Parrot 2.7.0 "Australian King" Released! | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 25 tickets; merge branches; review Git conversion plan
Set by moderator on 7 September 2010.
00:19 kid51_at_dinner is now known as kid51 00:25 patspam left
kid51 taptinder reporting build failure in src/hash.c 00:25
luben it's fixed in next revision 00:27
00:29 patspam joined, patspam left
NotFound (In case no one noticed, C pointers have stars * ) 00:30
dalek rrot: r48946 | jkeenan++ | trunk/src/global_setup.c:
[codingstd] No trailing whitespace.
 00:36
rrot: r48947 | NotFound++ | trunk (5 files):
remove deprecated function Parrot_get_runtime_prefix, TT #1191
00:43 whiteknight joined
chromatic Hm, looks like the ops changes cut out a lot of the speed gains of hash inling. 00:44
luben there is small benefit of hash inlining, but not so dramatic 00:45
plobsing what? how? 00:46
purl what? how is it ad hominem?
dalek TT #1191 closed by NotFound++: Deprecate Parrot_get_runtime_prefix
TT #1191: trac.parrot.org/parrot/ticket/1191
00:46 kid51 left
luben plobsing, I have merged a branch that moves runtime indirection in hashes to compile-time indirection 00:47
plobsing yes. I understand that, but the ops changes (I'm assuming the ones I made and not some changes I'm not aware of) should be independant of that. 00:48
luben I'm not sure that this are ops changes
dalek tracwiki: v14 | NotFound++ | ParrotDeprecations 00:49
tracwiki: trac.parrot.org/parrot/wiki/ParrotD...ction=diff
tracwiki: v11 | NotFound++ | ParrotDeprecationsFor2.9
tracwiki: trac.parrot.org/parrot/wiki/ParrotD...ction=diff
cotto NotFound++ for properly nuking deprecated stuff 00:52
dalek rrot: r48948 | NotFound++ | trunk (5 files):
remove remaining mentions of RetContinuation in comments and docs
00:53 patspam joined 01:00 kid51 joined
kid51 Does anyone know how I can post at www.parrotblog.org/ 01:01
whiteknight kid51: that blog is basically defunct and unused, as far as I know 01:04
certainly nobody is using it
plobsing I can get about a 2% performance improvement on running a null PIR program by pre-sizeing the const_cstring_hash. worthwhile? 01:05
01:05 dngor_ joined, dngor left
whiteknight plobsing: we should know at compile-time exactly how large that hash needs to be 01:05
cotto plobsing, any improvement for a non-null pir program?
kid51 whiteknight: Well, in TT #1164, coke pointed out that it could be used by any developer who didn't have a blog of his own 01:06
plobsing cotto: sure, but less dramatic to be sure.
cotto ship it
whiteknight kid51: I just checked, I don't have permissions to add new editors there
kid51 And since I don't blog as frequently as some people I know, I figured that was a good place. It's supposed to be aggregated to planet.parrotcode.org, as well. 01:07
whiteknight kid51: Unless chromatic knows something relevant, I suggest sending Allison a msg about it and see if she knows how to add contributors and maybe even meta-contributors to it
kid51: parrot.org also supports blogs, and you already have an account there if you want to use them 01:08
(I'm not telling you where to go, just pointing out options)
dalek rrot: r48949 | NotFound++ | trunk (4 files):
delete deprecated mehtod is_tty in handle PMCs, TT #1689
 01:09
rrot: r48950 | plobsing++ | trunk/src/string/api.c:
pre-size the constant string hash. gets ~2% performance improvement on parrot startup.
01:09 contingencyplan left
cotto plobsing++ - good idea there 01:11
01:15 patspam left
NotFound kid51: posterous is good for low frequent posters. 01:16
01:17 jimk joined
dalek TT #1689 closed by NotFound++: Deprecate is_tty 01:19
TT #1689: trac.parrot.org/parrot/ticket/1689
jimk whiteknight Where at parrot.org are those blogs? 01:20
01:20 kid51 left
whiteknight jimk: login. On the right click "create content". in the submenu, click "blog entry" 01:21
jimk Hmm, my submenu only offers: poll scratch 01:22
dalek rrot-linear-algebra: 4790ad1 | Whiteknight++ | setup.nqp:
update setup.nqp to include all the necessary metadata for a release. With this I can now make the sdist target and produce a tarball of a working distribution.
jimk I hate losing my identity
dalek tracwiki: v15 | NotFound++ | ParrotDeprecations 01:23
tracwiki: trac.parrot.org/parrot/wiki/ParrotD...ction=diff
tracwiki: v12 | NotFound++ | ParrotDeprecationsFor2.9
tracwiki: trac.parrot.org/parrot/wiki/ParrotD...ction=diff
01:24 jimk left 01:25 kid51 joined
NotFound I also don't have the "blog entry" item. 01:26
whiteknight jimk: what's your username there? I'll try to give you more permissions 01:32
NotFound: if you're interested, I can set that up for you too
more blogging about parrot == more awesome 01:33
NotFound whiteknight: if doesn't take much work, do it just in case.
kid51 jimk is back to being kid51 ... but at parrot.org, he's jkeenan
whiteknight kid51: try now 01:36
NotFound: you try also 01:37
kid51 whiteknight: Got it. Thanks. 01:38
NotFound My god! It's full of items! 01:39
whiteknight awesome. I've learned something new
NotFound whiteknight++
02:07 whiteknight left
kid51 has blogged: www.parrot.org/content/gathering-p...october-16 02:09
dalek website: jkeenan++ | A Gathering of Parrot Developers in the Pacific Northwest, Saturday, October 16 02:11
website: www.parrot.org/content/gathering-pa...october-16
cotto kid51, you forgot an important point 02:16
two words: "Ballmer peak"
kid51 only partly gets cotto's joke 02:17
Is that related to "Ballmer speaks"?
cotto ballmer peak?
purl hmmm... ballmer peak is xkcd.com/323/
kid51 Heh 02:19
cotto Sometimes it doesn't hurt to ask purl.
kid51 notes that if we had dumped purl, we wouldn't have gotten that link. 02:20
purl++
cotto I was going to say "never", but that'd be a lie
aloha, status
02:20 cotto is now known as aloha
aloha not present 02:20
02:20 aloha is now known as cotto
kid51 purl botsnack 02:20
purl :)
cotto thanks, aloha
kfsahafdhnlkshflsahfisnhfdbotsnackfilledwithtoxicwasteajjflkewjljladas 02:21
purl thanks cotto :)
02:35 janus left 02:36 kid51 left 02:42 Coke left
chromatic plobsing, 2% is worthwhile. 02:43
cotto I figured you'd say that. 02:45
It's hard to see a cost to that optimization in any case. 02:46
chromatic Anything we can know at compilation time we should fix at compilation time. 02:47
Hm, parrot_hash_oplib() is more expensive than I remember. 02:48
plobsing chromatic: that's because I just changed it to not be a horrible hack on core_ops' internal hash. 02:50
it is slightly more expensive as a result.
chromatic Cleanliness is good. 02:51
plobsing the benefit is that it now uses the same hashing as everything else
also the oplib specific hashes can be made static data now (trying to work out exactly how I'd do that atm) 02:52
particle cleanliness is next to 'dude it fucking worksiness!' 03:02
plobsing cleanliness?
purl cleanliness is next to godliness.
plobsing aw. I was hoping purl had picked that one up.
particle purl, no, cleanliness is next to 'dude it fucking worksiness!'
purl okay, particle.
particle biked 100 miles today, had a muscle relaxer known as 'beer', and is ready for a shower 03:04
03:04 atrodo joined 03:19 janus joined 03:55 petdance joined 03:59 tcurtis left 04:07 dngor_ is now known as dngor 04:56 bacek joined, aloha joined 05:30 elmex left 05:37 petdance left
cotto dukeleto, ping 05:53
06:04 fperrad joined 06:28 jhelwig joined 07:00 chromatic left 07:52 x3nU left 07:53 x3nU joined
cotto dukeleto, unpong 08:01
er, unping
08:02 elmex joined 08:34 tadzik joined
dalek rrot: r48951 | fperrad++ | trunk/runtime/parrot (4 files):
[library] completes the implementation of skip_all
 09:29
rrot: r48952 | fperrad++ | trunk/t/dynpmc/gziphandle.t:
[t] refactor with skip_all
09:51 M_o_C joined 10:03 M_o_C left 10:06 JimmyZ joined
JimmyZ msg kid51 '2030 UTC Tuesdays' is typo? see www.parrot.org/content/gathering-pa...october-16 10:08
purl Message for kid51 stored.
aloha OK. I'll deliver the message.
10:08 JimmyZ left 10:41 contingencyplan joined, mikehh joined
tadzik wklej.org/id/387906/ -- you might want to look at this guys 10:42
it's a perl 6 script
10:46 mikehh left 10:55 mikehh joined 11:17 whiteknight joined
dalek kudo: fdae00f | colomon++ | src/core/Rat.pm:
Add infix:<**>(Rat $a, Int $b).
 11:37
ast: 508e297 | colomon++ | S32-num/rat.t:
[t/spec] Add two tests to verify that Rats raised to positive Int powers are still Rats.
ast: 367d08a | colomon++ | .gitignore:
[t/spec] Add .gitignore set to ignore *.rakudo.
11:41 plobsing left, plobsing joined
dalek ast: 3dfdedd | KodiB++ | S02-builtin_data_types/keyhash.t:
KeyHash.elems should return the sum of the values, not the number of values.
 11:44
rrot: r48953 | nwellnhof++ | trunk (3 files):
Make some functions in hash.c static
 11:58
NotFound tadzik: there is a problem in ByteBuffer, I'll fix it, thanks. 12:08
tadzik NotFound++
12:18 Coke joined
NotFound tadzik: fixed, hope that it solves your problem. 12:28
dalek TT #1762 closed by nwellnhof++: Memory leak in Parrot_find_method_with_cache 12:29
TT #1762: trac.parrot.org/parrot/ticket/1762
rrot: r48954 | NotFound++ | trunk (2 files):
fix getting string from uninitialized ByteBuffer and a add a test for it, tadzik++ for the catch
 12:32
12:39 kid51 joined 12:45 jan left 12:49 jsut joined 12:53 jsut_ left
kid51 msg NotFound trac.parrot.org/parrot/ticket/1777 isn't actually closed, but it might be a good idea to keep it open for 2 days so that doughera has a chance to respond. 12:57
purl Message for notfound stored.
aloha OK. I'll deliver the message.
NotFound Uh... I got distracted and forgot to check the radio button. 12:58
kid51 msg Cotto Is trac.parrot.org/parrot/ticket/749 still pertinent? 13:03
purl Message for cotto stored.
aloha OK. I'll deliver the message.
kid51 NotFound If you get a chance, can you glance at trac.parrot.org/parrot/ticket/1315 ? Thanks. 13:07
Hmm, it appears our 'aloha' bot doesn't yet know how to erase messages. 13:08
NotFound kid51: that ticket is so old that who knows what the comment refers to. We have no one, but several packfile apis. 13:16
13:17 Psyche^ joined 13:18 Patterner left, Psyche^ is now known as Patterner
NotFound kid51: surely the way the packfile functions, the packfile pmcs, the eval pmc and the interpreter interact must be clarified and improved, but references to years old comments hardly helps. I'll recommend closing that ticket and open a new one, a web page on track, or both. 13:20
tadzik NotFound: I'd probably wait til Rakudo gets into the new revision, as when I built it some of my codes tend to segfault with no reason 13:33
NotFound tadzik: no problem 13:34
dalek rrot: r48955 | fperrad++ | trunk/runtime/parrot/library (2 files):
[distutils] fix TT#1784
 13:38
13:44 ash__ joined 13:45 ash_ left, ash__ left, ash_ joined
kid51 NotFound: Thanks. Deleted comment; took and closed ticket. 13:46
13:47 fperrad left
dalek TT #1784 closed by fperrad++: distutils fails sdist and bdist if zlib is not installed 13:53
TT #1784: trac.parrot.org/parrot/ticket/1784
TT #1315 closed by jkeenan++: src/pmc/eval.pmc: create PF API, move it there
TT #1315: trac.parrot.org/parrot/ticket/1315
rrot: r48956 | jkeenan++ | trunk/src/pmc/eval.pmc:
Deleting ancient, now meaningless comment as per �trac.parrot.org/parrot/ticket/1315.
 13:55
14:06 patspam joined
dalek TT #1140 closed by jkeenan++: config/gen/platform/generic/env.c unsetenv() out of sync 14:10
TT #1140: trac.parrot.org/parrot/ticket/1140
Infinoid moritz: I propose adding an optional "branches": ["master, "ng", "omgspiffy"] field to the json, which defaults to ["master"]. The gitorious and googlecode plugins don't support a branch list, but github does. 14:13
14:28 jan joined
moritz Infinoid: +1 14:36
purl 1
dalek nxed: r632 | NotFound++ | trunk/winxedst1.winxed:
simplify base class emit in stage 1
Infinoid moritz: Do you have an example of a non-master branch on a github repo that's worth tracking? 14:39
(semi-longterm) 14:40
moritz Infinoid: not really
Infinoid Well, I'll leave it in, we can test it if/when we need it 14:41
It's just passing stuff through to githubparser, and I know that already works.
14:45 Paul_the_Greek joined
Paul_the_Greek Hey ho. 14:45
Can someone help me with a branch merge?
kid51 which branch? 14:46
purl which branch is the one I need to pull again?
kid51 purl, forget which branch
purl kid51: I forgot which branch
Paul_the_Greek I need to merge sleeker_boolean. 14:47
I don't understand the branching guide.
kid51 sleeker_boolean: Sounds like the name of a geek skateboarding team ;-)
Paul_the_Greek Good name for a rock group, too. 14:48
Do I commit my changes to the branch first?
kid51 Do you have a local checkout of the branch? 14:49
Paul_the_Greek Yes, and I've changed five files.
kid51 Have you committed those 5 files to the branch?
Paul_the_Greek No.
kid51 Do so now. 14:50
Paul_the_Greek Question first.
When I diff those files, I see differences other than the ones I made.
Why?
dalek nxed: r633 | NotFound++ | trunk/token.cpp:
allow dollar sign in identifiers in stage 0
 14:51
kid51 Are you using 'svn diff' to compare them with the existing versions in that branch?
Paul_the_Greek Yes, but never mind. I don't see differences other than the ones I made. Just spaced out.
I'll commit those files now. 14:52
14:52 dalek left 14:53 dalek joined
Paul_the_Greek Done. 14:53
Now the guide tells me to merge with the trunk. 14:54
like this: svn merge -r48775:HEAD svn.parrot.org/parrot/trunk/ 14:55
kid51 Well, let me tell you what I do. 14:56
Paul_the_Greek Okay.
kid51 Which others will carp as long-winded
But it WFM
In your branch: Call: svn up;svn st;svn log . --stop-on-copy
Make sure all files in your branch are up-to-date. Note the oldest revision number, i.e., that at which you originally branched from trunk. 14:57
Paul_the_Greek But svn update won't find anything, since it's comparing against the branch files, right? 14:58
Infinoid moritz: I'm not sure the perl6.org and modules.perl6.org feeds will work, because of their names. But it's ready for testing
kid51 Paul_the_Greek: As I said, this may be long-winded, but it's reliable for me
nopaste "infinoid" at 192.168.1.3 pasted "New #perl6 feeds dalek will track" (21 lines) at nopaste.snit.ch/23303
Paul_the_Greek Okay ...
It said "At revision 48958"; status lists nothing. 15:00
kid51 And: svn log . --stop-on-copy
Paul_the_Greek Lists the original branch revision and the commit revision.
kid51 Fine. What is the oldest branch revision listed there? 15:01
Paul_the_Greek r48775, the branch revision.
kid51 Oh, by the way: Have you run 'make' and 'make test' in the branch?
Paul_the_Greek Yes, but I will do it one more time.
kid51 If not, do so, and handle any test failures.
Paul_the_Greek Let me do that now. Will you be around for a few minutes? 15:02
kid51 Yes, but while you're running 'make test', I'll type the next steps.
dalek nxed: r634 | NotFound++ | trunk/winxedst0.cpp:
initial incomplete support for $load directive in stage 0
rrot: r48957 | Paul C. Anagnostopoulos++ | branches/sleeker_boolean (5 files):
New Boolean that does not inherit from Integer
rrot: r48958 | luben++ | branches/hash_inlined_func:
Remove mergerd branch
kid51 Once we know the branch PASS, switch to a checkout from trunk. If I've changed many files, I'll frequently do a fresh checkout, just to be sure I don't have files lying around in my sandbox for which I can't account. 15:03
In that checkout from trunk, I call: svn up;svn st
... to make sure my trunk is up-to-date 15:04
If, from observing #parrot, I sense that trunk is FAIL, I consider holding off the merge until the FAIL is cleared.
Assuming trunk is in good shape, I perform the merge:
NotFound kid51: up after a checkout? You really want to be sure ;)
kid51 NotFound: As I said above, others will carp that this is long-winded, but it WFM 15:05
Paul_the_Greek You do a fresh checkout for this, or is an update of an existing checkout ok?
kid51 (And we haven't even gotten to the point where the git fanbois carp about the time it takes subversion to do merges ;-) ) 15:06
NotFound kid51: if you are going to put that steps in the guide, put the svn up after "trunk is in good shape"
kid51 Paul_the_Greek: In this case, you probably don't need a fresh checkout because your branch touched few files and is short-lived
But if, say, you had been working on this branch for 3 months and had changed 20 files, .... well, you get the picture. 15:07
Paul_the_Greek Got it.
kid51 svn merge svn.parrot.org/parrot/trunk@[REVISION NUMBER AT BRANCH POINT] svn.parrot.org/parrot/branches/sle...olean@HEAD . 15:08
Paul_the_Greek Tests PASS.
kid51 (I have a Perl script that simplifies that, but that's the official command.)
Paul_the_Greek The revision number at the branch point is the original revision when I did the branch? Just checking. 15:09
kid51 Yes.
Paul_the_Greek Does it want an 'r' in front?
kid51 No: nnnnn
Paul_the_Greek Okay, here we go ... 15:10
kid51 You're comparing head in the branch to the branchpoint in trunk
The merge will pick up a lot of stuff that is not relevant to what you are doing. e.g., revisions in the svn properties of files that have changed.
Don't sweat it.
Paul_the_Greek Okay, four of the files were listed with U.
kid51 Good sign. 15:11
Paul_the_Greek The new file is listed with C and it says Tree conflicts: 1
kid51 Okay, then you'll have to edit the file to resolve conflicts. Once conflicts are resolved, you call 'svn resolved file_that_had_conflicts'
Rinse, repeat. 15:12
Paul_the_Greek How do I resolve the fact that it is a new file?
kid51 Hmm, I'll have to think about that.
Paul_the_Greek You would think the commit would have complained. 15:13
It said "Adding ..."
Oh, but that's just in the new branch.
kid51 Well, the commit *in branch* had nothing to complain about.
Paul_the_Greek Exactly. 15:14
15:15 dalek left, dalek joined
kid51 Paul_the_Greek: When you added new files *in branch*, did you update MANIFEST *in branch*? 15:16
Paul_the_Greek Yes.
It was one of the files flagged 'U'.
kid51 okay, I've been shadowing you here. When I attempt the 'svn merge' command, the only file in which I get a conflict is MANIFEST. 15:17
Paul_the_Greek Shadowing using which branch? 15:18
kid51 Shadowing: I did a checkout of that branch, then attempted to merge branch into trunk as per instructions I just gave you.
Paul_the_Greek Oh, after I did the commit so you had the new file? 15:19
Did you get this output: --- Merging differences between repository URLs into '.':
kid51 Yes. 15:20
Paul_the_Greek How was boolean.pir flagged? With an 'A'?
kid51 Yes
Paul_the_Greek Interesting.
And I got no conflicts with MANIFEST, which actually surprised me. 15:21
kid51 Is the checkout from trunk into which you are merging a directory in which you were working on this project already?
Or is it a fresh checkout?
Paul_the_Greek Sorry, don't understand.
This is the checkout that I did right after I made the branch. 15:22
kid51 I'm trying to understand how you could get a 'C' on boolean.pir
Perhaps I was unclear.
Before executing 'svn merge', cd to a sandbox from trunk. 15:23
In that sandbox, call 'svn up;svn st' to get any recent changes to trunk.
Then execute the 'svn merge' command there.
Paul_the_Greek Oh, sorry, I thought you said that was unnecessary in this simple case. 15:24
Few files, not much time.
kid51 No, the thing that was probably unnecessary was to do an *additional, fresh* checkout from trunk -- as opposed to simply re-using your regular sandbox from trunk. 15:25
When I'm working on a project, I create a new branch in the repository, checkout the branch, hack and test in the sandbox for branch, commit to branch, then merge from branch to trunk. 15:26
Paul_the_Greek That's what I did.
You do that last step in a separate checkout, though?
kid51 Well, that last step can be conducted in any up-to-date, clean checkout from trunk. 15:27
Paul_the_Greek So then I'm not sure what this will do, but I've updated a separate checkout.
svn status is clear. 15:28
kid51 And a fresh checkout from trunk is, by definition, up-to-date and clean.
Paul_the_Greek Okay to do the merge again?
kid51 k. Execute svn merge command
Paul_the_Greek Okay, now a conflict in MANIFEST. Which option is the best to pick?
kid51 You have a couple: 15:29
1. Manual editing.
2. perl tools/dev/mk_manifest_and_skip.pl
Paul_the_Greek: I have to stop you at this point. In sleeker_boolean branch, 'make' is failing for me. 15:30
Paul_the_Greek I don't know why there would be a conflict with a new file.
kid51 And it's failing post-merge in trunk as well.
Paul_the_Greek Oh, what's happening?
Wait, first, what should I do with this merge?
Infinoid msg bacek github.com/bacek/pir doesn't seem to exist any more. Should I remove it from trac.parrot.org/parrot/wiki/Languages, or is there an updated link? (dalek scrapes that page looking for links, and gets fetch errors trying to follow that feed.) 15:31
purl Message for bacek stored.
aloha OK. I'll deliver the message.
kid51 Abandon ship. You've got to get trunk to build first.
Paul_the_Greek Just Ctrl/C out of it?
Have we actually changed the trunk yet?
kid51 Pause. 15:32
Paul_the_Greek Crap, I Ctrl/C'ed out of it.
Okay, calming down. 15:33
nopaste "kid51" at 192.168.1.3 pasted "'make' fails in sleeker_boolean branch" (413 lines) at nopaste.snit.ch/23304 15:34
kid51 So, return to the branch, correct build failure, run make test, etc.
Paul_the_Greek Interesting, I don't get that error. 15:35
NotFound What's FALSE? 15:36
purl it has been said that FALSE is "", undef, 0, "0". true is NOT these things. or at language.perl.com/ppt/src/false/ or '', 0.0, ( ) or unknown
kid51 This was run in a fresh checkout of sleeker_boolean branch at r48958 on linux/i386
Paul_the_Greek Is there a universal #define for FALSE, or should I just use 0?
NotFound Paul_the_Greek: there ar too many universal defines, usually that is the problem.
Paul_the_Greek Er, but TRUE and FALSE would be right up there with NULL as fundamental. 15:37
I'll use 0.
whiteknight Parrot_hash_put is called 2475 times at Parrot startup, without executing anything 15:38
kid51 Parethetical question: Do we have to run 'make headerizer' on this file?
NotFound Paul_the_Greek: yes, but you need to go back in time to pre-ansi C to fix that.
Paul_the_Greek It doesn't need headerizer, but I should run it anyway. 15:39
Can I run it on just one file?
NotFound In a pmc file? You shouldn't 15:40
The build process takes care.
Paul_the_Greek Okay.
kid51: Have these merges changed the branch yet? 15:41
kid51 NotFound: Thanks, I was unclear on that. 15:42
Paul_the_Greek Sorry, changed the trunk yet?
Did make. Running tests ...
kid51 No, assuming you have not said 'svn commit' in trunk.
Paul_the_Greek Ah, that makes sense. 15:43
kid51 In your checkout from trunk, you can say 'svn revert -R .' to revert the files you attemmpted to merge in.
Paul_the_Greek I wonder why I had a FALSE defined?
NotFound Paul_the_Greek: ar you building on windows? 15:44
Paul_the_Greek Yes.
NotFound Paul_the_Greek: some windows header should be included in some parrot header
dalek nxed: r635 | NotFound++ | trunk/winxedst0.cpp:
qualified base classes in stage 0
Paul_the_Greek Ah, so we get FALSE on Windows.
Should we undefine TRUE and FALSE at the bottom of some fundamental include file? 15:45
NotFound Paul_the_Greek: That's why I rant about too many universal defines
Paul_the_Greek I agree there are too many. I just like TRUE and FALSE. :D
NotFound Paul_the_Greek: we just shouldn't use too generic words.
Paul_the_Greek Languages without first-class Booleans make me crazy. 15:46
Okay, PASS.
moritz like, Perl 5? :-)
NotFound Paul_the_Greek: as I said, you need to go back in time an force the first C standard commitee to accept that ;) 15:47
Paul_the_Greek I know. But the mistake keeps being made in new languages. Ah well, gotta have something to grouse about. 15:49
kid51: Tests PASS, committed boolean.pmc
NotFound Winxed hasn't first class boolean X-)
But it has predefs true and false, at least.
Paul_the_Greek Which of these many conflict options from svn merge are the easiest to use to resolve the conflict? 15:51
dalek rrot: r48959 | Paul C. Anagnostopoulos++ | branches/sleeker_boolean/src/pmc/boolean.pmc:
Eliminate use of FALSE macro
 15:52
rrot: r48960 | jkeenan++ | branches/sleeker_boolean/examples/benchmarks/boolean.pir:
Set SVN properties on examples file.
Infinoid seen sorear 15:53
purl sorear was last seen on #parrot 19 hours, 2 minutes and 21 seconds ago, saying: micro or macro?
aloha sorear was last seen in #perl6 5 hours 2 mins ago saying "thanks".
Infinoid sorear: The more I think about your idea of push-based karma updates, the more I like it. Dalek is watching more than 50 github feeds now, and the polling interval is more than 5 minutes for each of those. It's a lot of latency and a lot of overhead 15:54
kid51 Paul_the_Greek: Am fixing some codingstd errors in your branch, so do 'svn up'
Paul_the_Greek Okay.
Do it now? 15:55
kid51 Yes.
Paul_the_Greek Update boolean.pmc, skipped boolean.pir 15:56
kid51 And, in branch, after running 'make test' and getting a PASS, you should probably do 'make codetest' as well ...
.... unless I do it first :-)
Paul_the_Greek Now I'm confused about where we are.
kid51 Alright 'make' succeeded in branch on linux/i386
We're still fixing the branch. 15:57
The more problems we correct in branch, the fewer problems we'll have with the merge.
Paul_the_Greek Did you commit the changes to boolean.pmc?
kid51 Yes: trailing whitespace. Do svn up and make test 15:58
You'll need to add a copyright notice to src/pmc/boolean.pmc
Paul_the_Greek There is one.
kid51 Hmm, the codingstd test isn't detecting it. 15:59
Paul_the_Greek I did an update, but I don't see any changes.
Oh, I bet the /* can't be on the same line. 16:00
kid51 I fixed it. 16:01
Not exactly clear what was wrong, so I just copied from another file.
svn up
Some additional codingstd problems: 16:02
run: prove -v t/codingstd/c_code_coda.t t/codingstd/c_parens.t 16:03
Paul_the_Greek Kid51, I'm getting confused. 16:04
kid51 about?
Paul_the_Greek Where do I do that svn update?
kid51 At any time, but particularly before crucial things like 'svn commit' and 'svn merge'. 16:05
In this case, we have a situation we don't often encounter ...
Paul_the_Greek But in my branch sandbox or in the clean one I'm using for the merge?
kid51 ... two people working in same branch simultaneously.
In branch sandbox.
Paul_the_Greek Okay.
kid51 We need to get the branch in tiptop shape before reattempting the merge.
I should have advised you to check for codingstd failures before attempting the merge earlier. 16:06
You can get a coda to place at the end of boolean.pmc out of any other .pmc file in the same directory.
Paul_the_Greek Well now, look at that complaint. Of course there should be a space between a function name and the open paren in the function definition.
It's the calls that shouldn't have a space. 16:07
kid51 Paul_the_Greek: For better or worse, we have coding standards that a codingstd test enforces.
If you want to argue about coding standards, file a TT and prepare for theological war ;-)
Paul_the_Greek Actually, I asked about this particular thing a few weeks ago and someone said it was good.
He didn't know that it was part of the standard, I guess. 16:08
kid51 As cage cleaner, I'm just the cop on the beat here. Don't make the law, just enforce it.
Paul_the_Greek So shall I make these changes?
kid51 Yes.
dalek rrot: r48961 | jkeenan++ | branches/sleeker_boolean/src/pmc/boolean.pmc:
[codingstd] No trailing whitespace.
 16:09
rrot: r48962 | jkeenan++ | branches/sleeker_boolean/src/pmc/boolean.pmc:
[codingstd] Fix copyright statement.
kid51 Incidentally, in the branch 'make' is now succeeding for me, as is 'make test' ...
... which suggests that once we fix codingstd, we can reattempt merge.
Paul_the_Greek No coda.
Can you see what's wrong with the coda? 16:10
kid51 (If you've fixed the c_parens failure, please svn commit the file.)
Paul_the_Greek It's complaining about no coda. Ignore? 16:11
kid51 Again, it's just a formatting problem (which exposes a limitation in the test) 16:12
Paul_the_Greek Okay.
kid51 Place "Local variables" on the 2nd line of the C-comment, not the first.
Paul_the_Greek Committed.
Oops, doing a make just to be sure. 16:13
Make okay.
kid51 Did you fix the coda problem?
Paul_the_Greek No, doing that now... 16:14
kid51 Compare, e.g., src/pmc/eval.pmc 16:15
Paul_the_Greek Okay, coda fixed and committed. Passes coding standards.
Try merge again? 16:17
kid51 Yes. cd to your sandbox for trunk. If you still have the remnants of the earlier attempt at a merge, perform "svn revert -R ." 16:18
Then: svn up; svn st --- to make sure you catch any commits to trunk performed since you were last there.
Paul_the_Greek All set.
dalek nxed: r636 | NotFound++ | trunk/winxedst0.cpp:
rename ClassBase to ClassSpecifier and rearrange its usage in stage 0 compiler
 16:19
Paul_the_Greek Now which conflict option should I pick for MANIFEST?
kid51 The simplest is 'P' for postpone ... because a run of 'perl tools/dev/mk_manifest_and_skip.pl' will take care of that easily. 16:20
Paul_the_Greek I got U and A on the other files.
kid51 So: perl tools/dev/mk_manifest_and_skip.pl
svn resolved MANIFEST
svn st
then, ideally: perl Configure.pl; make test 16:21
then: svn commit -m "Merging sleeker_boolean branch into trunk as per TT #xxx"
Paul_the_Greek svn status show lots of modified files.
kid51 But most of them are changes in SVN properties (2nd column); don't worry about them. 16:22
Paul_the_Greek Oh cripes, how cryptic.
kid51 As long as there are no 'C' and all the 'A' and 'U' are those you have brought in from branch, you're fine.
16:22 patspam left
kid51 Have you worked with other version control systems? 16:23
Paul_the_Greek Yup, first column of status as expected. 16:24
I've only worked with systems that checked out and locked files.
kid51 Ah
Paul_the_Greek This is still not obvious to me.
kid51 And if *this* isn't obvious, wait till we switch from Subversion to git in a few weeks! ;-) 16:25
Paul_the_Greek I figure I'll spend more time trying to grok git. I read the git book and played with it while on vacation.
make test ...
dalek rrot: r48963 | Paul C. Anagnostopoulos++ | branches/sleeker_boolean/src/pmc/boolean.pmc:
Fixed coding standards issues with parens
 16:26
rrot: r48964 | Paul C. Anagnostopoulos++ | branches/sleeker_boolean/src/pmc/boolean.pmc:
Fixed formatting of coda
kid51 ... to continue list of commands above ...
Once you've done 'svn commit' in trunk, you call:
svn rm svn.parrot.org/parrot/branches/sleeker_boolean -m "Branch has been merged into trunk and is no longer needed at HEAD."
And at that point, toss out all your local sandboxes that pointed to that branch. 16:27
Paul_the_Greek On that commit above, I presume I list the five changed files, as usual?
kid51 That way, you won't be tempted to make further corrections in a branch that's already been merged.
No, you don't need that in the log message.
Paul_the_Greek No, on the command. Or does it use just the changed files by default? 16:28
kid51 The way I've typed it above will commit all changes in your sandbox ... which is what you want at this moment.
Paul_the_Greek Good. 16:29
Test failure in archive_zip.t 16:30
I get that's a Windows thing.
s/get/bet/
kid51 Yes, I believe there was a Trac ticket about something like that in the past day.
Paul_the_Greek Okay, then I'm good.
kid51 So, you're ready for the 'svn commit' and 'svn rm' commands described above. 16:31
Paul_the_Greek Let me try the ciommit.
It sending all the files.
kid51 The transmission message is a bit cryptic. 16:32
It prints out all files including those where differences since branch point have occurred.
Paul_the_Greek Oh. Sweet mother.
kid51 But the number of dots in the *final* statement should match the number of 'A' or 'U' files. 16:33
Patience, my son.
16:33 tcurtis joined
Paul_the_Greek What? 16:33
I got four dots after "Transmitting file data"
kid51 At this point, the git fanbois interrupt to tell us how much quicker git is at merges than Subversioni.
4 dots sounds correct
You changed/updated 4 files.
Paul_the_Greek svn status is clear.
svn remove worked. 16:34
kid51 You're done. Give that young man a gold star.
Paul_the_Greek Forget my gold star. 16:35
purl Paul_the_Greek, I didn't have anything matching my gold star
Paul_the_Greek kid51++ 16:36
kid51 purl is so fashion-conscious
Paul_the_Greek Thanks so much. I never would have figured that out.
kid51 np.
Paul_the_Greek Get rid of my branch sandbox ...
kid51 Believe me, when we switch to git, *everybody* will have an ordeal like that
It's a question of getting a workflow in your fingertips.
At this time 3 years ago, very few Parrot devs were using branches at all. 16:37
s/very/relatively/
So there were many more direct commits to trunk
And, hence, much more breakage to trunk :-( 16:38
The more you test in your branch, the better.
Paul_the_Greek Yes, that makes sense. 16:39
But why wouldn't it add the new file from the branch sandbox?
kid51 Also, once you've had your knuckles wrapped by 'make codetest' a few times, you'll get a sense of what the codingstds require and will avoid those errors before they get tested.
16:39 patspam joined
kid51 Paul_the_Greek: I'm not sure. I'd have to peer into your sandbox for that. 16:39
Paul_the_Greek Well, next time.
purl Next time, Gadget. NEXT TIME!
kid51 Which 'new file' had the problem?
Paul_the_Greek examples/benchmarks/boolean.pir 16:42
dalek rrot: r48965 | Paul C. Anagnostopoulos++ | trunk (98 files):
Merging sleeker_boolean branch into trunk as per TT #1770
 16:43
rrot: r48966 | Paul C. Anagnostopoulos++ | branches/sleeker_boolean:
Branch has been merged into trunk and is no longer needed at HEAD.
purl i already had it that way, dalek.
kid51 Hunch: You didn't have Subversion properties set on that file. Your non-propertied version was sitting in a directory into which you attempted a merge.
16:43 patspam left, patspam joined
kid51 At one point above, I did a commit which added those properties. 16:43
dalek nxed: r637 | NotFound++ | trunk/winxedst0.cpp:
use ClassSpecifier in new statement in stage 0 compiler
 16:44
kid51 You might want to post in TT #1770 that you have performed the merge. If no one reports any problems in a couple of days, ticket is closable. 16:45
Ah, I see you've already posted. Good. 16:46
afk
Paul_the_Greek So it's okay that the changeset lists all those files that were "Sending"? 16:48
dalek TT #1770 closed by Paul_the_Greek++: Boolean PMC without Integer 17:00
TT #1770: trac.parrot.org/parrot/ticket/1770
TT #1770 reopened by Paul_the_Greek++: Boolean PMC without Integer
TT #1770: trac.parrot.org/parrot/ticket/1770
kudo: d1eb87c | pmichaud++ | src/core/ (2 files):
Rewrite C<.reverse> to be more efficient, move from Any-list to List.

from 44s to 4s when run on Rakudo's README.		 17:04
NotFound Oh, my, you reverted some changes with that merge
NotFound Duplicated VTABLE function: get_integer 17:05
Duplicated VTABLE function: shift_pmc
I already fixed that!
Uh, no, it should have been reverted before, don't see it in the changeset. 17:08
tcurtis Infinoid: I think bacek/pir is now at github.com/parrot/pir 17:11
NotFound Oh, shit! 17:14
17:29 dalek left
Infinoid tcurtis: Thanks, updated. 17:30
17:30 dalek joined
Infinoid msg bacek Never mind, tcurtis++ pointed out the new URL and I've updated the Languages page. 17:30
purl Message for bacek stored.
aloha OK. I'll deliver the message.
Paul_the_Greek NotFound: Did I screw it up with that merge?
NotFound Paul_the_Greek: no, it was a confusion, sorry.
Paul_the_Greek Whew.
I was scared by the huge list of files included in the commit. 17:31
Kid51 said it has something to do with file attributes?
kid51 Paul_the_Greek: Yes, this is something that appeared only in recent versions of Subversion.
I'm sure it's helpful to somebody, but the first time any average dev ($self included) sees it, it's puzzling. 17:32
Paul_the_Greek It's really scary because it shows up in the file list on the revision page.
dalek tracwiki: v142 | Infinoid++ | Languages 17:39
tracwiki: Update link to PIRATE, tcurtis++
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
NotFound kid51: looks you and me heard different discussions 17:43
kid51: Why do you want to keep wrong, unuseful and unused code in the repo? 17:45
nopaste "kid51" at 192.168.1.3 pasted "Conclusion of 9/11 discussion re duplicated vtable functions" (5 lines) at nopaste.snit.ch/23306 17:48
kid51 NotFound: We discussed that thoroughly yesterday. The code has to be put thru a deprecation cycle. 17:49
NotFound kid51: that was for the dying, but not for the erroneous vtables detected!
dalek rrot: r48967 | jkeenan++ | trunk/runtime/parrot/library (2 files):
[codingstd] No trailing whitespace.
 17:50
kid51 That was not my understanding. My understanding was that pmichaud had reported that a HLL had inadvertently depended on the interface as it was, such that our correcting the code without putting it thru a deprecation cycle was contrary to our deprecation policy. 17:51
I created and took the ticket and implemented that straight away.
NotFound kid51: how on earth he can depend on a vtable function that doesn't get compiled? 17:52
kid51 NotFound: That's not for me to say.
The correctness of the code was not the only issue. Whether a revision to the code needed a deprecation cycle was also at issue. 17:53
pmichaud and you were getting pretty riled up about the issue.
NotFound This is plain insanity.
kid51 I intervened and propsed a solution. 17:54
Which, as far as I could tell, both of you accepted.
I then implemented the solution.
NotFound I'll give up. I won't touch parrot for a week... or more.
kid51 NotFound: If we have two of our expert developers at loggerheads with each other, we have to take an approach that resolves the dispute in accordance with our customary procedures. 17:56
And one of those procedures is the deprecation policy.
True, it's not going to make you -- or pmichaud -- or me -- happy all the time.
plobsing I was under the impression that the vtables themselves were not the issue of deprecation, but the error checking introduced into pmc2c. 17:57
kid51 It's more of a "reduce the total quantity of tears shed" approach.
plobsing: Well, if pmichaud is around and if he agrees with your impression, I'm willing to stand corrected. 17:58
jnthn As far as I understand, the detection is totally fine (in fact, good). However, pmichaud++ felt that dying rather than just warning required a deprecation cycle.
(I helped pm to track down and remove the duplicated v-table method we had somewhere in Rakudo. It's good to detect them - our duplication was a straightforward mistake.) 17:59
kid51 jnthn Well, can you (and plobsing) post your impressions in trac.parrot.org/parrot/ticket/1785 ? 18:00
that way, when pmichaud comes on line, he can give his impression as well and we can move toward a resolution.
jnthn oh wth
The ticket is a complete mis-understanding.
The duplicates serve no useful purpose and should be tossed.
NotFound Sigh... Again, the duplicated vtable functions don't get compiled, they aren't put in the generated C file. Look at tapir2.ro.vutbr.cz/cover/cover-resu...b-pmc.html
jnthn There's no way we could have depended on them.
It was just a simple "die vs warn" issue. 18:01
NotFound++ is right here.
NotFound Lines 115-117
kid51 jnthn Please post all this in the ticket.
jnthn *sigh*
Posted 18:03
NotFound kid51: I appreciate your intention, but that things cause me headeaches.
kid51 jnthn pmichaud saw what I posted/committed yesterday. Perhaps you're correct. But since what I did was in response to discussion among NotFound, pmichaud and myself, I think we should hear from pmichaud before proceeding further.
18:03 ingy left
jnthn kid51: That's fine, but I'm correct. :-) 18:03
</arrogance> :-) 18:04
18:04 ingy joined
kid51 NotFound: Feel free to post the headaches in the ticket. (But, perhaps, cool the description a bit -- headaches -> problems -- so that we don't get overheated about this.) 18:05
jnthn: No way am I going to try to referee a dispute between you and pmichaud ;-) 18:06
If the collective judgment is that nothing needs to be deprecated, I will go with that.
If the collective judgment is that the wrong thing has been deprecated, I will go with that. 18:07
jnthn bbiab, dinner
NotFound I hope there is no dispute, just a misunderstanding. 18:08
kid51 I want to reduce the total volume of tear shed over this issue, and as quickly/effectively as possible. But I can't guarantee that no tears will be shed.
s/volume of tear/volume of tears/
plobsing trac-- # Warning: Sorry, can not save your changes. This ticket has been modified by someone else since you started 18:09
jnthn NotFound: I think it's just a mid-understanding between a bunch of people all trying to do the right thing. It happens. 18:10
18:10 bluescreen joined
jnthn really stops hacking to feed himself 18:11
kid51 plobsing Yes, I've encountered that Trac problem previously. Happens when there's a hot issue and >1 person is commenting on it simultaneously ;-)
NotFound jnthn: The problem is that solving the disputes by hand fighting will be expensive because of the long distance flying implied ;) 18:12
plobsing purl: rsp? 18:13
purl hmmm... rsp is the most trivial perl code inverter possible. or remote strangulation protocol.
18:15 kid51 is now known as kid51_at_dinner
dukeleto 'ello 18:16
dukeleto has internets for 5 mins
dukeleto just updated the parrot github mirror and pushed some branches 18:23
18:41 davidfetter joined 18:54 kid51_at_dinner left
Infinoid Does anyone know if there's a migration schedule for parrot -> github? I want to make sure dalek follows along. 18:59
cotto Infinoid, there's a relative schedule on GitMigration, but it's not certain when all the prerequisites will be done. 19:44
GeJ Bonjour everyone. 19:47
20:01 patspam left 20:02 patspam joined
dalek TT #1779 closed by coke++: test failures on OSX 20:05
TT #1779: trac.parrot.org/parrot/ticket/1779
kudo: 9993bb2 | moritz++ | docs/ChangeLog:
[docs] update ChangeLog
 20:07
cotto closed tickets? 20:12
20:13 tadzik left
dalek nxed: r638 | NotFound++ | trunk/winxedst0.cpp:
look for classes in namespace scopes in stage 0
 20:37
20:49 mj41 left 20:50 mj41 joined
luben there seems to be issues with running rakudo after boolean branch merge 20:59
nopaste "luben" at 192.168.1.3 pasted "Rakudo after boolean branch merge" (6 lines) at nopaste.snit.ch/23308 21:00
NotFound "cloned Boolean can change value" Can change rom 1 to 1? Amazing ;) 21:11
Paul_the_Greek Oh good, I figured there would be some. 21:15
cotto Paul_the_Greek, now you know why we like to ask "Did you test it with Rakudo?". 21:17
NotFound Paul_the_Greek: the clone test you changed was right, the value of the clone is 1, doesn't make sense change it to 1
Paul_the_Greek cotto: So I should never listen to anyone who says "Just commit it and let's see what happens"? :D
NotFound Paul_the_Greek: you did it, and now you see what happens ;) 21:18
cotto Yes. It's often wise to ignore me.
Paul_the_Greek I haven't the foggiest notion how to test it with Rakudo.
cotto: I think chromatic said that, too. 21:19
cotto Just Configure.pl --prefix=/something/something and build Rakudo normally.
Paul_the_Greek That nopaste is gibberish to me. Can someone interpret it?
Build Rakudo normally, you say.
cotto It'll take some digging to figure out exactly what broke. 21:20
NotFound Paul_the_Greek: don't worry too much, rakudo don't target trunk except for testing purposes. It can survive a few commits failing
cotto Well, make sure to nuke parrot_install/ in Rakudo's build dir/
s,/,.,
Paul_the_Greek Rakudo's build directory?
None of this means anything to me.
tcurtis Was Rakudo working on trunk before the merge?
cotto Have you build Rakudo? 21:21
NotFound I find easier to build plumage and then ./plumage install rakudo
cotto Ah. I'm not sure how plumage does its thing.
NotFound It does it well, that's enough for me ;) 21:22
cotto It's shaping up to be a nice tool, especially once it doesn't have to be run from its source dir.
21:23 mj41_ joined
Paul_the_Greek No, I haven't built Rakudo. I should learn how to do that. 21:23
cotto It's pretty simple on Linux. 21:24
Paul_the_Greek How bad is it on Windows?
cotto no idea
NotFound No idea 21:25
Paul_the_Greek Well then, it should be fun.
cotto istr that jnthn does it though
Paul_the_Greek Yes, he probably does.
21:25 mj41 left
Paul_the_Greek Having fun hacking on the debugger. 21:25
jnthn Builds fine for me. :-)
(using MS VC++ compiler) 21:26
cotto Paul_the_Greek, you can aks him then.
luben tcurtis, it was working
Paul_the_Greek Is it a build problem, or a test failure?
NotFound Paul_the_Greek: pleas fix the clone test 21:27
luben Paul_the_Greek, runtime failure
Paul_the_Greek NotFound: Sorry, I need more information. 21:28
tcurtis luben: okay. I was just wondering, since I recalled lots of troubles with other recent revisions.
21:28 mj41 joined
NotFound Paul_the_Greek: it was changing the cloned value from 0 to 1. With your commit it changes 1 to 1, which is not exactly a good test. 21:29
I mean, it was from 1 to 0
Now is from 1 to 1
21:30 mj41_ left
NotFound Is a bit confusing... very appropiate X-) 21:30
Paul_the_Greek I did that for a reason, but now it eludes me. Seems stoopid. 21:31
I'll change the test, run all the tests, and then commit it. 21:33
NotFound Paul_the_Greek: in the PMC file, there is a confusing mix of open and closing comments between lines 153-161. I'm not sure about what pmc2c does with it, 21:34
Paul_the_Greek It ignores it correctly. 21:35
I kept that code in case we have trouble with 'neg boolean'.
NotFound Paul_the_Greek: our coding standards is agianst that, if you need old code it is on the repository. And in PMC files is specillay prone to confusion because o pmc preprocessing 21:36
Paul_the_Greek Okay, I'll remove it and commit the file along with the test. 21:37
NotFound Paul_the_Greek++
Paul_the_Greek make and test first ... 21:38
21:44 PacoLinux left
Paul_the_Greek NotFound: All done. 21:44
dalek rrot: r48968 | Paul C. Anagnostopoulos++ | trunk (2 files):
Correct boolean test, remove obsolete code
 21:45
NotFound Paul_the_Greek: thanks 21:47
Paul_the_Greek My pleasure.
A little dinner, methinks.
luben Paul_the_Greek, the last fix does not fixes rakudo, it gives the same errors 21:59
NotFound luben: it was just fixing a test and a style correction. 22:03
luben ok
dalek nxed: r639 | NotFound++ | trunk/winxedst0.cpp:
complete implementation of $load directive in stage 0
 22:06
pmichaud 17:57 <plobsing> I was under the impression that the vtables themselves were not the issue of deprecation, but the error checking introduced into pmc2c. 22:15
plobsing is absolutely correct.
It's not the duplicated vtables that are an issue, it's the fact that pmc2c stops the compilation process when the situatione xists.
*situation exists. 22:17
whiteknight pmc2c--
pmichaud I never claimed we need to keep the duplicated vtable entries.
NotFound pmichaud: just a misunderstanding, then. Good to know. 22:22
luben wow, our fixed_size allocator is horrible... 2-3 time slower than system malloc 22:24
pmichaud I've updated the ticket with my comments. 22:26
22:40 patspam left
dalek nxed: r640 | NotFound++ | trunk/winxed (2 files):
refactor driver and installed driver to use recent features
 22:50
nxed: r641 | NotFound++ | trunk/pir/winxed_ (2 files):
update installable pir files
luben ops, my mistake, its slower, but not significantly slower 22:58
23:08 nwellnhof joined
dalek nxed: r642 | NotFound++ | trunk/winxedst0.cpp:
keyed base classes in stage 0
 23:15
nxed: r643 | NotFound++ | trunk/t/harness:
use recent features in harness
 23:20

Powered by Cro and the Raku® Programming Language.

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

Thank you!