Parrot 1.0 Released | parrot.org | 380 RTs left!
Set by moderator on 28 March 2009.
00:13 Casan joined 00:15 TiMBuS joined 00:16 RobertLJ joined 00:22 tetragon joined 00:23 rg joined 00:26 RobertLJ joined
RobertLJ register RobertLJ 08**augo bobjohnson11@yahoo.com 00:28
00:31 AndyA joined
RobertLLJ Hey, I think I need to change my psswd now :) 00:33
cotto good idea 00:41
purl cotto: Good Idea: Finding Easter eggs on Easter morning. Bad Idea: Finding Easter eggs on Christmas morning.
00:46 darbelo left 00:52 RobertLJ joined
dalek rrot: r37823 | allison++ | trunk (4 files):
[build] Add a --disable-rpath option to allow Linux packagers to build

Resolves TT #476.		 01:13
rrot: r37824 | pmichaud++ | trunk (10 files):
Merge pge1 branch back into trunk (partial resolution of TT #460).

  * Retrieve text of a Match is now .Str (was .text)
  * Setting a result object in a Match is now match.'!make'(obj) (was .result_object)
  * Retrieving the result object is now match.'ast'() (was .item)
  * NQP's $(...) now calls '.ast' instead of '.item'
  * get_string/integer/number on Match objects now work from .Str instead of result object		 01:27
01:30 kid51 joined
wayland76 Do those recent commits mean it's a good time to ask about trac.parrot.org/parrot/ticket/442 again? (and ticket 503 too maybe)? 01:53
rg no allison isn't here. you should try tomorrow after #parrotsketch
wayland76 rg: ok, thanks. I'll try to remember to do that. 01:54
02:01 cognominal joined
dalek kudo: c015556 | pmichaud++ | (6 files):
Bring Rakudo up-to-date with recent PGE and compiler tools

accurately reflect recent changes in S05).		 02:08
shorten dalek's url is at xrl.us/bem7ff
02:48 janus joined
dalek rrot: r37825 | jkeenan++ | trunk/tools/dev/nopaste.pl:
Applying patch submitted by fperrad in trac.parrot.org/parrot/ticket/520: use pod2usage() where possible.
 02:52
02:57 mikehh joined
bacek_ kid51: don't forget TT#522 :) 03:07
03:08 hudnix joined
kid51 bacek_ : I was only correcting a spelling error in the description. I don't claim to understand the issues involved, i.e., why it might have been deleted. 03:16
03:17 mncharity joined 03:18 mncharity left
kid51 I'll cc Allison on the TT (... but she hasn't been responding to cc-s on my tickets, so I can't guarantee a response). 03:18
kid51 must sleep
purl $kid51->sleep(8 * 3600);
03:37 jq joined 04:23 Andy joined 04:40 ilia joined 04:56 tetragon joined 05:06 tuxdna joined 05:16 dukeleto joined 06:04 TiMBuS joined 06:11 uniejo joined 06:24 tuxdna joined 06:42 flh joined 06:45 clunker3 joined 06:47 zostay joined
Infinoid bacek_: I will test #518 and #522 on mingw tomorrow morning. I suspect #518 will need some minor help, as PF_LOCAL isn't defined on win32 07:32
if I can just use #ifdef and use -1 or 0 if the define wasn't found, I should be able to fix that kind of problem easily if I see it 07:33
I suspect we might run into similar issues in socket_unix on older unices which don't have PF_INET6, which can be fixed in the same way 07:35
but overall I like the design
bacek_ Infinoid: -1 should be fine 07:37
Infinoid: Whole idea of this mapping was to handle such situations 07:38
eternaleye Hi, does anyone know what the current state of extending Parrot with C++ is? 07:57
08:15 masak joined
mikehh oops - testj fails - have been rubbung on Amd64 where make fulltest passes - I just switched to i386 and have a failure 08:17
srubbu/runni/
t/op/trans.t - Tests: 22 Failed: 1 - Failed test: 13 08:22
make fulltest - testj - Kubuntu Intrepid i386 at r37825 08:23
smoke PASS - smolder - 19505 08:24
08:52 TiMBuS joined
bacek eternaleye: it should be possible 09:14
09:24 cognominal joined 09:34 amoc joined 11:08 tuxdna joined 11:33 Debolaz joined 12:35 rg1 joined 12:46 Andy joined
Coke msg wayland76 you're probably better off commenting on the original tickets. 12:49
purl Message for wayland76 stored.
13:03 Khisanth joined, ilia joined 13:21 jan joined 14:04 NordQ joined 14:09 ilia joined 14:17 NordQ joined 14:31 Theory joined
dalek kudo: 5374f73 | pmichaud++ | docs/spectest-progress.csv:
spectest-progress.csv update: 339 files, 8039 passing, 0 failing
 14:45
shorten dalek's url is at xrl.us/bem8wz
14:49 Tene joined 15:08 Psyche^ joined 15:11 ilia joined 15:19 ilia_ joined 15:25 jan joined 15:27 amoc joined 15:28 Khisanth joined 15:30 Khisanth joined
Coke is tempted to pull perlcritic.t out of 'codetest'. =-) 15:31
(not really, but it's sloooow)
Infinoid perlcritic.t makes codetest take longer for me than "make test" does... including the building parrot part! 15:32
bacek: there exists a runtime/parrot/include/sockets.pasm file which has the same problems your socket5.patch had, in its first version 15:41
I think either we should use that version (modifying it if necessary), or get rid of it to avoid confusion 15:42
dalek rrot: r37826 | coke++ | trunk (6 files):
[t/docs] add more function doc signatures.
Infinoid Putting the constants in a non-autogenerated file like that does have the benefit of making the resulting constants without the "PIO_" prefix, which seems a little more readable to me 15:43
but I don't know what the normal thing is, for pir constants. Do you have an opinion?
particle wonders if perl::critic has been run through devel::nytprof 15:44
generally our constants are generated from header files, so they're available to c/pir/pasm 15:46
Infinoid yeah, that's what I'm looking at. bacek++ has a patch to add socket constants (like PF_INET and IPPROTO_TCP) so they can be used from pir 15:48
One problem is that these values vary from one platform to the next... PF_INET6 is 10 on linux and 23 on mingw, for example
So in the current implementation, we have the system values which vary, and the parrot values which don't vary (thus keeping bytecode portable), and some tables to map between the two 15:49
Coke Infinoid: we would need to standardize them.
Infinoid: there you go.
Infinoid Yeah, bacek++ has done that
But the resulting names in pir space are like ".PIO_PF_INET"... the extra prefix indicates the parrot version 15:50
Is that ok?
Coke i have no problem with a parrot-specific prefix, no.
Infinoid Great, thanks.
Coke should help avoid confusion.,
should we go with something super verbose like PARROT_PIO_PF_INET ? =-) 15:51
Infinoid if you're writing in pir, it's pretty obvious that you expect to be using parrot. This year, at least.
So these constants will be available as PIO_* in PIR space... and then languages like c99 can map those back to their common names I guess 15:52
even though the values might be different, so c99 will have to be careful to stay away from the system headers
That's where I start getting confused. :) 15:53
particle don't you worry about c99 on parrot, it'll probably never happen. 16:00
and in this case, punting to the hll is totally appropriate
only when hll devs complain loud enough and long enough will we consider addressing their issues. right, pmichaud? ;) 16:01
Coke particle: I'm sorry, I have to re-architect partcl AGAIN, what were you saying?
particle come back when i care :P 16:03
Coke go document your function and hush. 16:06
16:11 masak` joined, baest_ joined, elmex joined 16:12 dalek joined
Infinoid bacek: I've modified the comment in io.h because these things are now not the same as the native linux values, and I've made the PF_* constants conditional. Now I've fixed a couple of C90 warnings, and it tests fine on linux... testing on mingw next 16:16
I suspect we might need to standardize the SOCK_* definitions the same way we're doing for PF_*. But I'll see whether this works first 16:17
if the examples work on mingw, I'll commit this and add the SOCK_* stuff in a later commit 16:18
I've also got some warnings in socket_win32.c I want to fix.
16:20 ilia joined 16:22 ilia joined
pmichaud #parrot sketch in 123 16:27
moritz I'll probably miss that 16:29
16:30 Whiteknight joined 16:31 uniejo joined
dalek rrot: r37827 | Infinoid++ | trunk (4 files):
Apply patch from bacek++ in TT #518:
 16:35
Infinoid r37829 builds and tests fine for me on linux and mingw... I would love to get some reports from osx, mingw and solaris as to whether it builds, tests, and whether the examples in examples/io/ work 16:36
uh, reports back from msvc, not mingw 16:37
dalek rrot: r37828 | Infinoid++ | trunk/examples/io/http.pir:
Apply patch from bacek++ in TT #522:
 16:39
particle rebuilds on msvc 16:42
dalek rrot: r37829 | Infinoid++ | trunk (3 files):
[io] Make socket constants conditional. (They aren't all defined on all architectures, and should fall back to -1 where they aren't implemented)

  * Also, fix the following warning: src/io/socket_unix.c:171: warning: ISO C90 forbids mixed declarations and code
  * Also, fix the following warning: src/io/socket_win32.c:117: warning: ISO C90 forbids mixed declarations and code
Infinoid reduces dalek's polling interval 16:43
particle: There are probably some warnings from src/io/socket_win32.c. I'm going to be poking around in there to fix some warnings I'm seeing on mingw... if you have the chance, a list of msvc warnings would also be helpful 16:44
particle i'll see what i can do.
rg is running smoke on solaris. it'll show up in 1-2 hours 16:50
Infinoid thanks! 16:51
16:52 dalek joined
Coke_lunchies rg++ 16:52
Coke uninstalls parrot, and tries from the official macport version, and fails. wtf. 17:03
(pointing to the build path again. wtf.) 17:04
17:05 Enso joined
Coke anyone here try to install using the version in macports? 17:05
17:06 RobertLJ joined
dalek rkdown: ae0b52b | (Francois Perrad)++ | src/parser/ (2 files):
now, works with Parrot r37824
 17:09
shorten dalek's url is at xrl.us/bem9h4
RobertLJ register 08**augo bobjohnson11@yahoo.com 17:10
moritz RobertLJ: pick a new password ;-)
Coke wtf. I removed the previously installed parrot bits, new install worked. ah well. 17:17
17:18 darbelo joined
Coke I wonder if we should change parrot_config's output to be, e.g. YML. 17:18
(or json. or something that we conceivable reuse.)
17:25 uniejo joined
dalek rtcl: r327 | wcoleda++ | trunk/ (7 files):
Apply patch from issue #73 - track string renames
 17:28
particle Infinoid: my smoke report shows all pass 17:29
smolder.plusthree.com/app/public_pr...ails/19517
shorten particle's url is at xrl.us/bem9k5
Infinoid great, thanks! any chance you can try the stuff in examples/io/ ? there's an HTTP server that listens on localhost port 1234, and a simple client that connects to www.ibm.com and spits out some http headers
Coke grumbles, yes, we DO need a parrot-devel port for osx. 17:30
Infinoid the output of httpd.pir indicates it's meant to browse the html docs, but you can go to localhost:1234/NEWS or localhost:1234/Configure.pl even if you haven't done a "make html" 17:31
17:31 barney joined
Coke is it common for the -devel- port to include everything in the original port? 17:32
(or is it normally just the extras that the non-devel wouldn't have.)
(or does anyone care.)
Infinoid I dunno how ports does it. if it's anything like how debian/redhat/etc do it, the -devel- package will depend on the base package and just add headers and linking stuff 17:33
masak` oh, the parrotsketch meeting is still another hour away, right?
moritz yes 17:34
barney got confused as well
moritz DST and all. 17:35
Coke Infinoid: if we had an install-dev that was only the extras, I'd do it that way. =-)
(but it looks like 1.0 at least has install-dev : install, which means we get everything.)
ah well.
masak that's nice. then I have almost an hour to hack on November. :)
Infinoid yeah. the binary packagers tend to install to a tmpdir and cherry-pick individual files/dirs for each package it builds
Coke I am wondering if I should just update the parrot portfile to do an install-dev. 17:36
is there any point to a non-dev version?
aside from "ooh, look, I have parrot"?
Infinoid if you have a binary package of parrot and a binary (well, pbc) package of rakudo, that should be enough to run perl6 code 17:37
particle Infinoid: actually, i need to run it from the parrot build dir to get it to see NEWS
Infinoid so in that case, I'm not sure the parrot-dev package should be necessary
particle and yes, that works.
Infinoid particle: ah. I had assumed that's where you would run it
particle Infinoid: i got a 302 from the ibm site 17:39
so, that's working, too
Infinoid cool, thanks!
Coke tries to figure out what gets installed with 'install-dev'
particle i might like to see that script updated to use one of pafo's sponsor pages
heck, or parrot.org. 17:40
Infinoid yeah, probably. It also has some obsolete comments at the top
Coke: I'm guessing headers for embedders, all the extra tools (pbc_dump for example), stuff like that? 17:41
Coke Infinoid: I'd like a list. =-)
I can't seem to make tools/dev/install_dev*.pl generate one.
Infinoid I'm not authoritative, but I think the distinction is, parrot-dev is needed for building HLLs, whereas the base parrot package is all the stuff necessary to run the resulting .pbc files
but in parrot's case, quite a lot of stuff is wrapped into the one executable
Coke aha. --dry-run=1, not --dry-run 17:42
rant. tools/dev/ is the wrong dir for this. should be tools/build
particle yeah, the pmc2c files are in install-dev
Coke why does runtime/parrot/library/PCT/PAST.pbc -> /usr/lib/library/PCT/PAST.pbc 17:43
?
shouldn't there be a parrot in the installed dir?
hurm. perhaps that'll be resolved with appropriate options i can specify.
bah. I thought I was done with this crap. 17:44
particle i thought it should be /usr/lib/parrot/1.0.0/library/PCT/PAST.pbc or something 17:46
particle knows little of the new installer 17:47
17:50 uniejo joined
Coke wonders why nqp is installed under languages. 17:50
(all compilers, not just nqp) 17:52
alright. looks like install-dev doesn't have any other executables, just a bunch of docs and PBC runtimes and source. should be trivial to add a new portfile.
particle: yes, that's what it turns out to be. it's just the defaults that look scary. 17:53
17:54 allison joined 18:00 eternaleye_ joined
Coke hurm. can't seem to make the parrot-devel port look for a 'parrot' binary (instead of a parrot-devel binary) 18:01
aha. 18:04
(baha. tried to copy a try from another -devel port, nope. 18:06
18:06 cognominal joined
Coke there we go. yay cribbing. 18:07
18:16 jhorwitz joined 18:19 Tene joined, chromatic joined
chromatic I may be late for #ps; on the phone. 18:19
Coke bah. creating the devel port slightly more annoying that just hijacking parrot. 18:28
Coke opens a ticket in case someone is inspired. 18:30
Infinoid Coke++ for working on it - if we can't figure out, what chance to third party packagers have
Coke Infinoid: your timing is awful, I just gave up. =-) 18:31
Infinoid haha, but you tried!
cotto chromatic++ #working on jit issues 18:36
chromatic jit--
18:40 confound joined 18:42 Andy joined 18:43 eternaleye joined
dalek kudo: 78cb4c3 | (Moritz Lenz)++ | t/spectest.data:
add passing test to t/spectest.data
 18:48
shorten dalek's url is at xrl.us/bem94u
Infinoid purl, cla? 18:54
purl i heard cla was Contributor License Agreement or www.perlfoundation.org/contributor_..._agreement or www.parrot.org/foundation/legal
dalek rrot: r37830 | allison++ | branches/pcc5:
Removing calling conventions refactor branch from the repository
 18:57
particle we need an announcement of some sort that the nlnet grant has been completed. 19:08
barney: are you planning a pipp developer release? 19:10
moritz it has? 19:11
barney not really. There is too little working. 19:14
rg infinoid: solaris tests ok. how long is http.pir supposed to run (on a slow machine)?
barney 71/4xxxx tests succeeding, of the PHP test suite
moritz hey, it's a start 19:15
barney 71/4780
purl 0.0148535564853556
moritz barney: rakudo's record started with 223 passing tests, but I'm sure the real start was much lower 19:16
Infinoid rg: It takes a few seconds here. It tries to connect to www.ibm.com port 80 and dumps the http response; if your solaris machine doesn't have direct internet connectivity, it will probably time out
19:16 ron joined
rg it does. lwp-request took maybe 2 seconds. 19:17
19:17 ron left
particle yikes. barney, now that i'm coding in php, i may have some interest in working on pipp 19:17
barney particle++ 19:18
rg http.pir hangs after saying "connect returned 0"
Infinoid Ok. Are you up for a little interactive debugging?
rg sure
barney There are a lot of low hanging fruits, e.g 'switch', 'include'
Infinoid rg: great. Please try: ./parrot -t1 examples/io/http.pir 19:19
rg hangs on 64 callmethodcc P1, "recv" P1=Socket=PMC(0x10026f4f0)
nopaste "Infinoid" at 75.140.33.106 pasted "The output of "./parrot -t1 examples/io/http.pir" on linux/x86-64" (54 lines) at nopaste.snit.ch/16019
Infinoid rg: Hmm. First guess, maybe it's not sending the second newline so the server is still waiting for the end of the http request 19:20
rg ok, let me tcpdump that
Infinoid Ok. I've got a sniffed connection here to compare against 19:22
19:23 PhatEddy joined
Infinoid Another thought, maybe solaris recv() is trying to fill the buffer before returning 19:23
PhatEddy diakopter: ping 19:24
particle barney: is there a ml/trac/other where i can read more on what's needed?
rg uhm ... why would it use udp?! 19:25
that looks like a hardcoded constant difference
Coke tries to debug a problem in JS on IE. wtf. 19:26
barney particle: I might go for a Trac-queue, until then there is wiki.github.com/bschmalhofer/pipp/i...ion-status
shorten barney's url is at xrl.us/benab7
rg oh yes, looks a lot like it: 19:28
particle thanks, barney
rg #define SOCK_STREAM 2 /* stream socket */
#define SOCK_DGRAM 1 /* datagram socket */
barney particle: a Test.php would be great, in order to get rid of Perl 5 test scripts 19:29
rg works with sock.'socket'(2, 2, 0)
i believe there's already a ticket saying we need proper constants? ;)
Infinoid rg++ # good find
particle do we have a porting guide?
that's a good faq item for one
Infinoid rg: I don't know if there's a ticket, but I was thinking about this being a possible issue earlier today 19:30
[09:17] <@Infinoid> I suspect we might need to standardize the SOCK_* definitions the same way we're doing for PF_*. But I'll see whether this works first
rg ok then there's your proof that you need them ;)
19:31 eternaleye_ joined
Infinoid rg: When I've worked up a patch, I'll bug you about this again if you don't mind :) 19:32
chromatic wonders if the 64-bit problems in the JIT have something to do with hardcoded stack offsets which seem to assume an 4-byte pointer size.
rg sure, no problem
Infinoid this will have to wait until I get home from work tho 19:33
barney particle: Nope, over Easter I hope to find some time for looking into extensions. Getting an idea wheter the PHP extensions can be reused
Infinoid I can't find a ticket. I'll create one
bacek good morning
rg infinoid: no worries. 19:34
bacek Infinoid: thanks for applying and fixing my patches
Infinoid bacek: thanks for sending them :)
It sounds like we'll need another lookup table for the SOCK_* defines; they differ on solaris according to rg++'s testing. I was just writing up a ticket, it'll be a few hours before I can work on that
19:36 kj joined
bacek Infinoid: I can implement it. 19:36
Infinoid TT #527 19:37
bacek++
barney added particle as Pipp collaborator 19:38
bacek Infinoid: not yet :)
particle barney++
bacek Infinoid: udp/tcp selected by "PIO_PROTO_*", not SOCK_*... 19:40
rg: can you check /etc/protocols for "tcp" and "udp" values? 19:42
darbelo I have intermittent failures in t/pmc/packfiledirectory.t on OpenBSD 19:45
diakopter PhatEddy: what 19:47
darbelo half the time it crashes with "parrot in free(): error: bogus pointer (double free?)"
rg bacek: the values in /etc/protocols are standards that can't change anywhere
tcp 6, udp 17, icmp 1 to name the most common ;) 19:48
darbelo a backtrace shows that it happens in Parrot_exit().
bacek rg: I know... That's why I wonders how it can create udp socket instead of tcp...
rg it's stream vs. dgram sockets 19:49
barney Were there changes in PAST::Val ?
bacek rg: ah, ok. It makes sense.
Infinoid: you have to change description in ticket :)
PhatEddy rakudo: sub foo($v, $w?, $x?, $y?){ say $v~"|"~$w~"|"~$x~"|"~$y}; 19:51
tewk chromatic: did you see my GSoC proposal. Just curious, but What JIT work are you doing? getting x86_64 to work?
chromatic tewk, working on cotto's patch to make *ManagedStruct use ATTRs. 19:52
tewk ahh
chromatic TT #519
You might have some insight on it. 19:53
tewk Looking
darbelo is this a known issue or should I open a ticket? 19:55
chromatic darbelo, I haven't seen it.
Can you create a ticket and include a backtrace?
particle how's our coverity report looking post-1.0? 19:56
chromatic Haven't looked in a while.
tewk, look at t/pmc/nci_66.pir 19:59
in nci_vp (a function from src/nci_test.c), inOpaque is 0xa instead of a pointer.
20:00 ilia joined
Infinoid rg: Can you verify the values of IPPROTO_TCP == 6 and IPPROTO_UDP == 17 on solaris? 20:02
davidfetter allison, what's the good word on sandboxing?
Infinoid Those are the actual values that go into the ip headers on the wire, so I was hoping they wouldn't vary
rg infinoid: i don't think they have a choice, but i can confirm those are there. 20:03
allison davidfetter: reprioritized 20:04
rg i'd be very surprised if any ip stack would do translation on those constants.
Infinoid bacek: TCP is the default for IP sockets of type SOCK_STREAM, and UDP is the default for SOCK_DGRAM, so I'm not convinced
davidfetter hopes that means, "placed at a higher priority"
chromatic cotto, I think we need to change Pointer first.
allison davidfetter: lower. I don't suppose you'd be interested on working on it? I could help you get started. 20:05
cotto chromatic, how
?
davidfetter well, i can probably get People(TM) on it. what do you have?
cotto switch it to ATTRs too? 20:06
chromatic Yes.
allison davidfetter: a spec and a rough prototype that treats a sandbox as a lightweight thread
davidfetter: though I'm happy for People(TM) to take a non-threaded approach
davidfetter: in fact, I'll just chuck the prototype (twas *very* rough) 20:07
Infinoid bacek: rg checked this directly, solaris's SOCK_STREAM and SOCK_DGRAM have the reverse values of our constants. irclog.perlgeek.de/parrot/2009-03-31#i_1029542
jhorwitz +1 for non-threaded approach -- mod_parrot needs it. 20:08
cotto chromatic, Pointer's in DEPRECATED. Should it not be?
bacek Infinoid: anyway. Patch attached to ticket :)
Infinoid great, thanks 20:09
chromatic cotto, maybe so. It's used in t/pmc/nci.t though.
At least this is the problem with test #66 there. 20:10
barney Is there a --no-prefix option for Configure.pl ? Never look for libs at the location of an installed Parrot ?
20:10 dngor joined
bacek I have a question about Ref. It's deprecated, but used internally by "Undef" and "scalar". 20:11
davidfetter allison, what concurrency approaches are ok to use apart from threads?
bacek Should we deprecate "scalar" as well and merge semantic of Ref into Undef?
allison davidfetter: open to all suggestions, just bring the design discussion to me 20:13
rg bacek: could you also update the http.pir (and maybe httpd.pir, i haven't checked that one yet)? 20:15
+example(s)
bacek rg: ah. just a second.
allison davidfetter: consider it a security expert's dream: we're willing to make extensive changes for security sandboxing over the next couple of years 20:16
rg i guess a test that depends on an internet connection is not a good idea
20:16 ilia joined 20:19 PhatEddy left
dalek pp: 7b0c4c1 | (Bernhard Schmalhofer)++ | src/pct/actions.pm:
grab a magic fix from Rakudo
 20:19
shorten dalek's url is at xrl.us/benajv
bacek rg: patch attached to ticket.
rg: it is not a test. It's example :) 20:20
rg i know. i was wondering if a test might be more suitable
bacek Is ParrotThread fully functional? I can create self-contained test. 20:21
tewk chromatic: inOpaque is 0xa instead of a pointer is this in x86_64 or x86_32, 20:27
rg no idea. might be an interesting thing to try even if we need to todo it on some platforms
tewk iirc Pointer is used in several different ways by several different systems, each use case should probably have its own PMC. NCI needs a PMC that represents a raw unmanaged pointer. 20:28
NCI also needs a different managed pointer that will free itself in some cases 20:29
chromatic x86_32 20:31
tewk, I suspect this is a bad example though. cotto's patch makes all 'p' elements in an NCI signature look at the UnManagedStruct attribute storage for their data.
darbelo chromatic: Done. * Back to Ticket #529 20:35
Coke bacek: please ask the questions about Ref on the ticket. 20:36
bacek Coke: ok
tewk p elements probably aren't always UnManagedStructs, looking at UnMangedStructs code 20:38
chromatic Shouldn't they be either UnManagedStructs or ManagedStructs? 20:39
tewk I'd have to think through the use cases, but I don't think those two options are sufficient to cover the common ffi cases. 20:41
chromatic Maybe we need to call get_pointer then instead of dereferencing it ourselves.
moritz tewk: do you have a proposed project for this year's GSOC? 20:42
davidfetter allison_afk, oops 20:43
tewk moritz: whatever will get me money:)
moritz: I'm going to try to prototype LLVM Jit for parrot.
moritz tewk: I'm just asking because I've got a mail over the mentor list to ask students to submit their proposals now, so please feel informed and motivated ;-) 20:44
tewk I just saw allisons comments on my proposal so I'll probably rework it tonight and submit, will that be ok? 20:45
moritz tewk: yes 20:46
cotto chromatic, I thought that using get/set_pointer would be better, but I don't know how to make the jit code do that. 20:50
chromatic cotto, basically we have to push the two arguments (interp and PMC) onto the stack, call the function, then pop off the return value (or leave it in the right place).
... not that our current JIT code makes that simple to write. 20:51
20:51 RobertLJ left
tewk chromatic: Yeah that is the right answer, its just complicated. 20:52
have to call get_pointer. 20:53
chromatic That means we have to manipulate all of the stack state to make that possible.
tewk Let me look, it shouldn't be too bad. 20:54
cotto tewk++
20:54 bobke joined
cotto is afk 20:54
Tene tewk: do I remember accurately that you did some bytecode gen stuff? 20:55
dalek a: 73ef9d9 | (Francois Perrad)++ | src/ (5 files):
Now, works with Parrot r37824 (PGE merge)
 20:56
polyglotbot OUTPUT[Parrot VM: Can't stat languages/lua/lua.pbc, code 2.ā¤main: Packfile loading failedā¤]
polyglotbot OUTPUT[Parrot VM: Can't stat languages/lua/lua.pbc, code 2.ā¤main: Packfile loading failedā¤]
polyglotbot OUTPUT[Parrot VM: Can't stat languages/lua/lua.pbc, code 2.ā¤main: Packfile loading failedā¤]
polyglotbot OUTPUT[Parrot VM: Can't stat languages/lua/lua.pbc, code 2.ā¤main: Packfile loading failedā¤]
shorten dalek's url is at xrl.us/benaqv
dalek a: 9999c7d | (Francois Perrad)++ | t/markdown.t:
fix Markdown test
polyglotbot OUTPUT[Parrot VM: Can't stat languages/lua/lua.pbc, code 2.ā¤main: Packfile loading failedā¤]
polyglotbot OUTPUT[Parrot VM: Can't stat languages/lua/lua.pbc, code 2.ā¤main: Packfile loading failedā¤]
shorten dalek's url is at xrl.us/benaqx
polyglotbot OUTPUT[Parrot VM: Can't stat languages/lua/lua.pbc, code 2.ā¤main: Packfile loading failedā¤]
Tene ack! 20:57
rg aiieeee. didn't we already have a polyglotbot that ignored dalek? 20:58
tewk Tene: what do you mean by bytecode gen? I know the bytecode stuff pretty well.
chromatic: cotto I'll apply cotto's patch and write up a calling get_pointer, I know that code better than anyone. 20:59
give me 15minutes
Coke ETOOMANYBOTS 21:00
moritz rg: I think we did... 21:02
rg: but it went down, and it seems I restarted the wrong one 21:03
Tene: ping
rg bacek++ # patch works. i also see that patching the example is not strictly necessary, but it's nicer ;) 21:09
nopaste "tewk" at 155.98.69.7 pasted "building/testing now, but this is the fix for Jit" (11 lines) at nopaste.snit.ch/16021 21:18
tewk chromatic: cotto see paste 21:19
maybe not fails tests 21:20
chromatic t/pmc/nci.t Failed tests: 8, 29-39, 51, 54-55, 66 ? 21:21
It should pass at least 66.
tewk I'm going to have to take a closer look, if fails 20 tests 21:24
chromatic Hm, the Pointer PMC doesn't have a get_pointer vtable function. That might account for some of it. 21:25
tewk had to pop out for a minute, that was the next thing I was going to check. 21:27
it will need a get_pointer impl 21:28
Tene moritz: pong
tewk: I was having trouble figuring out why some pirc-generated pbc was failing so I could fix it 21:29
moritz Tene: after feather's reboot I started the evalbot in ~tene/pbot/ - is that the right one?
tewk Tene: tried pbc_dump -d, and diffed the outpu? is it the bytecode or some other part of the packfile? 21:31
Tene moritz: yes
tewk: probably some part of the packfile. the bytecode is equivalent afaict 21:32
moritz Tene: it doesn't seem to ignore dalek...
Tene it's the last test (failing) in compilers/pirc/t
moritz: I never told polyglotbot to ignore dalek... if someone else did I don't remember it. 21:33
tewk: Will you be online tonight?
tewk opps Parrot_UnManagedStruct_get_pointer won't work I need VTABLE_get_pointer 21:34
Infinoid I did, I think
Tene Infinoid++
Infinoid perl6: say "Infinoid-- # it stopped working!" 21:35
polyglotbot OUTPUT[Infinoid-- # it stopped working!ā¤]
chromatic Yeah, grab the vtable from the PMC and then grab the get_pointer pointer. Yay.
tewk Tene: really busy right now, maybe, send me an email description and I'll try to take a look.
Tene tewk@tewk.tewk?
tewk chromatic: yep, going to need a VTABLE_OFFSET macro
tewk.com
Tene oright, we emailed once before.
tewk tewk@tewk.com 21:36
Infinoid Tene: Is there a secret decoder ring or something I need to get into feather3? Or is polyglotbot somewhere else now?
Tene Infinoid: that's where I saw it last.
Infinoid hmm, hostname says feather
Tene apparently moritz has the secret handshake
particle mails ten.e@e.net just to see... 21:37
Tene there's an e.net now? 21:38
particle no :(
particle plays with htpp://domai.nr 21:39
er, domai.nr
moritz no, I'm on feather1
... which might not be such a good idea
Tene that splains it 21:40
Infinoid the modification I made to ignore dalek was on feather3. if you've got another checkout from source control somewhere, I didn't commit it
particle which feather vm is the special one for bots? 21:41
Tene 3
particle thought so
moritz Infinoid: can you start polyglotbot on feather 3 then? I'll kill the one on 1
Infinoid I can't ping feather3
Tene where's juerd these days? 21:42
Infinoid I would if I could log into it. But I don't think the vm was ever restarted after the server's last reboot a couple weeks ago or whenever it was
particle i think he declared irc bankruptcy 21:43
moritz maybe write a mail to him?
Infinoid I think polyglotbot on feather3 was running from daemontools... so it should start automatically when the vm comes up 21:44
... I'm not positive of that.
21:47 TiMBuS joined
PerlJam so feather3 is dead dead dead? 21:47
Infinoid No idea. 21:48
All I know is it isn't reachable from the intarwebs
Tene maybe I'll run one at home.
Infinoid moritz: Was your copy using that two-parrot-checkout rotation script I wrote? I don't have a backup of that. 21:50
moritz Infinoid: I didn't even look at the build scripts...
Infinoid: so it looks like "no"
Infinoid Eh. The logic wasn't too complicated 21:52
rg so nobody has access to start/check the vm? 21:53
moritz Juerd++ has 21:54
Tene juerd does
particle who else is in .nl? maybe they can walk over :) 21:56
Tene it can't be much bigger than portland
rg and knock him out of bed? i'm not sure he's going to be too happy about that ;) 21:57
Tene they both end in 'land'
chromatic bedland? 21:58
Tene man, I love that place 21:59
rg dreamland :)
Tene that's where I'm a Parrot.
jonathan No way is a quarter of 2009 already over... 22:00
PerlJam Tene: is each feather representative of a programming language?
Tene PerlJam: yes, but half of them are scheme 22:01
PerlJam Tene: and one of them is quite a bit larger than the others and looks like a camel feather? :)
Tene ).) 22:02
22:09 Limbic_Region joined
Infinoid feathered camels: www.discogs.com/image/R-1239369-1205608920.jpeg 22:24
(one of the weirder albums I own.)
22:25 kid51 joined 22:28 tetragon joined 22:31 tetragon_ joined 22:44 dduncan joined 22:45 dduncan left
cotto braces himself for slew if April 1st articles. 22:45
s/if/of/
dalek rrot: r37831 | jkeenan++ | branches/dir_simplify:
Create branch to work on TT #524.
kid51 At $job, we got our March 31 No Fooling today: Salary freeze; no more company matching 401K contributions. 22:46
davidfetter kid51, ouch! looking for another one? 22:58
szbalint it's already april 1st here
szbalint braces for the idiocy
haha, I'm getting april 1st malware spam, that's at least mildly refreshing 23:00
moritz I liked that one: www.smashingmagazine.com/2009/03/31...es-leaked/ 23:01
shorten moritz's url is at xrl.us/benar6
jonathan What's so special about April 1st? 23:03
cotto It's April Fool's Day, when everyone (in the US at least) tries to play tricks on eachother. 23:04
jonathan HAH you fell for it ;-)
jonathan knew full well :-)
cotto I realized that just after I pressed enter.
jonathan++
japhb Had to trash my git svn clone of parrot ... is there a public clone/snapshot I can grab? 23:06
23:07 shucho joined
rg "Server-side code decompiler" *lol* 23:10
Coke_afk wow, like 3 damian posts in as many days. 23:20
wayland76 In Australia, there's a rule that if you play an April Fools' joke after noon, you're the fool instead of the other person 23:21
Coke oh, I thought it was that the joke day was actually on October 1st. 23:27
tewk chromatic: an afternoon later... 23:34
t/pmc/nci.t (Wstat: 1792 Tests: 69 Failed: 7) Failed tests: 33-34, 36-38, 51, 54
chromatic That's one fix, anyway.
tewk patch in a minute 23:35
23:36 Whiteknight joined
Infinoid japhb: git://squawk.glines.org/parrot-trunk/ 23:39
nopaste "tewk" at 155.98.69.7 pasted "jit_defs.c.patch" (67 lines) at nopaste.snit.ch/16023
japhb Infinoid: Does that really only contain trunk? 23:40
tewk chromatic: the first hunk is the fix, the second hunk is from cottos patch
Infinoid japhb: yes
tewk so is there an official git mirror for parrot yet? 23:41
japhb Infinoid: if I clone from there, how do I re-root to the real svn repo?
tewk: I wish.
Actually, I wish parrot would just switch to git
Infinoid japhb: I suspect you just add a couple of lines to your .git/config, but I haven't tried it 23:42
one moment, I will
japhb Infinoid: If it was a real git repo, I'd just add another remote ref for it, but I never know where the edges of git-svn's magic are (as in, what would horribly break things) 23:43
Infinoid oh my, adding a [svn-remote "svn"] clause to .git/config and "git svn fetch" started at -r1 23:44
I can always just tar it up for you.
japhb Nah, don't worry about it.
Infinoid: thanks, though 23:45
Infinoid ok. anyway, it's as close to an official git mirror as we seem to have. A script updates it once every 10 minutes, and it's been there for months
japhb Ah, now that's interesting.
push'able?
or r/o?
Infinoid r/o
It started out as my bounce repo for dcommits, but I've started being more careful now that others are using it 23:46
If you've got ideas on how to make it pushable and get the svn usernames right and all of that, I'm happy to host it 23:47
japhb Let me think on it ... have to go AFK for a bit
Infinoid seeya

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!