it later ;) 21:47
plobsing blue no green!
darbelo goes with imageio.pmc for now. 21:48
cotto_work BinaryBytecodeFreezer? 21:51
plobsing +1
purl 1
cotto_work +purl 21:52
21:54 brooksbp joined
darbelo wonders if the BinaryBytecodeFreezer PMC should have a freeze() VTABLE 21:57
21:59 eternaleye joined
dalek tracwiki: v2 | plobsing++ | FreezeThawTasklist 21:59
tracwiki: trac.parrot.org/parrot/wiki/FreezeT...ction=diff
tracwiki: v3 | plobsing++ | FreezeThawTasklist
tracwiki: trac.parrot.org/parrot/wiki/FreezeT...ction=diff
22:03 cotto_work joined 22:07 rurban joined 22:08 rurban left 22:20 Whiteknight joined
darbelo Hmm. At the cost of some efficiency we *could* make ImageIO subclassable from PIR. 22:27
Whiteknight to what benefit?
darbelo Custom serialization. 22:28
Whiteknight how would the user even get access to the ImageIO PMC in the first place/
?
darbelo I'll add a variant to the freeze opcode. 22:29
It gives back a string now, which I intend to keep, but it could also return the PMC. 22:30
chromatic I'd rather consider that a possibility for future expansion than jump into it now.
darbelo Oh, sure. Just noting the fact. First I want to get this working, and then do a few passes of cleanup before considering that. 22:31
cotto_work +1 to chromatic
Whiteknight that's a good idea. I guess I never understood why freeze returned a string
darbelo: where you doing this, a branch? 22:32
darbelo Whiteknight: Because the PMC never got implemented?
Whiteknight: Yep. pmc_freeze_with_pmcs
I've just created a ImageIO PMC and I'm moving stuff from pmc_freeze.c into it. 22:34
cotto_work darbelo, commit early and often 22:36
22:37 Austin joined
cotto_work hi Austin 22:37
Austin hello, cotto_work 22:38
Whiteknight darbelo++
Austin Frustratingly, my enormous TV has arrived, but I must wait for my wall-mounting bracket. :(
On the plus side, I think I'm getting skin cancer from playing Wii Sports Resort. 22:39
Whiteknight maybe a PMC should serialize to normal bytecode
hello Austin
Austin Hello, Whiteknight.
darbelo I've half a mind to make the freeze format a packfile.
But that means extending the packfile format a bit. 22:40
Austin That won't work, since the pf format changes every couple of weeks.
darbelo It shouldn't explode too badly if we stick to the manipulation API. 22:41
'sedes we're already using packfile primitives for this stuff.
Whiteknight darbelo: couldn't we just stick a frozen PMC into the constants table of the PBC?
cotto_work There'd be some nice isomorphism there.
22:42 ruoso joined
Whiteknight yeah, and some nice festizio 22:42
see, I can make up words too
cotto_work Your wartnak is showing. 22:43
chromatic You send him to school, you get him an engineering degree, you make Google pay him money, then his fork()ed child process writes all over his memory. 22:47
darbelo I don't have an engineering degree yet ;) 22:49
chromatic I meant Whiteknight.
darbelo But I can make the whole fork()ed child process thing happen all the same ;)
It's not a prereq.
Whiteknight my fork()ed child process is preventing me from going to sleep() because he keeps raising SIGCRY 22:50
chromatic ENOWHINING
Whiteknight not whining, just making my day open source 22:51
NotFound Austin_away: you need a TV to sustain your wall? 22:52
Austin_away NotFound: Yeah. The TV's bigger. 22:53
chromatic It's a load-bearing TV.
NotFound Is an interesting idea. Instead of paying home repairs, buy big TV.
Austin_away Sure. A large enough TV will cover all kinds of home-decorating problems. 22:54
NotFound And use the wall mounting to reforce the walls. 22:56
22:56 plobsing joined
NotFound thinks about photo retouching, TimToady and "Wall mounted TV"... 22:57
22:58 mikehh joined 23:02 mikehh joined 23:03 patspam joined 23:04 brrant joined 23:47 mikehh joined

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!