Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Sat Mar 15 | tinyurl.com/2pmnlq
Set by moderator on 8 March 2008.
Coke "public" means "even though I'm logged in, treat me like a guest." 00:41
I find it somewhat annoying when I go to a Public/ link but mean to be doing business.
00:42 kid51 joined
moderator Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Mar 31 | tinyurl.com/2pmnlq 00:44
Coke kid51: heh. I just saw that use perl article. =-) 00:46
Infinoid yeah, would be nice if Public would redirect or at least link to the real page, if you're logged in 00:51
kid51 Earlier this week I was on another channel with jmcadams. He says he has lots of good submissions, but felt he had enough lead time to extend the deadline a little. 01:01
svnbotl r26434 | coke++ | trunk: 01:02
: [tcl]
: Going forward, no tests are expected to fail in tcl's test suite during
: a release. Smoke away.
diff: parrotvm.org/svn/parrot/revision/?rev=26434
Coke seen kj? 01:16
purl kj was last seen on #parrot 2 days and 4 hours ago, saying: coke: yy_switch_to_buffer error is probably caused by the macro stuff in tcl (you got it with tcl right?) [Mar 14 14:25:00 2008]
Coke msg kj (TODO: { local $TODO; $TODO = 'pwd is broken on windows' if $^O eq 'MSWin32';
purl Message for kj stored.
Coke msg kj er... ignore that last one. yy_switch_to_buffer_error was in tcl, yes. 01:17
purl Message for kj stored.
Coke I wonder if we could add :default handling to our already copious arg handling. 01:31
kid51 In what part of Parrot? 01:34
Coke PCC
(the calling conventions)
kid51 Oh. Not my territory.
Coke if I want a hunk of C code that follows the PCC, is a method on a PMC my easiest choice? 01:56
02:05 Theory joined
svnbotl r26435 | coke++ | trunk: 02:23
: [tcl]
: minor doc updates
diff: parrotvm.org/svn/parrot/revision/?rev=26435
Tene What's the difference between 'resolved' and 'rejected' in RT? 02:49
There's a ticket there for the old implementation of cardinal, which isn't around anymore.
Infinoid one means "fixed", the other means "not going to fix"
Tene Which would be more appropriate for this ticket?
Coke rejected. But with a nice comment explaining why.
Coke finds a bug in tcl that he introduced back in August. 03:05
jdv79_ you really like that tcl, huh?
Coke At this point, I feel responsible for inflicting it on everyone. 03:09
yay, fixed. 03:15
svnbotl r26436 | coke++ | trunk: 03:19
: [tcl]
: Fix a bug I introduced in the conversion of this method from PIR to C back in
: r20422: lists that contain quoted elements would abort list processing.
: This reclaims 2 TODO'd tests.
Coke stares at PAST. 04:20
Tene stares at Coke. 04:21
Coke Tene: care to help me fix something in tcl? =-) 04:22
Tene Sure.Tell me what's going on and I'll offer what i can. 04:24
Coke ok. tcl's commands are like: [puts foo] - that is parsed as a command of puts, with a single arg of foo. 04:25
Tene 'kay
Coke tcl8.5 introduces a syntax that lets you do {*}{some list of things}, and that gets expanded at compile time to replace the whole thing as if you had done:
[list some list of things] - it turns a single word into multiple words.
this works in partcl now except when the {*} is on the first word in the command (the command itself). 04:26
Tene Okay, and {some list of words} is a quotation of some sort?
Coke so you should be able to do {*}{puts hi} and have it execute the command puts with the argument hi.
Yes. { is like ' in perl.
Tene 'kay 04:27
Coke so, in pge2past.tg, in transform past (command) {; between command = shift words and command= tree....; I've added a check there to see if 'expand' is defined. If so, I have to transform the command ... I haven't quite figured out the proper transformation step there. :| 04:28
Tene looks. 04:30
nopaste "Coke" at 72.228.52.192 pasted "wip" (26 lines) at nopaste.snit.ch/12520 04:31
Coke So, I'm about 90% certain I've got the hook in the right place there; Now I don't know what to do about it. =-)
Tcl has a .dumper() macro you can use to dump a particular node to test. 04:32
(delete the comment line for that to actually do anything. :( 04:34
er, the comment line with the {*}
Tene Hmm. I don't understand what's going on there. Reading more... 04:36
Coke (the code that is commented out is also suspect) ; well, at the moment, it's not doing anything different, really; just demarking the place where something that did something would have to go.
Tene The acronym associated with a .tg file is TGE, yes? 04:37
Coke yes. 04:38
tcl isn't using PCT yet.
Tene Right.
Tene goes to read TGE docs.
Coke Hurm. I wonder if I should be doing this in the next layer down (past2pir instead of pge2past) 04:40
Tene pge2past looks like the right place to me. 04:43
the thing-to-expand is going to be in the [0] of <word>, yes? 04:45
Coke I believe so. moment.
hurm. I think it is rather the node that is left when you delete the expand entry. 04:47
Tene Hmm. Okay.
So you need to split it on spaces? 04:48
Coke ... basically, though tcl's parsing is more complicated than that. 04:49
(needs to deal with all the various forms of substitution which affect "what is a word.
Tene Okay, is there a way from pge2past to call that parsing rule on that object? 04:50
Coke Iunno. =-) 04:53
Tene Would "split on spaces" work for testing right now? 04:54
Coke I can get to the string "puts hi" (which is the canonical form of {puts hi} at this level). I've been trying to come up with a way to transform that via TGE *again* to something that the command node can handle by itself, and am failing. 04:55
(for testing) not convinced, as I can probably come up with a test that will fail it shortly.
Tene That is , skip this part to see if you can get the rest right, and if this is the right place to be doing this. 04:56
Coke {*}{puts "hello world"}
Tene Would it help if I migrated tcl to PCT? ;) 04:57
Coke It might. I had convinced KJ to poke at that as well.
Tene Why is functions.pir in src/grammar/expr instead of src/builtins/ 04:59
?
Coke those functions only make sense in the [expr] commands sub-language.
and historical reasons. 05:00
Tene Okay.
they're called at runtime, or at compile-time?
Coke hurm. Depends on how clever mdiep was.
it would only be when the particular [expr] command was compiled-to-be-run, though. 05:01
Tene (note: I don't know tcl)
So the .tg is basically PIR with some magic sugar, yes? 05:02
Is there a rule in expr.pg that you'd like to call to parse this string? 05:05
Coke command seems like the logical choice. 05:06
(at least in this case where we are expanding into a command) 05:07
Tene So get_hll_global tclexpr::grammar::command, then call it on the string?
or whatever get_whatever is appropriate here?
Coke You should be able to achieve the same effect with a TGE dispatch call, I believe.
Tene grep can't find 'dispatch' mentioned anywhere in compilers/tge 05:08
Coke grep (ack) for dispatch in that file. 05:09
Tene Okay.
Looks like 'dispatch' is a local rule defined in that file, not a TGE thing. 05:11
Coke ja.
I think it's a common tge idiom, fwiw.
svnbotl r26437 | coke++ | trunk: 05:12
: [tcl] make this TODO test harder to pass with more alternatives.
diff: parrotvm.org/svn/parrot/revision/?rev=26437
Tene Okay.
Reading through 'transform dispatch', it doesn't quite look like what we want.
It tries to pull the past out of an already-parsed object.
We need to do some actual parsing. 05:13
Yeah?
purl totally dude!
Tene purl: don't forget Yeah?
purl no idea, tene
Coke ah. currently, the Expand stuff is basically deferred until the past2pir stage, where __list (a helper PIR function) converts it to a TclList.) 05:15
Tene Huh.
Coke Ok. I'm more convinced now that it needs to go in the past2pir layer, but haven't figured out how to let the DynamicCommand entry there know its dealing with something that ISA past:expand. Presumably that needs to be in the PAST as well. 05:30
"blah". 05:31
Thanks for taking a look.
-> zzz
Tene g'night 05:33
Coke: is there a way to get a past dump from tcl? 05:55
It was pretty easy to get that grammar working in PCT. 05:57
Coke Tene: zzzzz don't know of an easy way no. zzz 06:01
06:03 jrt4 joined
Tene SleepCoke: thanks. 06:12
Okay, tcl doesn't look too bad. I might be able to hack on it tomorrow night. 06:19
06:58 IllvilJa joined
svnbotl r26438 | chromatic++ | trunk: 07:28
: [Tcl] Minor cleanups in the documentation. (First I saw a typo, then I went
: crazy.)
diff: parrotvm.org/svn/parrot/revision/?rev=26438
07:30 uniejo joined
svnbotl r26439 | fperrad++ | trunk: 08:18
: [WMLScript]
: - ppd17pmc
r26440 | fperrad++ | trunk:
: [WMLScript]
r26442 | fperrad++ | trunk: 08:25
: [Lua]
: - add lua_getmetatable() & lua_newuserdata()
: - many docs
r26443 | fperrad++ | trunk:
: [Lua]
: - refactor IO with lua_newuserdata & lua_getmetatable
r26444 | fperrad++ | trunk:
: [Lua]
: - add md5 library
diff: parrotvm.org/svn/parrot/revision/?rev=26444
08:29 AndyA joined
svnbotl r26445 | fperrad++ | trunk: 08:32
: [Lua]
: MANIFEST
diff: parrotvm.org/svn/parrot/revision/?rev=26445
purl well, : MANIFEST is updated accordingly, and file props are set. yay!
svnbotl r26446 | fperrad++ | trunk: 09:16
: [build]
: - fix the following error
: src\\packdump.c:39: error: nonnull argument references non-pointer operand (arg 1, operand 2)
diff: parrotvm.org/svn/parrot/revision/?rev=26446
09:33 mire joined
spinclad (can't get purl to forget silly ': MANIFEST' factoid) 10:30
10:43 ruoso joined 10:50 kj joined
Coke : MANIFEST ? 12:03
purl, : MANIFEST ?
purl no idea, coke
Coke : MANIFEST
the trick is to set the factoid to be: "<reply>" - the empty reply is saved, but not announced. 12:04
12:42 skv_____ joined 13:05 gryphon joined 13:51 Andy joined
svnbotl r26447 | fperrad++ | trunk: 14:12
: [digest]
: - remove SHA256 & SHA 512 (not available with libssl 0.9.6)
: src/dynpmc/Makefile must be regenerated.
diff: parrotvm.org/svn/parrot/revision/?rev=26447
14:41 rdice joined 14:56 davidfetter joined 15:05 skids joined
cognominal someone working on rakudo triggering a bus error when evaluating something after catchin an undefined variable error? 15:29
I see in gdb that the flag for outer context is set when there is none. 15:30
outer_ctx = 0x0 15:31
Coke rt.perl.org/rt3/Ticket/Display.html?id=49758 15:37
no one is assigned to it, though, so I'm guessing it's not being worked on.
cognominal I don't know much of the internal... 15:38
but clearly the flag is wrong
15:38 parrot-poke joined
cognominal I don't know if anynone else gives a label to push_eh instead of a sub. 15:39
15:46 lidden joined
Coke sure. 15:57
(pretty much any exception handler in tcl works that way) 15:58
nopaste Someone at 82.67.232.89 pasted "rt 9758. proposed fix : reset $?SUB between to compilations" (15 lines) at nopaste.snit.ch/12523 16:29
cognominal I am not sure it does not break anything else but if fixes the problem.
I ahe not figured out yet how to edit a ticket in RT 16:30
ok , edited it. 16:35
Coke if no other tests fail, I suspect you're good. 17:07
ah. but the patch is in PCT, so it's hard to tell. 17:11
? Why is PCT setting @?BLOCK in the first place?
cognominal I am not so familiar with the whole compilation chain 17:13
17:21 pjcj joined
Coke more of a question for patrick. 17:22
the answer may be "parrot needs something like that, and we needed the perl6 version *anyway*, so...
17:25 adc joined
Coke "/win 3 17:26
17:36 cout joined 17:43 Theory joined
kj Coke: @?BLOCK: that's the "scope stack" 18:02
not sure if i'm referring to the same @?BLOCK as you; I'm referring to the one that is set in many languages, such as rakudo 18:03
cognominal that means I probably set $?SUB according to the content of @?BLOCK instead of forcing it to null. 18:07
s/set/should set/ 18:08
18:24 Tene joined
Coke kj: it's not a *parrot* scope stack, though. 18:26
it's specifically being managed by the HLLs here, using a very perl6-ian name. neh?
kj Coke: right, I think. In any case, I know @?BLOCK as a HLL managed thingie, that is used in the action methods to put PAST::Block's on 18:27
Coke: right (your last remark)
Coke ok.
kj but I just saw your remarks, and figured it sounded familiar to me, so I replied
cognominal pct is supposed to be language agnostic? 18:28
kj cognominal: can you rephrase what you mean?
18:29 Psyche^ joined
cognominal I supposed that pct is designed to compile any language 18:29
PerlJam cognominal: Yes.
cognominal: pct *is* language agnostic afaik 18:30
Tene Someone should talk more about the first item in the topic.
cognominal but I suppose that some concept like blocks happens in any language
kj yes, i think so too
Tene bbl.
kj cognominal: if you're referring to the @?BLOCK "trick": it's used in pretty much all PCT based languages
cognominal so it is reasonnable to name them like they would be in Perl 6.
kj (I use it all the time) 18:31
svnbotl r26448 | chromatic++ | trunk:
: [WMLScript] Fixed a segfault, making more string tests pass. Now only one
: fails, and it's not a segfault.
diff: parrotvm.org/svn/parrot/revision/?rev=26448
cognominal after all. this @?BLOCK lives in the parrot namespace.
I see that rakdo does not yet use Perl 6 namespace.
Coke I would argue about the reasonability of using perl6-style names in a language-agnostic vm. 18:32
cognominal I have tentative code that generates .HLL thelanguage but it would break any other language using pct 18:33
at least Perl 6 is documented so It makes things easier so far as this varaible lives in a separate space. 18:34
18:50 skids joined 18:56 barney joined 19:27 adc joined
Coke (newfrom) ... huh. didn't realize that was gone. 19:59
20:00 mire joined
barney Being paranoid, I also looked for new_from :=) 20:03
svnbotl r26449 | coke++ | trunk: 20:05
: [tcl]
: Avoid using "new $I0"
r26450 | bernhard++ | trunk:
: No need to check for Perl 5.006, when Parrot
Coke Parrot? 20:06
purl Parrot is our teacher, our mother, our secret lover or the reason Dan started or the reason Dan left or pretty onionish:)
cognominal on croirait ma mere...
oops
Coke wonders if babelfish's translation there is accurate. 20:09
20:10 sjansen joined
barney Who need Perl 5.6 when you have Parrot? 20:12
svnbotl r26451 | bernhard++ | trunk: 20:21
: [config]
: A litte code cleanup, no need to check for $] >= 5.006
diff: parrotvm.org/svn/parrot/revision/?rev=26451
PerlJam I just tried to compile parrot on the RHEL4 box at work and it doesn't compile, so I'd say perl 5.6 is quite a bit better than parrot on that box :)
Tene Hm. Lemme try it on a fresh RHEL5 box. 20:25
kj good day 20:28
purl every day above ground is a good day
kj I read in readme_cygwin.pod that parrot builds on cygwin "out of the box"
well,not from my box :-/
barney Perljam: If it doesn't compile, then you don't have it 20:30
PerlJam yeah, and I'm slightly surprised that it didn't compile. But I don't have time right now to debug, I just thought I'd try it on a lark. 20:32
Coke I haven't had a working cygwin build in aeons.
Coke tries again on a new windows box.
PerlJam Coke: cygwin has been working fine for me for a while now.
(in fact, that's what I've been using at home)
kj PerlJam: i remember some path setting that must be done wrt blib 20:33
Coke PerlJam, kj: even with that path setting, it NEVER worked for me.
tries again.
kj coke: right, I always get this error 53 thingie
Coke (PerlJam- if it works for you, I'd appreciate it if you could go through the queue and start weeding out the old cygwin failure messages.)
PerlJam Coke: do you have strawberry perl installed at all?
Coke PerlJam: I do on this box, yes. 20:34
PerlJam I did notice that stawberry perl and parrot don't seem to play well together.
Coke I didn't on the one it never worked on.
(I had a system activestate, but wasn't using that one.)
PerlJam hmm.
Tene Looks like Parrot is building fine... the only thing I had to install was gcc.
20:34 Ron joined
kj that's strange isn't it? Cygwin installs should be the same everywhere, except for installed packages I thought 20:35
Coke which readme is this in that says it works?
kj README_cygwin.pod
a new file
purl well, a new file is truncated.
Tene Let's run 'make test' now. 20:36
Coke WOOHOO! 20:38
Congratulations!
Your organization "The Perl Foundation" has been accepted in to the
purl Congratulations! are you joel?
Coke Google Summer of Code(tm) 2008
svnbotl r26452 | bernhard++ | trunk:
: [test]
: Run the examples tests and the manifest tests with 'make fulltest'.
diff: parrotvm.org/svn/parrot/revision/?rev=26452
Tene Coke: So, what's the significance of the first item in the topic? 20:39
PerlJam Tene: more mentors needed perhaps.
Coke every student needs a mentor.
PerlJam I'd mentor someone but I think I'd suck at it :-) 20:40
Tene What's involved in that?
Coke basically, keep them on track, answer any questions they may have, etc.
I unno, never done it.
mentor?
purl mentor is best known for writing www.phrack.com/show.php?p=7&a=3 , which was quoted in the movie "Hackers" or Lloyd Blankenship
Coke soc mentor?
soc?
purl well, soc is www.perlfoundation.org/perl5/index.cgi?gsoc2008
Tene I could probably do that. 20:41
The only error in 'make test' when built on clean RHEL5 is t/examples/pasm.t 20:42
error:imcc:syntax error, unexpected IDENTIFIER, expecting $end ('use') in file 't/examples/pasm.t' line 5
barney Did you run with './parrot' ? 20:45
Tene Yes.
svnbotl r26453 | bernhard++ | trunk:
: [test]
: Run the tests in t/run when running 'make fulltest'.
diff: parrotvm.org/svn/parrot/revision/?rev=26453
Infinoid hrm, I'd make a horrible mentor, I have enough trouble just keeping *myself* on track 20:46
Tene Ack, looks like I didn't update the expected output of examples/pasm/fact.fib
Coke it would be nice if the cygwin readme listed the bare minimum of packages required. (like, e.g. gcc-core) 20:48
kj Coke: did you get cygwin running?
Coke (and perl-whatever)
kj parrot on cygwin
Coke kj: haven't touched it since I mentioned it earlier.
kj oh ok
Tene I didn't realize examples were tested.
Coke Now you know.
purl and knowing is half the battle!
barney Tene: mkraii had sent a patch in RT#51806
20:49 Psyche^ joined
kj PCT tutorial readers: episode 3 is up! 20:49
barney Tene: rt.perl.org/rt3/Ticket/Attachment/3...orial-test 20:50
Coke kj++
Tene kj: you don't mind if I steal some of your structure and organization for the presentation I'm working on, right?
kj sure 20:51
eh, sure not mind
whatever that is in english :-P
what presentation is that?
Tene kj: I'm presenting on Parrot/PCT at penguicon.org/ 20:52
kj if you want I can send you Squaak's --the tutorial showcase language-- implementation (I already completed it myself)
Tene I put an example language together myself already.
kj oki
Tene barney: thanks. I'll add that, too. 20:53
20:53 liona29 joined
Tene kj: github.com/tene/gil/tree/master 20:53
20:53 Psyche^ joined
kj Tene: maybe you could mention the tutorial at the end of your presentation (A) 20:57
Tene sure!
kj I'm planning to finish it this week; 5 more episodes
cotto_work kj++ 20:58
barney Are there any news regarding 'events' ? 20:59
Infinoid is anyone working on this failure I'm seeing in t/dynpmc/digest.t ? 21:03
svnbotl r26454 | rblasch++ | trunk:
: [src] Put const on right side (literally.)
diff: parrotvm.org/svn/parrot/revision/?rev=26454
Infinoid seen fperrad 21:10
purl I haven't seen 'fperrad', Infinoid
barney Infinoid: Do you happen which package contains the openssl header files in Ubuntu ? 21:12
Infinoid probably libssl-dev or libcrypto-dev, let me look 21:13
Tene barney: for the fix in Parrot_IO.t do we really want to be explicitly checking for git, or do we just want to skip on all non-svn? 21:14
On rt.perl.org/rt3/Ticket/Attachment/3...t-failures ?
shorten Tene's url is at xrl.us/bhs9y
Tene Would s/if/unless/ be better than the proposed s/svn/git/ on that line? 21:15
svnbotl r26455 | tene++ | trunk:
: Update t/examples/pasm to match longer output for fixed fact.pasm example.
: Skip fact.pasm when GPM isn't available. (RT #51806)
diff: parrotvm.org/svn/parrot/revision/?rev=26455
Infinoid detecting svk has typically been a problem in the past, too
Tene Does svk also have issues with expanding keywords? 21:16
nopaste "Infinoid" at 75.28.79.197 pasted "TODO t/dynpmc/digest.t failures caused by r26447" (22 lines) at nopaste.snit.ch/12527
Infinoid I don't think so, it mostly acts like svn 21:17
barney I'd play it safe, and skip all non-svn.
Tene Thanks. 21:19
barney Infinoid: Could you commit digest.t ? The crypt PMCs can be readded after the release. 21:20
Infinoid will do 21:21
barney: Committed revision 26457.
Tene I like closing tickets. :D 21:22
Infinoid barney: its libssl-dev. did that work? 21:26
svnbotl r26456 | tene++ | trunk:
: Fix the direction of an svn-only test. (RT #51806)
r26457 | infinoid++ | trunk:
: [t] r26447 removed SHA256 and SHA512. So TODO out the tests that check for them.
diff: parrotvm.org/svn/parrot/revision/?rev=26457
barney Infinoid: yes, after installing libssl-dev I also got those errors. 21:28
Infinoid r26457 = All tests successful on x86-64 linux :) 21:29
barney Probably not for 'make testC' 21:31
svnbotl r26458 | jkeenan++ | trunk: 21:32
: Correct typo in SKIP message: 'GMP' rather than 'GPM'.
diff: parrotvm.org/svn/parrot/revision/?rev=26458
barney jkeenan++ 21:33
Tene jkeenan++ 21:35
barney ANTLR3 seems to get optok parsing 21:37
Infinoid barney: under testC, the final test of digest.t segfaults calling MD5_Init() 21:40
Infinoid is clueless about testC
the md5 pmc's SELF pointer is null. 21:44
barney Infinoid: Sounds like RT#51790
Infinoid looks like it indeed, -S runcore has the same symptoms 21:46
Infinoid digs 21:50
barney Are GDBMhash and the crypt PMCs the only PMCs that load external libs? 21:53
21:55 lidi20 joined 21:58 slightlyoff joined
barney Infinoid: good luck with digging 22:07
I'm calling it a day. I plan on making a release branch tomorrow around 16:00 UT 22:08
22:08 Limbic_Region joined
Infinoid ok, thanks! 22:12
22:21 kid51 joined 22:24 jenny20 joined 22:57 adc joined
Infinoid I'm trying to debug rt#51790. Is there anything I can RTFM for the NCI interface? 23:01
Infinoid finds the pdd16 draft 23:02
23:35 rdice joined 23:53 skids 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!