Camelia, the Perl 6 bug

IRC log for #parrot, 2012-05-09

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:03 dalek parrot: 510b6c0 | Whiteknight++ | compilers/imcc/ (7 files):
00:03 dalek parrot: Fix all instances of global/static variables in IMCC which had been marked by plobsing++.
00:03 dalek parrot:
00:03 dalek parrot: In theory, this should make IMCC reentrant and thread-safe (pursuant to #565). Without a real threaded stress-test we won't know for certain.
00:03 dalek parrot: review: https://github.com/parrot/parrot/commit/510b6c07e1
00:08 dalek parrot: 465b364 | Whiteknight++ | src/pmc/scheduler.pmc:
00:08 dalek parrot: Don't clear the scheduler field from the interp when the Scheduler PMC is destroyed.
00:08 dalek parrot:
00:08 dalek parrot: First off, if the Scheduler is destroyed, that means the interp is probably not alive to have marked it. Second, since Scheduler isn't a singleton, it's entirely plausible (though not currently recommended) that the Scheduler might not be attached an an interp.
00:08 dalek parrot: review: https://github.com/parrot/parrot/commit/465b364fd7
00:09 aloha (parrot/parrot) Issues closed : 565 (pir compiler must be reentrant) by coke : https://github.com/parrot/parrot/issues/565
00:15 aloha (parrot/parrot) Issues closed : 510 (Order-of-destruction error with Scheduler PMC) by Whiteknight : https://github.com/parrot/parrot/issues/510, 313 (Bug in subroutine calling conventions) by parrot : https://github.com/parrot/parrot/issues/313
00:17 whiteknight msg cotto if you have a minute, can you take a peek at #514? I hope bacek has an answer, but if you have any input worth recording, it would be appreciated
00:17 aloha OK. I'll deliver the message.
00:20 aloha (parrot/parrot) Issues closed : 262 (Register allocator and lexicals bug) by jnthn : https://github.com/parrot/parrot/issues/262
00:22 cotto ~~
00:22 cotto whiteknight, 514 is fixed
00:22 cotto nm.  old trac issue
00:25 aloha (parrot/parrot) Issues closed : 338 (Build error on windows 2008 server) by parrot : https://github.com/parrot/parrot/issues/338
00:25 whiteknight cotto: I'm thinking that we have an is_writable flag in packfiles. You can't write if that flag isn't set, and you can't execute if it is.
00:25 whiteknight since our system does treat packfiles as more-or-less immutable after they've been loaded (and threading depends on that), I think that's a decent solution
00:28 cotto how deeply is the assumption of packfile immutability built into parrot?
00:28 cotto having a flag feels like a workaround
00:29 cotto not a bad short-term solution though
00:35 aloha (parrot/parrot) Issues closed : 244 (Investigate using an external hash library) by coke : https://github.com/parrot/parrot/issues/244
00:37 bacek_at_work ~~
00:37 bacek_at_work we should rebuild PackFIle header for repack
00:38 bacek_at_work (afair)
00:40 aloha (parrot/parrot) Issues closed : 251 (Pmc2c should generate mark and destroy VTABLES) by Whiteknight : https://github.com/parrot/parrot/issues/251
00:41 cotto bacek_at_work, what's the benefit from doing that rather than just enforcing "don
00:41 cotto 'd do that"
00:41 cotto ?
00:41 bacek_at_work cotto, no benefits at all.
00:42 bacek_at_work blocking of repacking PackFiles is fine with me.
00:42 cotto bacek_at_work, great.  I was hoping I wasn't missing anything obvious.
00:42 bacek_at_work Making them properly platform-independent is much more beneficial from my POV
00:43 cotto agreed
00:44 cotto whiteknight, there you go.
00:45 aloha (parrot/parrot) Issues closed : 646 (pluggable everything) by allisonrandal : https://github.com/parrot/parrot/issues/646
00:45 whiteknight cotto: immutability of bytecode and entries in the constants table, after the packfile has been created, are deeply ingrained
00:45 cotto whiteknight, I figured it'd be like that.
00:45 whiteknight cotto: But while the packfile is being constructed, we need to call multiple routines to construct it, so for a long time it needs to be writable, then it needs to not be
00:46 whiteknight because if we try writing to it after it's been loaded, blamo
00:46 cotto that's sensible
00:47 bacek_at_work we can apply "StringBuilder" approach within "Immutable Strings world".
00:48 bacek_at_work e.g. provide PackfileBuilder which will be used for building Packfiles
00:48 bacek_at_work but Packfiles by them self will be immutable
00:48 bacek_at_work after creation/loading
00:55 whiteknight bacek_at_work: There's a difference between the PackFile* struct and the PackFileView immutable PMC wrapper
00:55 whiteknight is that how we want to manage this, by relying on different wrappers?
00:55 whiteknight (that's not a bad idea, I'm just making sure I understand)
00:55 bacek_at_work whiteknight, I don't know anything about PackFileView...
00:56 whiteknight PackfileView is just a read-only wrapper around PackFile*
00:56 bacek_at_work good enough
00:56 bacek_at_work Than we just have to disable of fetching Packfile* out of it.
00:57 bacek_at_work (somehow)
00:57 whiteknight All the Packfile API routines take a PackFile*. Comes out of PackfileView with get_pointer
00:58 dalek parrot/rm_pasm: 589ca19 | bacek++ | t/pmc/packfileconstanttable.t:
00:58 dalek parrot/rm_pasm: Rework test to not rely on constants order.
00:58 dalek parrot/rm_pasm: review: https://github.com/parrot/parrot/commit/589ca195d6
00:58 whiteknight unwrapping repeatedly can lead to bad performance (every Sub call can be a switch)
00:59 alvis whiteknight++ # Wow! Awesome! (Sorry, I had to go to my daughter's recital)
00:59 whiteknight alvis: Sorry! I didn't realize you had called dibs
01:00 alvis whiteknight: Nbd. I'm just glad it's done and passes! :)
01:07 whiteknight alvis: While you're here, can you take a look at ticket #318? I suspect you would be the best person to look at that now
01:15 aloha (parrot/parrot) Issues closed : 498 (Coredumping Parrot with only 2 lines of PASM for fun and profit) by leto : https://github.com/parrot/parrot/issues/498, 374 (Whitespace before/after the dot in a methodcall is disallowed) by Whiteknight : https://github.com/parrot/parrot/issues/374, 297 (Unify
01:15 aloha ..VTABLE_invoke semantics between PMC types) by Whiteknight : https://github.com/parrot/parrot/issues/297
01:19 bacek_at_work meh... I pushed to wrong branch...
01:20 aloha (parrot/parrot) Issues closed : 330 (new_s and new_s_i opcodes) by Whiteknight : https://github.com/parrot/parrot/issues/330
01:26 whiteknight cotto: Also, can you look at #666? I suspect you'll be able to flip the switch pretty easily
01:28 cotto whiteknight, you mean implement it or close it?
01:28 cotto didn't we have almost 500 open issues earlier today?
01:39 whiteknight cotto: yeah, I'm inclined to just close it, considering the ratio of work/payoff
01:41 whiteknight cotto: yeah, we had 499 when I sat down at the keyboard this evening
01:41 whiteknight we're at 483 now
01:45 alvis whiteknight: (I was afk). I've taken assignment of #318 (though Coke may already have given the solution), and I'll get it done, tonight or tomorrow. I do have a question: Is there a title for the book other than draft?
01:46 whiteknight alvis: http://www.amazon.com/Parrot-Developers-G​uide-Allison-Randal/dp/0977920127/ref=sr_​1_2?ie=UTF8&qid=1336527953&sr=8-2
01:46 alvis whiteknight: Ah, ok. thanks.
01:46 whiteknight that's how it got published. I suspect that's the title we'd give it
01:47 alvis Sounds good. (I finally bought that book, btw :) Just hasn't arrived yet.
01:48 bacek_at_work whiteknight, you broke master in 510b6c07
01:48 whiteknight I did?
01:49 dalek parrot: dd22615 | bacek++ | compilers/imcc/main.c:
01:49 dalek parrot: Fix build
01:49 dalek parrot: review: https://github.com/parrot/parrot/commit/dd22615628
01:49 bacek_at_work whiteknight, -Werror=declaration-after-statement
01:49 bacek_at_work and this commit ^^^^
01:49 whiteknight ah, damnit
01:50 whiteknight bacek++
01:54 ttbot Parrot 465b364f MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/83620
01:54 kid51 joined #parrot
01:55 nopaste "kid51" at 70.85.31.226 pasted "t/dynoplibs/io.t: test failures" (16 lines) at http://nopaste.snit.ch/139199
01:56 nopaste "kid51" at 70.85.31.226 pasted "t/library/osutils.t: test failures" (13 lines) at http://nopaste.snit.ch/139201
01:57 whiteknight blah. Those are getting fixed tomorrow. I'm going to bed ow
01:57 whiteknight now
01:58 whiteknight goodnight
01:59 nbrown cotto: I now have a working linux vm which is much less forgiving than my windows box. Also, I fixed the segfault
01:59 nbrown linux++
03:13 benabik joined #parrot
03:17 benabik ~~
03:30 nbrown cotto: check out my last commit. It adds your feature request
03:31 nbrown cotto: as for how to test the debugger... I have no immediate ideas for how to do that, but I'll start thinking about it
03:47 eternaleye_ joined #parrot
05:07 cotto nbrown, I'm constantly impressed at how forgiving your main machine is
05:17 cotto automated debugger tests are tricky.  You might make it scriptable, but then you'd be maintaining an interface that's only (or primarily) useful for testing and you'd have to parse the output.
05:25 dalek parrot: 9fea01c | alvis++ | docs/index/ (2 files):
05:25 dalek parrot: Applied Coke's patch in issue #318 - plus a'bit of chapter reorganization for presentation purposes. This closes issue #318.
05:25 dalek parrot: review: https://github.com/parrot/parrot/commit/9fea01c4a1
05:32 aloha (parrot/parrot) Issues closed : 318 (Subject: Draft Parrot Book chapters no longer available on docs.parrot.org) by brianwisti : https://github.com/parrot/parrot/issues/318
06:08 adu joined #parrot
06:09 adu hey
06:14 cotto hi adu
06:15 adu I'm having some trouble with HLL stuff
06:16 adu I made a grammar, checked that it works in rakudo
06:16 adu and now I'm getting "Unable to obtain PAST from Regex;Match"
06:16 adu when I run my language
06:25 cotto If you're using Rakudo, #perl6 in freenode is the best place to ask.
06:26 adu I'm not, I'm using parrot-nqp
06:26 cotto ah
06:27 cotto do you have a repo you're working in?
06:28 adu nope, I can pastebin it
06:30 adu http://pastebin.com/7nTZ1kjq
06:36 cotto it'd be helpful if you could put it up on github.  I don't know nqp grammars well enough to spot what's wrong and digging any deeper means setting up some scaffolding to run the grammar.
06:36 bacek_at_work adu, you need some "Actions"
06:36 cotto that said, nqp is a subset of Perl 6, so the #perl6 crowd might be able to spot what's wrong pretty quickly.
06:36 cotto or you can ask bacek++
06:36 bacek_at_work adu, your grammar looks good (enough :)
06:37 adu bacek_at_work: I tried writing actions but I keep getting "Unable to obtain PAST" so I don't know if the Actions file is getting better or worse
06:38 bacek_at_work adu, try Actions with single "method TOP"
06:38 bacek_at_work or use "--target=parse" in CLI to stop at parsing phase
06:39 bacek_at_work But using NQP for parsing Scheme is... overkill ;)
06:40 adu I know
06:40 adu but so is JSON ;)
06:43 bacek_at_work adu, I wrote JSON parser without Perl6 Grammars :)
06:43 adu heh
06:44 adu but iirc, there's been a couple json .nqp implementations floating around
06:44 bacek_at_work mine is in nqp. But with hand-crafted parsing
07:04 zby_home joined #parrot
07:10 alin joined #parrot
07:11 fperrad joined #parrot
07:49 dalek rakudo/nom: 2abb9b7 | moritz++ | src/core/Rat.pm:
07:49 dalek rakudo/nom: fix Rat.floor
07:49 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/2abb9b7205
07:49 dalek rakudo/nom: b0cd3cb | moritz++ | src/core/Rat.pm:
07:49 dalek rakudo/nom: rewrite Rat.Str to use code by TimToady++
07:50 dalek rakudo/nom:
07:50 dalek rakudo/nom: in particular it does not go through Num now
07:50 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/b0cd3cbb00
07:50 dalek rakudo/nom: 24ffc9c | moritz++ | src/core/Str.pm:
07:50 dalek rakudo/nom: Revert "add native str concatenation op"
07:50 dalek rakudo/nom:
07:50 dalek rakudo/nom: This reverts commit ad3a7b6403d430a681a497b9ec4b67d0eeeacd52.
07:50 dalek rakudo/nom: It lead to occasional segfaults in splice.t
07:50 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/24ffc9c725
08:04 alin joined #parrot
08:28 lucian joined #parrot
10:24 wknight-phone joined #parrot
10:53 alin joined #parrot
11:10 JimmyZ joined #parrot
11:12 lucian joined #parrot
11:53 alin_ joined #parrot
11:58 alin_ joined #parrot
12:05 crab2313 joined #parrot
12:22 whiteknight joined #parrot
12:24 whiteknight good morning, #parrot
12:24 tadzik good afternoon whiteknight
12:27 whiteknight hello tadzik
12:28 bacek joined #parrot
12:39 Coke cotto, whiteknight, notfound: +1 from me on closing wishlist items. My only concern is that we don't close tickets merely because they are old (which in the past has happened a bit) - also in teh past, the list would get cc'd on commented tickets. now that that's not happening, is there a double check?
12:41 whiteknight Coke: I'd be more than happy to start dumping wishlist items to a wiki page, so long as they aren't cluttering up our ticket queue
12:42 whiteknight Coke: I don't want those ideas to be lost, they just shouldn't be tickets
12:42 Coke alvis: I don't see my name anywhere on Issue#318. What?
12:54 bacek Coke, jfyi I fixed packfile test fail in rm_pasm branch.
12:56 smash joined #parrot
12:56 PacoAir joined #parrot
12:56 Coke bacek: oh, it was fixable? nifty.
12:56 Coke feel free to commit to coke/rm_pasm.
12:57 smash hello everyone
12:57 Coke er, s/commit/push/ (old svn verbs die hard.)
12:57 Coke SMASH!
12:57 Coke ... now I want to go see avengers!
12:57 Coke (well, to be fair, I've been wanting to see it since january or so. ;)
12:58 dalek parrot: 95b238c | Whiteknight++ | t/library/osutils.t:
12:58 dalek parrot: Fix small test failure in osutils.pir.
12:58 dalek parrot:
12:58 dalek parrot: We have an rindex op now, so we need to quote functions of the same name. It seems like IMCC should be smart enough to see that this is a method call not a bare op, but it isn't. Shame. kid51++ for reporting this failure.
12:58 dalek parrot: review: https://github.com/parrot/parrot/commit/95b238cbd5
13:00 PerlJam joined #parrot
13:03 dalek parrot: 5b4324e | Whiteknight++ | t/dynoplibs/io.t:
13:03 dalek parrot: Remove bad test from t/dynoplibs/io.t.
13:03 dalek parrot:
13:03 dalek parrot: Overwriting a class by name in the global class hash IS NOT how we provide the ability to use custom classes in place of native ones. The mechanism to use is .hll_map. Relying on the behavior (or even continued existance) of the global class hash is a bad idea in all user-facing cases. kid51++ for reporting this failure.
13:03 dalek parrot: review: https://github.com/parrot/parrot/commit/5b4324ec3b
13:06 whiteknight actually, if people really want the ability to overwrite the default class mappings in an interp-global way that will affect multiple HLLs, we can add an interface for that
13:06 whiteknight but modifying the class hash directly is not a good idea
13:09 JimmyZ_ joined #parrot
13:10 aloha (parrot/parrot) Issues closed : 273 (readline_interactive method no longer returns PMCNULL on eof) by flh : https://github.com/parrot/parrot/issues/273
13:14 JimmyZ joined #parrot
13:15 aloha (parrot/parrot) Issues closed : 632 (compacting/copying gc) by allisonrandal : https://github.com/parrot/parrot/issues/632, 625 (ordered destruction) by allisonrandal : https://github.com/parrot/parrot/issues/625
13:18 whiteknight I wish aloha mentioned who closed the ticket, instead of mentioning the person who created it twice
13:18 whiteknight And, if some karma happened along the way.... :)
13:18 whiteknight in a couple years, I'm cashing mine in for that donut
13:19 whiteknight retire comfortably
13:39 travis-ci joined #parrot
13:39 travis-ci [travis-ci] parrot/parrot#334 (master - 5b4324e : Whiteknight): The build was fixed.
13:39 travis-ci [travis-ci] Change view : https://github.com/parrot/par​rot/compare/95b238c...5b4324e
13:39 travis-ci [travis-ci] Build details : http://travis-ci.org/parrot/parrot/builds/1284985
13:39 travis-ci left #parrot
13:39 whiteknight damn skippy
13:47 crab2313 joined #parrot
13:48 JimmyZ_ joined #parrot
13:51 brrt joined #parrot
13:53 ttbot Parrot 5b4324ec MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/83884
14:02 whiteknight Uggh, we really need to fix sprintf and friends
14:02 whiteknight this engine is...suboptimal
14:03 whiteknight It's still written as if we have mutable strings, and is doing a lot of concats which create extra GC pressure
14:03 whiteknight we should create a StringBuilder instead, or use a similar mechanism
14:06 whiteknight What we need to do first is make sure we have a decent set of tests and benchmarks for sprintf. Then we can start rewriting it to keep the tests happy and make the benchmarks better
14:10 * JimmyZ only cares about 6model and m0 ...
14:11 brrt testing sprintf is easy enough
14:11 whiteknight I'm just trying to get some tickets closed before the release
14:11 Jashwanth joined #parrot
14:16 aloha (parrot/parrot) Issues closed : 364 (a ticket about the ticket system) by gerd : https://github.com/parrot/parrot/issues/364
14:17 adu joined #parrot
14:21 alin_ joined #parrot
14:22 brrt joined #parrot
14:26 aloha (parrot/parrot) Issues closed : 660 (load_language "nqp" fails) by Whiteknight : https://github.com/parrot/parrot/issues/660
14:27 alin_ joined #parrot
14:28 dalek parrot/coke/rm_pasm: b3a2163 | bacek++ | t/pmc/packfileconstanttable.t:
14:28 dalek parrot/coke/rm_pasm: Rework test to not rely on constants order.
14:28 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/b3a21635a8
14:31 * brrt wonders how to invoke sprintf from winxed
14:33 brrt or, how tests are written in winxed
14:35 crab2313 joined #parrot
14:37 whiteknight brrt: winxed has an sprintf builtin
14:37 whiteknight sprintf("foo", [args as an array])
14:37 benabik joined #parrot
14:38 brrt ok, right :-)
14:38 brrt is that the same sprintf you want to test?
14:38 whiteknight yeah
14:38 whiteknight winxed is a very thin layer over parrot's features
14:38 brrt explains a lot about the errors i got by the way
14:38 brrt winxed is javascriptish
14:38 whiteknight yes
14:38 brrt i like javascript, ergo, winxed is good enough for me
14:39 brrt (also, how and when did it happen that people started comparing languages to javascript?)
14:39 benabik whiteknight: +1 to StringBuilder
14:39 benabik _very_ required for efficient string building w/ immutable strings
14:40 whiteknight sprintf is allocating a huge STRING buffer to work with, then it does a concat operation which effectively blows that away and GC reclaims it
14:40 benabik eep
14:42 whiteknight I'm actually surprised sprintf works, because it's passing that buffer to the C library snprintf to render integers and floats, and that buffer might not be as big as advertised
14:43 benabik brrt: When javascript became ubiquitous.  Everything used to be C because every Unix system came with it.  Now _everybody_ has a browser, and many servers are running node.
14:47 benabik "Benabik closed the issue in a few seconds."  I will close the ticket shortly?  What?
14:47 tadzik yeah, always funny :)
14:47 * benabik wonders if github just adds a second or two to the close time when you "comment & close" just to keep the timeline in order.
14:47 * brrt brb
14:47 tadzik brrtbrb
14:48 benabik tadzik: Gazunteight?
14:48 tadzik ;)
14:48 whiteknight benabik: quantum entanglement. You're entangled with the ticket, Github knew before you clicked the button that it was going to be closed
14:51 aloha (parrot/parrot) Issues closed : 611 (bytecode generation from post) by allisonrandal : https://github.com/parrot/parrot/issues/611
14:59 dalek winxed: 8a6d47a | NotFound++ | winxedst (2 files):
14:59 dalek winxed: show the name, not the 'const' word in errors about const statements
14:59 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/8a6d47ae27
15:02 JimmyZ_ joined #parrot
15:03 benabik whiteknight: I created a central Tasklist page, and discovered that [[Text | Wiki Page]] links work.  :-)
15:03 whiteknight fun
15:08 NotFound brrt: winxed is javascriptish mainly because most these days people has some knowledge of javascript (even if only to criticize it).
15:08 * benabik saves the research on hashing from #244 in the wiki.
15:10 * brrt back
15:10 preflex joined #parrot
15:13 whiteknight JavaScript syntax is very good, except for automatic semicolon insertion
15:13 whiteknight JavaScript object model semantics are crazy and embarrasing
15:14 NotFound The language designer main choice: either you are ready to write long documents and examples about how to write even the most simple things with your new clever syntax, or you borrow most of C/java/javascript.
15:19 NotFound whiteknight: even in that there is no unanimity. Some people says that automatic semicolon insertion is beautiful and that people who has problem with it are just lazy.
15:20 NotFound (Maybe they do it just to provoque long discussions in his blogs, though)
15:20 whiteknight ASI is a huge source of bugs, because the things that constitute a statement break are not clear, and subject to change in future revisions
15:29 brrt is there a testing framework thing for winxed, or parrot in general?
15:30 Coke define "testing framework"?
15:30 Coke like smolder.parrot.org ?
15:30 brrt something that allows you to say is('foo', bar, 'foo is bar')
15:30 Coke yes. look at t/
15:30 Coke half those tests are written in PIR
15:30 brrt ok, will do
15:31 NotFound In winxed tou can do: using extern Test.More plan, ok, is;
15:31 benabik runtime/parrot/library/Test/More.pir
15:32 benabik NotFound: Will using extern automatically do the load_bytecode?
15:32 NotFound Than injects plan, on and is into the current namespace.
15:32 NotFound benabik: yes
15:33 NotFound benabik: it loads and import at runtime, and declare the symbols in the scope at compile time.
15:33 benabik NotFound: Have I mentioned that winxed is awesome lately?
15:35 bluescreen joined #parrot
15:38 preflex_ joined #parrot
15:39 brrt joined #parrot
15:43 dmalcolm joined #parrot
15:44 whiteknight brrt: Rosella also provides a testing library and harness, which I prefer
15:44 whiteknight it's a little bit of extra effort to get the dependencies installed
15:56 alvis Coke: See https://github.com/parrot/parrot/issues/318 (Coke commented ... a year ago)
16:12 brrt joined #parrot
16:15 Coke alvis: huh. must have been looking at the wrong ticket. danke.
16:15 Coke alvis++
16:42 cotto ~~
16:43 crab2313 left #parrot
16:53 PacoAir joined #parrot
17:12 preflex joined #parrot
17:18 contingencyplan joined #parrot
17:22 whiteknight mercurial is showing me a commit with a message of "1" and no modified files
17:22 dukeleto ~~
17:22 dukeleto whiteknight: yeah, mercurial is LTA
17:23 whiteknight Everybody who uses it claims it's just as good as git or better
17:23 dukeleto whiteknight: depends on who you talk to, I guess.
17:23 whiteknight but in the past few months of use, I have been absolutely unimpressed in every possible way
17:24 benabik I think it was mercurial that implemented branches as a file in trunk.
17:24 dukeleto whiteknight: mozilla switched from CVS to Mercurial, but there is a large faction of the company which is using Git+Github
17:24 tadzik working with it on $dayjob?
17:24 benabik Each branch was a line with branch name -> commit id
17:24 whiteknight branches in mercurial are global and permanent, so if you create one it always pushes by default and can never be deleted
17:24 whiteknight so, nobody makes branches
17:24 tadzik whiteknight: ^
17:24 benabik Which removes the point of using a DVCS
17:24 dukeleto whiteknight: precisely because mercurial is dog slow compared to git, especially with repos with >10 years of history
17:25 whiteknight tadzik: yeah, $dayjob. One of our clients insists on using mercurial instead of the SVN we use internally. Honestly, I far prefer svn
17:25 tadzik oh wow
17:25 tadzik I didn't think I'll ever hear thar
17:25 whiteknight yeah, it's that bad
17:25 tadzik wow
17:26 whiteknight in their defense, Mercurial does have a notion of bookmarks or something, which is like what "branches" are for every other VCS ever
17:26 tadzik what besides this branches thing?
17:26 whiteknight The branches thing is my biggest annoyance, though the performance is pretty bad too
17:26 tadzik I never really worked with hg, but I always had the impression that it's just a "Python git"
17:31 cotto svn is nice because you can put on your git-svn gloves
17:33 whiteknight The svn repo here at work is so screwy too. We've got hundreds of projects in a deeply nested folder structure in a single repo
17:34 whiteknight and even THAT is preferrable to mercurial, in my limited experience with i
17:34 whiteknight it
17:34 crab2313 joined #parrot
17:40 brrt whiteknight: we actually recently changed to such a layout
17:41 brrt experience has not universally been pretty
17:42 benabik Big SVN repos leads to corruption, AFAICT
17:42 whiteknight I've never had an issue with file corruption in SVN before
17:42 whiteknight and I've seen some pretty big repos
17:43 benabik It's probably gotten better.  I got bit by SVN back in the 1.1 or so days.
17:43 whiteknight oh, okay
17:44 benabik "Hey this is better than CVS!  Hm, dealing with these databases is a pain."
17:44 dukeleto this seems dangerous: https://github.com/blog/112​1-introducing-team-mentions
17:44 moritz SVN had these fun things where your checkout ended up in irrepairable states just from moving files around
17:44 brrt i have seen the following: copy a directory (we work with 'stub' directories, don't ask me why), work work work, commit inside this directory
17:44 brrt work goes to stub directory
17:44 moritz and I don't mean .svn files/dirs
17:44 whiteknight dukeleto: yes, but powerful. If we are more fine-grained about our team layouts it could be very useful
17:44 dukeleto whiteknight: agreed
17:45 whiteknight A feature I'm really wanting, after Coke mentioned it yesterday, is auto-posting all issues and comments to a list
17:51 whiteknight apparently, I need to start learning Ruby. I've just received word through the grapevine that the whole company is transitioning to a new Rails-based platform in the future
17:52 dukeleto whiteknight: I like Ruby, but Rails is a bit too cultish for me
17:52 dukeleto whiteknight: but Dancer (Rails ported to Perl 5) is pleasant to use
17:52 whiteknight Our new platform is going to be rails on mongodb
17:53 dukeleto whiteknight: because nosql document databases solve everything?
17:53 whiteknight dukeleto: no, not everything. But a pretty convincing case was made that this particular one fits our usage patterns pretty well
17:54 whiteknight I just hope they keep up a good abstraction layer, in case we need to swap out data layers in the future
17:54 dukeleto whiteknight: that seems more reasonable. I see too many people choosing nosql for no good reason other than it is the new shiny
17:54 brrt if i may ask, why mongo over couch?
17:54 tadzik I don't think Dancer is Rails in Perl
17:55 tadzik more like Sinatra
17:55 tadzik Rails seems to be a pretty big beast
17:55 dukeleto tadzik: i meant sinatra
17:55 * dukeleto gets them confused still
17:55 dukeleto tadzik++ # somebody was wrong on the internet :)
17:55 tadzik :)
17:58 dalek parrot: 89cbb85 | jkeenan++ | src/pmc/bigint.pmc:
17:58 dalek parrot: [codingstd] Correct 2 POD formatting errors.
17:58 dalek parrot: review: https://github.com/parrot/parrot/commit/89cbb85724
18:00 whiteknight brrt: I don't know all the details about why they picked mongodb over couchdb or ravendb or any of the other options
18:00 whiteknight Either the differences aren't worth worrying about and we picked one out of a hat, or the differences were important and mongo came out on top
18:03 dukeleto whiteknight: i highly recommend this: http://openmymind.net/2011/3​/28/The-Little-MongoDB-Book/
18:04 dukeleto mongo has a bigger community, from what I can see
18:04 brrt mongo is more popular, yes
18:10 benabik If you're using Rails, hopefully you're using ActiveRecord on top of Mongo.  AR's a nice DB-agnostic storage layer.
18:21 whiteknight joined #parrot
18:21 whiteknight "I'm going to turn off the lights. I don't think your computer is on the same circuit, but you might want to save everything anyway"
18:21 whiteknight and then...blackness
18:32 benabik Uh
18:33 whiteknight yeah, I am glad I saved
18:38 ttbot Parrot 89cbb857 MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/83963
18:40 whiteknight dear ttbot, please shut up. Sincerely, Andrew
18:46 lucian joined #parrot
18:49 Coke isn't that the main we we find out we broke the build on c++?
18:49 Coke that error looks pretty bad, though.
18:56 dukeleto cannot find file 'src/pmc/fixedstringarray.dump' in path
18:56 dukeleto maybe ttbots disk is full?
18:58 Coke could be, ja.
19:17 aloha (parrot/parrot) Issues closed : 686 (sweep-free gc) by allisonrandal : https://github.com/parrot/parrot/issues/686
19:23 aloha (parrot/parrot) Issues closed : 377 ([RFC]: Realtime garbage collector for RTEMS) by leto : https://github.com/parrot/parrot/issues/377
19:28 aloha (parrot/parrot) Issues closed : 659 ([BUG] hash iterator interface changed, needs errata or deprecation notice) by pmichaud : https://github.com/parrot/parrot/issues/659
19:33 aloha (parrot/parrot) Issues closed : 492 ([RFE] Redesign Parrot NCI callback functionality) by japhb : https://github.com/parrot/parrot/issues/492
19:34 whiteknight 472 tickets now. Only 22 left to reach the goal of 450
19:35 whiteknight Far better than the 502 we were at last week
19:35 dalek winxed: a2cbc09 | NotFound++ | winxedst2.winxed:
19:35 dalek winxed: fix inlined return
19:35 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/a2cbc09b01
19:43 dalek winxed/version_1_8: 569699d | NotFound++ | winxedst2.winxed:
19:43 dalek winxed/version_1_8: fix inlined return
19:43 dalek winxed/version_1_8: review: https://github.com/NotFoun​d/winxed/commit/569699dc3c
19:43 dalek winxed/version_1_8: c1ab237 | NotFound++ | winxedst2.winxed:
19:43 dalek winxed/version_1_8: jump to 1.8.1
19:43 dalek winxed/version_1_8: review: https://github.com/NotFoun​d/winxed/commit/c1ab2374d8
19:43 dalek winxed/version_1_8: 9890fe7 | NotFound++ | NEWS:
19:43 dalek winxed/version_1_8: update news
19:43 dalek winxed/version_1_8: review: https://github.com/NotFoun​d/winxed/commit/9890fe71db
19:44 dalek winxed: d68fc48 | NotFound++ | NEWS:
19:44 dalek winxed: update news
19:44 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/d68fc48316
19:49 dalek Heuristic branch merge: pushed 81 commits to parrot/ayardley/READMEs by ayardley
19:52 whiteknight alvis: ping
19:52 alvis whiteknight: pong # Just caught me!
19:54 whiteknight alvis: I want to get as much doc improvements and README improvements into master before the release
19:54 whiteknight so let me know what you have planned, and what help you need
19:55 alvis whiteknight: ok, will do. I'm just about to post to parrot-dev and, then, I'm away until this evening.
19:55 whiteknight okay, thanks. Talk to you later
19:55 alvis later.
19:56 alvis left #parrot
19:56 dalek rakudo/nom: 9a49814 | jnthn++ | src/ops/perl6.ops:
19:56 dalek rakudo/nom: Consider .defined, not just .DEFINITE, when deciding whether to run KEEP/UNDO phasers. Fixes fail() not triggering UNDO, and that in turn fixes the 'let' handling.
19:56 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/9a49814bb0
19:58 crab2313 joined #parrot
20:00 dalek winxed/version_1_8: 7eea17f | NotFound++ | pir/winxed_ (2 files):
20:00 dalek winxed/version_1_8: update installable files
20:00 dalek winxed/version_1_8: review: https://github.com/NotFoun​d/winxed/commit/7eea17f35e
20:04 alin joined #parrot
20:11 dalek parrot: 9b49e86 | NotFound++ | / (3 files):
20:11 dalek parrot: update winxed snapshot to 1.8.1 - Bug fixes
20:11 dalek parrot: review: https://github.com/parrot/parrot/commit/9b49e8651a
20:27 travis-ci joined #parrot
20:27 travis-ci [travis-ci] parrot/parrot#336 (master - 9b49e86 : NotFound): The build was broken.
20:27 travis-ci [travis-ci] Change view : https://github.com/parrot/par​rot/compare/89cbb85...9b49e86
20:27 travis-ci [travis-ci] Build details : http://travis-ci.org/parrot/parrot/builds/1288495
20:27 travis-ci left #parrot
20:33 preflex_ joined #parrot
20:37 perlite_ joined #parrot
20:38 * Coke wonders if now is a good time to suggest to alvis that we just rip out POD and use markdown instead.
20:38 atrodo Coke> Why not?
20:40 ttbot Parrot 9b49e865 MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/83995
21:02 Coke atrodo: because he's in the middle of podifying things.
21:09 dukeleto Coke: i lost that battle
21:10 dukeleto Coke: i created various README.md files, but cotto++ declared that we should consistently use POD
21:10 dukeleto Coke: which is fine with me
21:12 Coke if we had it to do over, I'd drop pod from pir/pasm
21:12 Coke the pain we've had to generate HTML is insane. too much infrastructure code.
21:29 dmalcolm_ joined #parrot
21:30 dmalcolm__ joined #parrot
22:00 lucian joined #parrot
22:24 dukeleto Coke: agreed
22:34 aloha (parrot/parrot) Issues closed : 481 (Make RNG algorithm used by rand dynop pluggable) by japhb : https://github.com/parrot/parrot/issues/481
22:45 kid51 joined #parrot
23:20 eternaleye_ joined #parrot
23:21 arnsholt joined #parrot
23:21 Maddingu1 joined #parrot
23:22 alvis joined #parrot
23:23 dalek parrot: 74e3213 | jkeenan++ | lib/Parrot/Headerizer.pm:
23:23 dalek parrot: POD correction: method was lacking head2.
23:23 dalek parrot: review: https://github.com/parrot/parrot/commit/74e32131a3
23:25 nbrown joined #parrot
23:41 whiteknight joined #parrot
23:41 whiteknight good evening, #parrot
23:45 schmooster joined #parrot
23:57 dalek parrot: e3383c7 | jkeenan++ | include/parrot/extend.h:
23:57 dalek parrot: Remove references to Parrot_free_cstring from HEADERIZER section.
23:57 dalek parrot: review: https://github.com/parrot/parrot/commit/e3383c7905
23:57 dalek parrot: 708e44d | jkeenan++ | MANIFEST:
23:57 dalek parrot: Add docs/index/book.json to MANIFEST.
23:57 dalek parrot: review: https://github.com/parrot/parrot/commit/708e44d3ee
23:58 ttbot Parrot 74e32131 MSWin32-x86-multi-thread make error http://tt.taptinder.org/cmdinfo/84020
23:59 kid51 Hmm, I suspect that's a phony report from taptinder.
23:59 kid51 A commit which is just a POD change shouldn't prevent 'make' from completing.
23:59 kid51 And I build and ran make test before committing that.

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

Parrot | source cross referenced