Camelia, the Perl 6 bug

IRC log for #padre, 2009-12-31

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

All times shown according to UTC.

Time Nick Message
00:12 Sewi gn8
01:40 cognominal joined #padre
01:50 submersible ...braaaains....
01:51 submersible code4pay: sry was asleep ... reading ticket now
01:51 * Alias_ waves from work
01:51 Alias_ (continues hacking away at the Aspect.pm monster)
01:52 Alias_ submersible: I think I've FINALLY got it to the point I can remove the need for those horrible DESTROY-time self-executing blessed code refs
01:55 submersible Alias: I'm quite fond of those majicks ;) what happens instead?
01:55 Alias_ I've done an (exhausting) logical merge of the code generated by Aspect itself, and the code generated by Hook::LexWrap
01:55 Alias_ The blessed coderef thing was needed because of the API boundary
01:56 Alias_ H:LexWrap could never trust that the thing calling it would run the end-of-scope function
01:56 Alias_ Now that the two layers are merged, the coderef is passed to a trusted parent
01:56 Alias_ So it's just a raw anonymous sub to keep the closure scoping stuff, but without the blessed DESTROY part of the Magick
01:57 Alias_ The layer merging means you can ditch all the flexibility, and with it about half the code
01:57 Alias_ Also, on the plus side, it used to return one descoping hook per symbol table hack
01:58 Alias_ With the merge, all symbol table hooks for each Advice instance are able to share the same closure variable
01:58 Alias_ So 100 hooks results in only a single descoping function for all of them
01:59 Alias_ Although adds a little bit more Magick :)
01:59 Alias_ But at least it's more Magick of the same type, and as a result I can remove one entire class of other Magick
02:18 Hyppolit #390: Copy and Paste onto highlighted text (accepted defect) [ http://padre.perlide.org/t​rac/ticket/390#comment:20 ]
02:21 submersible joined #padre
02:23 submersible Alias: ping
02:24 * submersible is suffering optus netwrok pain again
02:24 submersible Alias_: I take it that Aspect will be able to catch die from aspected routines very soon then?
02:28 Alias_ I'm inching closer, yes
02:29 Alias_ I could probably implement it quite quickly now, but I'm delaying a bit until I've got as much weight stripped out as I can
02:29 submersible groovy.
02:29 Alias_ Easier to do while the code is smaller
02:29 submersible I had a play with swarm collaboration over the last few days
02:29 Alias_ Still some tricky bits though
02:30 Alias_ Theoretically, the "after" aspect is supposed to fire on BOTH return and throw
02:30 Alias_ Which means you aren't supposed to be able to set the return value
02:30 Alias_ Because there isn't one
02:30 Alias_ It appears you are supposed to use an explicit "after () returning" (in Java) if you want to hack the return value
02:30 submersible so maybe 'before' 'after' and 'postmortem'
02:31 Alias_ There's 5 main advice rtypes
02:31 Alias_ "before", "after", "after returning", "after throwing", "around"
02:31 Alias_ Where "after" is a combination of the two subtypes
02:31 submersible gotcha
02:31 Alias_ BUT
02:32 Alias_ The _current_ (old and stable) API for Aspect.pm lets you use before in a somewhat similar way to "around"
02:32 Alias_ And "after" is strictly "after returning"
02:32 Alias_ And I'm scared of changing the API to break that
02:32 * submersible smell an Aspect::Simple or Aspect::DWIM
02:32 Alias_ or Aspect2::
02:33 Alias_ ick
02:33 submersible Aspect2  what? :)
02:33 submersible Aspect2Prespect
02:33 submersible stop me
02:33 Alias_ The number thing is a relatively'ish common way to allow API break
02:33 Alias_ It's only used for huge things normally
02:33 Alias_ This seems... a bit trivial to bother
02:34 Alias_ http://www.google.com/codesearch?as_q=%22use+As​pect%22&btnG=Search+Code&hl=en&as_l​ang=&as_license_restrict=i&as_license=&​amp;as_package=&as_filename=&as_case=y
02:35 Alias_ I suppose it's not unreasonable to make after start handling exceptions as well
02:35 submersible at least the audience seems small
02:35 Alias_ That's the greypan audience
02:35 Alias_ Beyond lies 10-100 times that amount in the darkpan
02:35 submersible true
02:35 Alias_ But I should at least be able to audit the greypan to discover typical usage patterns
02:35 submersible on a different topic...
02:36 submersible how do you reckon we turn a regular padre doc/editor combo into a swarmified one?
02:36 Alias_ Not sure...
02:36 Alias_ Well
02:37 submersible presently my thinking is right click 'Swarm this document'
02:37 Alias_ document/project
02:37 Alias_ Or toolbar button "Swarm current document"
02:37 Alias_ Or maybe a separate icon on the notepad...
02:38 Alias_ Well, that would be when it's enabled
02:38 Alias_ To start with though, you could get away with a menu entry with a tickbox
02:38 submersible and do the same 'project relative filepatch' matching
02:39 Alias_ http://www.google.com/codesearch/p?hl=​en#TbLe_3gaALM/components/perl/Aspects.pm&q=%22use%20Aspect%22&sa=N&cd=20&ct=rc
02:39 Alias_ That's the kind of thing I'm a bit worried about
02:39 Alias_ I imagine SOME of those must throw exceptions from time to time
02:40 Alias_ But I suppose if there's a clear "after -> after_returning" workaround to get back to the old functionality...
02:40 Alias_ That's maybe not TOO bad
02:40 Alias_ Worse will be the idea of not allowing "before" to call the original function itself
02:41 submersible I can see that 'before' behaviour biting people badly - yes
02:41 Alias_ Worse, the current before behaviour doesn't cleanly fit the around pattern either
02:41 Alias_ I suppose the upside to Aspect.pm is that the code that is using it SHOULD be simple and in one place, not scattered around
02:42 Alias_ Since that's the entire point
02:42 Alias_ BTW, you have any plans for NYE?
02:42 Alias_ Just home and fireworks on TV with wife?
02:43 submersible we're going to try for the 9pm ones from a backyard party in N.syd
02:43 submersible then escape back across the bridge before the real chaos begins
02:43 Alias_ heh, good luck
02:43 submersible ty :D
02:50 waxhead what's Aspect.pm?
02:51 submersible Alias_: you're taking the stairs to the roof for NYE?
02:52 waxhead man.. how many pies do you have your fingers in Alias?
02:52 waxhead ok.. time for some pointless racing on the ps3...
03:03 Alias_ waxhead: I didn't WRITE it
03:04 Alias_ waxhead: I just took it over to make it work better and suck less
03:17 waxhead man alive.. rally is hard!
05:11 * Alias_ was wondering why Padre was so slow
05:11 cognominal joined #padre
05:11 Alias_ Totally missed I was running 0.52
05:11 Alias_ Which is awesome, because it means that the 0.53 speed up really is tangible
05:11 Alias_ (If everything feels shite-slow when you revert)
05:30 PacoLinux joined #padre
05:33 awnstudio joined #padre
06:30 patspam joined #padre
07:53 waxhead ok outer ear.. have a great NYE
07:54 szabgab joined #padre
07:55 pece joined #padre
08:30 pece2 joined #padre
09:14 kaare joined #padre
09:27 PerlJam joined #padre
09:36 Sewi joined #padre
09:36 Sewi hi all
11:12 pece joined #padre
12:01 siegfri3d joined #padre
12:10 pece2 joined #padre
12:28 pece joined #padre
12:52 Hyppolit svn: r10028 | szabgab++ | http://padre.perlide.org/trac/changeset/10028
12:52 Hyppolit add standard images to the blogs.padre.perlide.org site
12:52 Hyppolit trunk/blogs.padre.perlide.org/www/ trunk/blogs.padre.perlide.org/www/images/
12:57 Hyppolit svn: r10029 | szabgab++ | http://padre.perlide.org/trac/changeset/10029
12:57 Hyppolit replace the standard logo with the one of padre
12:57 Hyppolit trunk/blogs.padre.perlide.org/templates/ trunk/blogs.padre.perlide.org/www/images/
12:57 pece joined #padre
12:57 Hyppolit svn: r10030 | szabgab++ | http://padre.perlide.org/trac/changeset/10030
12:57 Hyppolit README - how to update the blogs.padre.perlide.org site
12:57 Hyppolit trunk/blogs.padre.perlide.org/
13:19 dam joined #padre
13:37 cognominal joined #padre
14:22 Alias__ joined #padre
15:06 CromeDome joined #padre
15:09 pece joined #padre
15:43 patspam joined #padre
15:59 Alias__ Happy New Year eveyone, for those sitting online on NYE still back there in the past :)
16:00 * Alias__ plans a Year in Review post next week for PAdre
16:00 Alias__ (Past year)
16:01 szabgab happy ny
16:12 kentnl the bug report link here is somewhat er, not helpful: http://search.cpan.org/~szabgab/Debug-Client-0.11/
16:12 * kentnl looked in and cant see where DC's bugs are
16:14 szabgab trac?
16:14 Hyppolit trac is http://padre.perlide.org/trac and also http://padre.perlide.org/trac/register
16:14 szabgab there
16:15 kentnl szabgab: I looked, but I can't work out where in trac "Debug Client" related bugs are/should be
16:15 Hyppolit svn: r10031 | szabgab++ | http://padre.perlide.org/trac/changeset/10031
16:15 Hyppolit link to the debugger
16:15 Hyppolit trunk/Debug-Client/
16:16 szabgab select component Debug::Client
16:17 kentnl ah. I see none have been filed.
16:18 kentnl http://www.nntp.perl.org/group/perl.​cpan.testers/2009/12/msg6593263.html  # /me has been getting failures that are lots like this one
16:20 szabgab kentnl, what os?
16:20 kentnl linux, gentoo.
16:21 szabgab perl version?
16:21 kentnl 5.10.1
16:22 szabgab first of all you can force install it and try to use it but I'd appreciate your help to work out the source of the problem
16:22 szabgab could you run    prove -bv t/09-io.t
16:22 kentnl it doesn't emit the same weird unicode character, the output of the test is confusing because as far as I can tell, the printed strings are identical
16:22 szabgab or whatever test fails
16:23 * kentnl has to run it though cat -A to see it
16:23 kentnl http://scsys.co.uk:8001/37708
16:24 szabgab then maybe you can try to wrap the prining in some sign around the string:   '$var'   so we might see trailing spaces
16:25 szabgab let me check it
16:27 szabgab could you put a diag(" '$out' "); on line 139 and rerun?
16:28 szabgab it seems to have some extra characters at the end
16:28 szabgab probably after line 139
16:31 kentnl diag won't show the magic character
16:31 kentnl its there, but cat -A is required to see it
16:33 kentnl szabgab: http://scsys.co.uk:8001/37709
16:34 kentnl \e[?1034h # is what its printing
16:36 Sewi Happy new year Alias :-)
16:49 szabgab could you try to do the whole session manually in the debugger?
16:51 szabgab I am pretty clueless where does this thing come from, and how to disregard it
16:53 szabgab oh, this is the expected standard output of the script, not even debugging commands
16:53 szabgab and as I can see it only happens after the we called quit
16:54 szabgab 1034 in hexa?
17:05 kentnl \e[?1034h appears to correspond with 'nterpret "meta" key, sets eighth bit. (enables the eightBitInput resource).'  from the XTerm escape sequence list
17:06 kentnl so with that in consideration, it might be a Term:: interaction if the debugger is internally using that somehow
17:10 kentnl szabgab: tada!, removing Term::Readline::Gnu solves the problem !
17:13 virtualsue joined #padre
17:20 Getty hey dudes, hope you get fresh into the new year
17:21 Getty szabgab: you dont party? you cleanup code/server? ;)
17:21 Sewi Getty: szabgab is not far away from our timezone, he still has some time left until midnight :-)
17:22 Getty Sewi:that doesnt mean he should work
17:22 Sewi You don't work atm?
17:22 Sewi shame on you
17:27 Getty once i may not work
17:27 Getty i actually cant work
17:39 szabgab kentnl, thanks
17:40 szabgab do you mean uninstalling  Term::Readline::Gnu solved it?
17:40 szabgab or what was the magic?
17:41 szabgab Getty, Sewi we don't really celebrate here, by chance tomorrow is not a workday as it is Friday but usually 1st Jan is just a regular workday
17:41 szabgab but kidz will have school tomorrow
17:53 Sewi szabgab: Do you have another "new year" event (from another calendar)?
17:54 Getty szabgab: wt.?
17:55 szabgab Sewi, yes, it is usually around September
17:56 szabgab many people really celebrate but wit kids we are sort of heavy
17:56 szabgab and besides I have a slight fever
18:43 code4pay joined #padre
18:44 code4pay Happy New Year
18:44 szabgab code4pay, its a bit early for you, isn't it ?
18:45 szabgab anyway, happy new decade!
18:45 code4pay yeah, I'm a morning person
18:53 Getty code4pay: you are already in?
18:53 Getty so you are a japanese or an australian ;)
18:53 code4pay Australia, it is 5:52
18:53 Getty oh so even  east australia?
18:54 code4pay yes, needless to say News Eve was quiet, in bed by 10:30 :-(
18:54 Getty For the last laughter in this year (or the first one in this year): http://www.raudssus.de/slashfil​m_the_phantom_menace_review.rar
18:54 Getty wait
18:54 Getty you slept over 0?
18:54 Getty wtf?
18:54 Getty give me the concept, i dont get it
18:55 Getty why someone sleeps at 0 ?
18:55 code4pay tired,  go to sleep :)
19:13 Getty what?
19:13 Getty dont get it
19:13 Getty you know that image that destroys the borg collective?
19:14 code4pay I once went to bed at 11:45 on new years eve
19:15 szabgab this year we celebrate new years eve in Iran
19:15 szabgab so we can go to sleep at 10:05 pm
19:21 Getty !??!???!?!!
19:21 Getty dudes
19:21 Getty you destroy my brain
19:21 Getty i dont get it
19:34 szabgab do you guys know if Thomas Fahle is on IRC ?
19:45 Getty you dont have any idea how it nick could be so "around"? so far that name doesnt trigger anything
19:46 Floyd-ATC joined #padre
19:48 szabgab Getty, nope but he just posted on my blog that he created a live DVD for Perl based on Ubuntu
19:48 szabgab and I wanted to talk to him
19:49 szabgab next year
19:49 Getty baeh...
19:49 Getty i hate ubuntu, you know it ;)
19:49 Getty you shouldnt have told me
19:49 Floyd-ATC Happy new year all
19:49 SvenDowideit_ joined #padre
19:50 Floyd-ATC I have a weird problem: tutorial scripts using Win32::GUI work from command-line, doesn't show anything when run with Padre
19:50 Floyd-ATC ...any idea why?
19:51 Getty define "Padre", how you installed it? standalone installer or via cpan?
19:51 Floyd-ATC Standalone installer, then upgraded to .53 via CPAN
19:51 Getty do you have perl installed beside padre or tried to make it, or do you just have padre installed to get perl
19:51 Getty no other perl involved?
19:51 Getty just padre given?
19:52 Floyd-ATC no other perl
19:52 Getty mh ok...
19:52 Getty sounds strange then
19:52 Getty feels a bit like indiana jones 4... it doesnt make any sense :D
19:53 Floyd-ATC The scripts appear to start, "checkpoint" print lines show as expected but no window ever appears
19:53 Getty yeah
19:56 Getty cause there is no other perl involved its strange, shouldnt happen, definitly some kind of bug in the involved modules and/or in padre
19:56 Getty (or a misunderstanding, win32 stuff always has its special magic)
19:57 szabgab Floyd-ATC, oh you are trying the debugger?
19:57 szabgab first try to run the script with F5 to see if that works
19:57 szabgab the debugger might be very buggy :-)
19:57 szabgab s/might be/must be/
19:57 szabgab s/must be/is/
19:58 Getty lol
19:58 Floyd-ATC I'm using F5 to run the script, not yetr familiar with the debugger
19:58 Floyd-ATC yet*
19:58 Getty i suggested this
19:58 Getty ;)
19:59 Getty Floyd-ATC: are you an ATC?
19:59 Floyd-ATC http://www.atc.no/
19:59 Getty ah ok
19:59 Getty so no real ATC ;)
19:59 Floyd-ATC "How do you define real?"
19:59 Getty Air Traffic Control
19:59 Floyd-ATC ;-)
20:00 Floyd-ATC oh... no :-)
20:00 Getty goes there is a /usr/games/atc on the unix world i define it as standard! ;)
20:00 Getty btw awesome game
20:00 Getty i love it
20:00 Floyd-ATC never seen it, sorry
20:00 Floyd-ATC anyway, you should know that case matters
20:00 Getty its an ascii terminal game, where you are an air traffic controller
20:00 Getty if you make one mistake the game is over
20:01 Getty its the best stress game ever
20:01 Floyd-ATC Oh I know how that works, I have a wife
20:01 Getty lol
20:02 szabgab Floyd-ATC, so you say if you press F5 your wife does not work
20:02 szabgab err, the script
20:02 Floyd-ATC None of the above, sadly
20:03 Getty you must press the on button on the vibrator
20:03 Getty or the on button on the electric collar
20:03 Getty whatever works best
20:03 szabgab does it get stuck or does it exit without a window?
20:03 szabgab Getty, please
20:04 Getty sorry, its new years eve :-P
20:04 Getty but YOU dont care about! <cry>
20:04 Floyd-ATC Let's see... If I start the debugger, it stops at the first line. I then set a breakpoint at Win32::GUI->Dialog; and tell it to run until breakpoint.
20:04 Getty beside... the bot with the profanity check is in the SVN :-P
20:04 Getty <reminder>
20:04 szabgab I have a dead-line in less than 2 hours
20:05 Floyd-ATC then "enter into", and Padre becomes unresponsive
20:05 Getty ok then kill it!
20:05 Getty Floyd-ATC: ask again in 3 hours ;)
20:05 Getty Floyd-ATC: when the deadline is over
20:05 szabgab Floyd-ATC, oh no worry I have plenty of time to finish this thing
20:06 Getty hehe
20:06 Floyd-ATC You're actually going through with the insane idea of a new year release?
20:06 Floyd-ATC :-D
20:06 szabgab and slowly I understand whan I need to do
20:06 szabgab switching servers
20:06 szabgab one website to go
20:07 Floyd-ATC Am I the only one here drinking champagne right now?
20:08 code4pay well its 7.00am in the morning here so I'm not
20:08 Floyd-ATC 9pm here
20:09 code4pay well I can tell you 2010 has been uneventful so far
20:10 code4pay how are things back in 2009?
20:11 Floyd-ATC We still have a lot of problems
20:12 Floyd-ATC things I'm sure you people in the future don't have to worry about
20:15 szabgab Floyd-ATC, so I don't understand have you run the script using F5 or the debugger? - you said you stepped in and so
20:15 Floyd-ATC I have tried with F5 and I've now tried with the debugger
20:17 Floyd-ATC With F5 it appears to run until halted with F6 but I can't see anything except text output. With the debugger Padre freezes when I enter Win32::GUI->Dialog(), otherwise everything's the same
20:17 szabgab I'd suggest inserting print statements for now and seeing where does it go, both from the command line and from padre using F5
20:18 szabgab the freeze is probably due to handing over execution to the event loop of the script and not noticing it...
20:20 Floyd-ATC prints statements emit as expected, the script runs fine and appears to enter the dialog loop
20:20 Floyd-ATC the only thing missing is the actual window
20:20 Floyd-ATC from the command-line, the window shows and can be closed (responds to clicks)
20:22 Floyd-ATC Stepping through GUI programs with a debugger usually presents a cartload of issues so I don't blame Padre for that :-)
20:23 szabgab Floyd-ATC, what if you configure padre to use external window for running?
20:24 Floyd-ATC Hmm let's see
20:24 szabgab Edit/Preferences/Run Parameters   checkbox
20:24 Floyd-ATC And voilá, there it is
20:24 Getty szabgab: you cant resist if there is a serious problem with padre, or? ;)
20:25 Floyd-ATC It works now, what exactly does that option do?
20:26 Getty then it opens a window
20:26 Getty instead of none
20:28 szabgab does your script expect anything on STDIN?
20:28 Floyd-ATC No
20:29 szabgab the external mode runs cmd and in that cmd it runs your script
20:29 Floyd-ATC I see the output appears in a separate console window, yes...
20:29 szabgab the internal mode tries to grab the STDOUT and STDERR of your script and to show it in our output window
20:29 szabgab the thing that we use for this or the way we use it might be problematic
20:30 szabgab and it cannot handle STDIN
20:30 szabgab so for now you can use it this way till we figure out how to fix the internal thing
20:30 Floyd-ATC should I file a bug report then?
20:32 Getty yes
20:38 Getty tickets are always good
20:43 Floyd-ATC If they provide new and useful information, anyway :-)
20:52 Getty it just needs that they mark out a problem
20:52 Getty and then some developer can take it and solve it
20:53 Getty i added more tickets then i solved, sadly, cause i'm not that good, but the tickets are there, and someone will solve it someday
21:17 rindolf joined #padre
21:18 rindolf Hi all.
21:36 cognominal joined #padre
22:27 cognominal joined #padre
23:01 Sewi happy new year GMT+1 :-)

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