Camelia, the Perl 6 bug

IRC log for #padre, 2010-01-05

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

All times shown according to UTC.

Time Nick Message
00:28 Hyppolit svn: r10051 | zipf++ | http://padre.perlide.org/trac/changeset/10051
00:28 Hyppolit Updated Spanish translation
00:28 Hyppolit trunk/Padre/share/locale/
00:40 Alias_ Is there a term for software designed for teams?
00:41 Alias_ Other than, ugh, Groupware
01:16 zamolxes joined #padre
01:23 CromeDome joined #padre
01:37 CromeDome joined #padre
02:22 jazzanova joined #padre
02:58 cognominal joined #padre
03:44 Alias joined #padre
03:47 CromeDome joined #padre
03:48 patspam joined #padre
03:56 CSJewell joined #padre
04:03 Hyppolit svn: r10052 | szabgab++ | http://padre.perlide.org/trac/changeset/10052
04:03 Hyppolit link to blog of Alias
04:03 Hyppolit trunk/template.padre.perlide.org/data/
04:21 Hyppolit #813: File in file cannot find $self (new defect) [ http://padre.perlide.org/trac/ticket/813 ]
04:21 Alias szabgab: Does everything in that post match approximately your thoughts?
04:24 Hyppolit #814: Find-in-files result window, clicking on filename does not do anything (new enhancement) [ http://padre.perlide.org/trac/ticket/814 ]
04:44 Hyppolit svn: r10053 | adamk++ | http://padre.perlide.org/trac/changeset/10053
04:44 Hyppolit Adding Padre::Test module to indicate that Padre is running inside of a test suite. You should only need to use this to suppress things like visible Padre windows during testing. It isn't needed everywhere if you don't have issues.
04:44 Hyppolit trunk/Padre/lib/Padre/
04:54 Hyppolit svn: r10054 | adamk++ | http://padre.perlide.org/trac/changeset/10054
04:54 Hyppolit When loading the test class, also disable the splash screen
04:54 Hyppolit trunk/Padre/lib/Padre/
04:55 Hyppolit svn: r10055 | adamk++ | http://padre.perlide.org/trac/changeset/10055
04:55 Hyppolit Adding support for --invisible
04:55 Hyppolit trunk/Padre/
04:55 Hyppolit svn: r10056 | adamk++ | http://padre.perlide.org/trac/changeset/10056
04:55 Hyppolit Newer style tidy-friendly Class::XSAccessor usage
04:55 Hyppolit trunk/Padre/lib/
04:56 Hyppolit svn: r10057 | adamk++ | http://padre.perlide.org/trac/changeset/10057
04:56 Hyppolit Hook ->Show to never show when Padre::Test is loaded
04:56 Hyppolit trunk/Padre/lib/Padre/Wx/
04:57 Hyppolit svn: r10058 | adamk++ | http://padre.perlide.org/trac/changeset/10058
04:57 Hyppolit Noting invisibility
04:57 Hyppolit trunk/Padre/
05:05 submersible joined #padre
05:06 Alias_ joined #padre
05:09 jazzanova joined #padre
05:18 jazzanova joined #padre
05:29 jazzanova joined #padre
05:36 * patspam gets experimental clickable filenames in console working
05:36 patspam for Perl error messages
05:37 patspam clicking filename opens file and line number
05:37 patspam but it's hacky
05:47 Hyppolit #815: Clickable filenames in Console (new enhancement) [ http://padre.perlide.org/trac/ticket/815 ]
05:53 Hyppolit #815: Clickable filenames in Console (new enhancement) [ http://padre.perlide.org/trac/ticket/815#comment:1 ]
05:55 Alias patspam: The first version of everything in Padre is almost always hacky
05:55 Alias It's considered a good thing :)
06:01 Hyppolit svn: r10059 | adamk++ | http://padre.perlide.org/trac/changeset/10059
06:01 Hyppolit Added missing copyright
06:01 Hyppolit trunk/Padre/lib/Padre/
06:03 Hyppolit svn: r10060 | adamk++ | http://padre.perlide.org/trac/changeset/10060
06:03 Hyppolit Added critic exception, since this is really test code and not core code
06:03 Hyppolit trunk/Padre/lib/Padre/
06:29 kaare joined #padre
06:38 patspam Alias: would also add Path::Class::URI dependency
06:38 patspam that ok?
06:41 cognominal joined #padre
06:41 patspam ..or are you saying i should just commit it and let other people fix it =p
06:45 Alias More dependencies are generally considered bad...
06:45 Alias How many more deps will it suck in we don't already have?
06:45 Alias Or is there a way to achieve it without the dep
06:45 Alias We've been trying to keep Padre outside the Top 100 heaviest modules
06:45 Alias And we're only barely succeeding
06:45 Alias http://ali.as/top100/
07:04 Alias ugh
07:04 Alias I'm sick of having no hot key for the vertical align feature
07:04 Alias So I'm taking CTRL-SHIFT-Space
07:05 Alias It slightly violates our normal pattern of "Shift means the same as the non-shift version, but smarter"
07:05 Alias But it will have to hold until we can find something better
07:08 Hyppolit svn: r10061 | patspam++ | http://padre.perlide.org/trac/changeset/10061
07:08 Hyppolit Adds experimental support for clickable filenames in Output panel.
07:08 Hyppolit
07:08 Hyppolit Currently only matches: <error> at <file> line 5.
07:08 Hyppolit Note that Padre::Wx::Output now subclasses Wx::RichText, which
07:08 Hyppolit may break stuff.
07:08 Hyppolit trunk/Padre/ trunk/Padre/lib/Padre/Wx/
07:09 patspam I removed the Path::Class::URI dependency
07:09 Alias Awesome
07:09 Alias Did it cause any problems?
07:09 Hyppolit #815: Clickable filenames in Console (new enhancement) [ http://padre.perlide.org/trac/ticket/815#comment:2 ]
07:11 Alias hrm
07:11 * Alias ponders the idea of "Close passing tests"
07:24 marcela joined #padre
07:27 jazzanova joined #padre
07:44 patspam joined #padre
08:07 patspam joined #padre
08:32 pece joined #padre
08:35 szabgab patspam, if you are already playing with the output window, could you also make it accept input ?
08:36 szabgab Alias, regarding the blog post, I'd be happy if the standard IDE stuff was already working well (e.g. if the debugger did not suck)
08:37 szabgab and if padre.exe was running the first time :-)
08:38 szabgab oh and on the big plans, to allow padre to be used as a platform to develop other stuff on it
08:39 Sewi joined #padre
08:39 Sewi gm
08:40 pece joined #padre
08:44 Sewi joined #padre
08:45 waxhead joined #padre
08:45 waxhead evening all
08:47 Sewi Hi waxhead
08:48 waxhead how goes the Padre dev?  Ready for a release this week?
08:53 Sewi brbbrb, reboot...
08:56 Sewi joined #padre
09:03 Hyppolit svn: r10062 | Sewi++ | http://padre.perlide.org/trac/changeset/10062
09:03 Hyppolit Upgrading to ProcessStream 0.24 solves #756
09:03 Hyppolit trunk/Padre/
09:04 kent\n joined #padre
09:06 Hyppolit #756: missing output lines (new defect) [ http://padre.perlide.org/trac/ticket/756#comment:2 ]
09:06 Hyppolit #756: missing output lines (closed defect) [ http://padre.perlide.org/trac/ticket/756#comment:3 ]
09:26 SvenDowideit_ joined #padre
09:29 Alias_ Evening all
09:30 Alias_ waxhead: I'd say we're ready for a release right now
09:30 Alias_ waxhead: Before someone goes and breaks it again (like me)
09:30 Alias_ Also, I can't do the PerlTidy and PerlCritic plugin upgrades to support project policy properly until it's out
09:31 waxhead really?
09:31 waxhead I promised a string freeze for the translators!  :)
09:31 Alias_ oh :(
09:31 waxhead it's only 2 days...
09:32 Alias_ Then you probably better email the list and say not to make any code changes if people can avoid it
09:32 waxhead I did... email sent about 15 mnutes ago..
09:32 Alias_ Or maybe make a release branch now, and just do a string merge up later?
09:32 Alias_ ah, ok
09:32 Alias_ I shall refrain from breaking things then :)
09:33 Alias_ And concentrate on the Aspect stuff
09:33 submersible famous last words..
09:33 Alias_ yeah
09:33 submersible :P
09:33 Alias_ submersible, I have a review of Test::Class
09:33 Alias_ It's fucking aweful
09:33 Alias_ awful
09:33 waxhead ooo I'm avoiding the branch thing at the moment.. don't really want to get stuck with a painful merge later
09:33 Alias_ waxhead: A string only merge wouldn't be that hard
09:33 submersible short .. pointed.. only 66% opinion.. not much of a review
09:34 Alias_ submersible: OK, so here's the longer version
09:34 submersible -1 for the typo.
09:34 submersible :)
09:34 Alias_ Imagine you took our entire test suite
09:34 Alias_ And decided that the startup overhead of loading K2 was annoying and stupid
09:34 submersible " in an xUnit/JUnit style"
09:34 Alias_ And hey, if you put the tests into real classes, you could install them all to the system and run them later
09:34 submersible yeah - have the code compiled and eval the damn tests
09:35 Alias_ So, as a pedantic zealot that wants fast tests and future test runs, you write it
09:35 Alias_ And clearly, because you're a pedantic zealot, it's awesome
09:35 submersible or never finished
09:35 Alias_ Everything is layed out is nice deep elegant trees
09:36 Alias_ in
09:36 submersible as is more commonly the case
09:36 Alias_ And all those temporary shitty test classes like My::Foo now have proper places in this tree
09:36 Sewi Could anybody test ProcessStream 0.24 on Win?
09:36 Alias_ Sewi: Will do
09:36 Sewi thanks
09:36 Alias_ submersible: But here's the problem
09:36 waxhead wow.. processtream is moving along..
09:36 submersible ...
09:36 waxhead all that time it stagnated causing us problems..
09:37 Alias_ 1. Everything is in huge deep trees that are almost impossible to explore properly if you've never seen it before
09:37 submersible unclean interpreter state??
09:37 Alias_ 2. EVERYTHING is properly names
09:37 Alias_ Which means, K2::Controller::BlahBlah::Taco::Someth​ingElse::test::TemporaryClasses::Blah
09:37 Alias_ named
09:38 Alias_ None of this My::Foo simple stuff
09:38 Alias_ And now the names are so long you have to keep all the class names in variables
09:38 Alias_ And of course, nothing gets cleaned up at END time until the entire suite is done
09:38 Alias_ So database connections and stuff bleed
09:39 Alias_ And instead of just a long linear "story" spewed down a .t file, you need to break it up into chunks, and name those chunks as methods, and then track the test counts for every single chunk
09:40 Alias_ And instead of just my $commonthing = foo; you have to have specialised my $commonthing; sub setup_method: blah { $commonthing =  }
09:40 Alias_ I just ripped all the Test::Class code out of Aspect.pm and moved it to Test::More and it's about half the size
09:40 Alias_ And I didn't touch the actual test code
09:40 Alias_ Just removed boilerplate
09:41 Alias_ It all definitely feels like write-once test code
09:41 Alias_ Oh, and the REALLY crazy dirty tricks are out of bounds
09:42 Alias_ So in test code, you can't cheat or be evil
09:42 Alias_ Which is half the damned point
09:43 Alias_ YES, IT'S THAT BAD!
09:43 Alias_ (that you had to leave) :)
09:46 Alias_ I guess I should just experiment over the next two days
09:46 Alias_ Maybe I can get the regexp context thing working
09:46 Ryan52 anybody know offhand if Padre 0.54 has any deps that 0.53 didn't?
09:47 Sewi Ryan52: At least Wx::Perl::ProcessStream 0.20 --> 0.24
09:47 Ryan52 ah, crap.
09:47 Ryan52 that ones easy tho :)
09:48 szabgab Ryan52, will you be able to put the Perl6 and Catalyst plugins into Debian?
09:48 Alias_ Ryan52: Quite a few updated versions
09:48 Alias_ Ryan52: SQLite, ORLite, Wx::Perl::ProcessStream
09:49 Alias_ Possible more
09:49 Alias_ I don't believe there's any actual new modules
09:49 Alias_ But I could be wrong
09:49 * Alias_ checks
09:50 szabgab Ryan52, and Smolder
09:50 Alias_ uh, what?
09:50 Alias_ That's a joke right? :)
09:50 Alias_ If you've already done the newer FindIndent etc from a few weeks back, that should be all
09:50 cognominal joined #padre
09:52 szabgab ctrl-M does not always work
09:52 Ryan52 szabgab: I'll work on the Perl6 plugin after I get rakudo into Debian (which will happen in February, hopefully).
09:52 szabgab good, though you know it does not depend on Rakudo, right?
09:53 Ryan52 oh, it doesn't?
09:53 szabgab nope
09:53 szabgab for editing you don't need Rakudo
09:53 Ryan52 editing perl6 code without a perl6 interpreter seems kind of silly..
09:53 szabgab only for execution
09:54 szabgab Rakudo is much more volatile than the plugin, people will
09:54 Alias_ The plugin does it's own installation though
09:54 Alias_ Or is that only on Windows?
09:54 szabgab want to use the latest home made rakudo
09:54 szabgab but will be ok with the plugin as it came with the OS
09:55 Ryan52 I have to package 8 things to get all of the plugins dependencies.
09:55 Ryan52 (for Perl6)
09:55 szabgab go Ryan52 go :-)
09:55 Alias_ Ah
09:55 Alias_ All of them hard, I imagine
09:57 Alias_ Ryan52, so there's 4 changed deps
09:57 Ryan52 what's the 4th?
09:57 Alias_ Text::FindIndent, DBD::SQLite, ORLite, Wx::Perl::ProcessStream
09:57 szabgab ok, here is a nice issue
09:57 Alias_ I'm farly certain that's it
09:57 szabgab if I try to comment out some lines in an apache conf file (that happens to be recognized as plain text file)
09:57 szabgab padre silently does not do anything
09:58 szabgab because the comment_lines_str method of the base Document class returns nothing
09:58 Ryan52 Alias_: okay, easy enough. thanks for looking
09:58 szabgab it is very annoying as I thought for some time now that someone broke the ctrl-m
09:58 szabgab so what would be the right solution here?
09:59 Alias_ Error message?
09:59 szabgab add # as the default comment character?
09:59 Alias_ "Cannot determine comment character for this document type"
09:59 szabgab error pop-up ?
09:59 szabgab I could do that but it will violate the string freeze of waxhead :-)
09:59 submersible joined #padre
10:00 Alias_ In that case, just do nothing
10:00 szabgab call die()
10:00 Alias_ Silent shortcut
10:00 Alias_ my $char = comment_lines_str or return;
10:00 szabgab I think I'll add the error message
10:00 szabgab that's what happening now, silence
10:00 szabgab frustrations
10:01 Alias_ You could do a system beep? :)
10:01 szabgab and I don't have so much hair any more
10:01 szabgab or add a non-translatable error pop-up
10:01 szabgab and then change it to translatable in the next version ?
10:01 Hyppolit svn: r10063 | Sewi++ | http://padre.perlide.org/trac/changeset/10063
10:01 Hyppolit Remove a debug output message
10:01 Hyppolit trunk/Padre/lib/Padre/Document/
10:02 Alias_ That sounds fine to me
10:02 Alias_ Or ignore the string freeze :)
10:02 Alias_ It's a rare error message after all
10:03 waxhead that's right.. no violating string freeze!  :)
10:04 waxhead is two days too long for a string freeze?
10:04 waxhead I'm sure not everyone has seen the email, so you could get in a few quick changes.
10:07 waxhead mind you, there are already 2 languages at 100%
10:08 Hyppolit svn: r10064 | szabgab++ | http://padre.perlide.org/trac/changeset/10064
10:08 Hyppolit add pop-up error message when the document type does not have a comment character type, szabgab will need to pay a lot a beer to the translators
10:08 Hyppolit trunk/Padre/lib/Padre/Wx/
10:10 szabgab and then View Document As... does not have a #@$@%@#   Apache Configuration file type
10:12 Hyppolit svn: r10065 | szabgab++ | http://padre.perlide.org/trac/changeset/10065
10:12 Hyppolit comment out previous change so the translators wont kill me
10:12 Hyppolit trunk/Padre/lib/Padre/Wx/
10:13 szabgab I think I'd better use a git repository for this time
10:25 submersible gah .
10:26 submersible help search looks so enticing ... but then it freezes me and ...GAH!
10:28 Hyppolit svn: r10066 | adamk++ | http://padre.perlide.org/trac/changeset/10066
10:28 Hyppolit Added missing "use 5.008005;"
10:28 Hyppolit trunk/Padre/lib/Padre/
10:35 awnstudio joined #padre
10:41 gshank joined #padre
10:56 Hyppolit svn: r10067 | adamk++ | http://padre.perlide.org/trac/changeset/10067
10:56 Hyppolit Adding a check for Bad->Method::names
10:56 Hyppolit trunk/Padre/xt/
10:57 * Alias_ ponders adding a badcode check for $Sewi_Style_Variables
11:06 daxim joined #padre
11:10 Hyppolit svn: r10068 | adamk++ | http://padre.perlide.org/trac/changeset/10068
11:10 Hyppolit Fixing badcode complaint
11:10 Hyppolit trunk/Padre/lib/Padre/Wx/
11:15 pece2 joined #padre
11:20 patspam joined #padre
11:21 Hyppolit svn: r10069 | adamk++ | http://padre.perlide.org/trac/changeset/10069
11:21 Hyppolit Removing weird usage of Perl. Bad code! BAD!!!
11:21 Hyppolit trunk/Padre/lib/Padre/
11:21 Hyppolit svn: r10070 | adamk++ | http://padre.perlide.org/trac/changeset/10070
11:21 Hyppolit Removing no longer needed note
11:21 Hyppolit trunk/Padre/lib/Padre/
11:25 waxhead night all...
11:30 Hyppolit svn: r10071 | adamk++ | http://padre.perlide.org/trac/changeset/10071
11:30 Hyppolit Still doesn't work, but at least it's legal
11:30 Hyppolit trunk/Padre/t/
11:39 szabgab joined #padre
11:53 Hyppolit svn: r10072 | adamk++ | http://padre.perlide.org/trac/changeset/10072
11:53 Hyppolit Add the --invisible flag
11:53 Hyppolit trunk/Padre/xt/
11:54 Alias_ Should we move 00-compile to xt?
11:54 Alias_ It's quite a slow test, and it probably doesn't matter if we don't load EVERY single module from scratch one at a time
11:55 szabgab_ joined #padre
12:09 cognominal joined #padre
13:01 Hyppolit #811: possibly copyrighted work without a license (share/languages/perl5/perlapi_current.yml) (assigned defect) [ http://padre.perlide.org/trac/ticket/811#comment:1 ]
13:27 spacechicken joined #padre
13:27 spacechicken left #padre
14:13 daxim joined #padre
14:42 CromeDome joined #padre
14:59 bruck joined #padre
15:06 rindolf joined #padre
15:08 rindolf Hi all.
15:09 rindolf szabgab: hi! the Planet FOSS-IL entry for http://szabgab.com/blog/2010/01/1262671068.html displays the HTML markup instead of a formatted document.
15:09 szabgab thanks, checking
15:10 rindolf szabgab: it also seems to be with your entire food at home.
15:11 szabgab what do you mean?
15:14 szabgab rindolf, ?
15:16 rindolf szabgab: http://www.szabgab.com/blog/szabgab.rss
15:16 rindolf This one.
15:16 szabgab chck again
15:16 rindolf At least in Akregator.
15:16 rindolf szabgab: what?
15:16 szabgab check it again pls
15:16 szabgab I stopped the CGI::escapeHTML on the feed
15:16 rindolf szabgab: OK, now it's OK.
15:17 szabgab I don't remember why and when did I add it
15:40 rindolf szabgab: don't you have version control?
15:41 szabgab yes, but sometimes I make emergency changes on the server and forget to move the code back to the version control, besides I just moved to a new server so I might have broken something then
15:44 mib_stoubi joined #padre
15:55 dapatrick joined #padre
15:55 rindolf szabgab: I always keep the source in the version control and upload it to the server using rsync.
16:03 zamolxes joined #padre
16:19 rindolf szabgab: and while you're in the neighbourhood please set $entry->author("Gabor Szabo");
16:39 siegfri3d joined #padre
16:52 Alias_ joined #padre
17:05 francisturner joined #padre
17:05 Ryan52 tsee has a bunch of modules that are very closely linked with Padre (and that's the reason they exist), and yet they aren't in SVN. meh.
17:09 Ryan52 szabgab, Sewi, Alias_, other-smart-ppl: do you know why ./share/languages/perl5/perlapi_current.yml is in Padre instead of Padre-Plugin-XS? It's only used by the plugin. and we in Debian have to remove it and change the code that uses it to generate based on the system installed version. the easiest way to do this would be to simply change the code that would use the .yml file rather than worrying about generating it at build. but
17:09 Ryan52 so what I would propose is to move the .yml file to the plugin.
17:10 Ryan52 I'll ask tsee about it before I do it, but does anybody know it wasn't done like this originally?
17:16 Alias_ Nope
17:16 Alias_ Check svn log
17:16 Alias_ Ryan52: tsee's modules are in my repo
17:16 Alias_ svn.ali.as
17:16 Alias_ As are all MY modules linked with Padre
17:16 Alias_ Repo has similar rules to Padre one, anyone vaguely competant can just have commit
17:18 Ryan52 Alias_: how do I get a commit bit?
17:18 Alias_ What's your cpanid?
17:18 Ryan52 RSN
17:18 Alias_ And now you have commit
17:19 Alias_ Email will send itself to you shortly
17:19 Alias_ Don't check out the root, seriously
17:19 Alias_ Start at http://svn.ali.as/cpan/trunk
17:19 Alias_ The root is about a 1.5 gig checkout
17:27 daxim I pulled that, and the complete history
17:42 francisturner left #padre
18:00 daxim I want to attach a 23 MB file to trac, but I just noticed the limit ist 256 kB
18:00 daxim please review http://padre.perlide.org/trac/wiki/GitMigration
18:00 daxim how do I share my result with yall?
18:00 Sewi Upload it somewhere and paste the link
18:01 Ryan52 daxim: can be annoying and ask you to please change my email address to ryanryan52@gmail.com
18:02 daxim changed, will sync to wiki later
18:02 Ryan52 thanks
18:02 daxim Sewi, where do I upload?
18:05 Sewi Up to you. Use any webserver you like, if you don't have one, last resort would be rapidshare or others
18:05 daxim I feared so
18:06 daxim then I have to delay until this evening when I come home and restart the computer
18:06 daxim we had a power outage today
18:06 Sewi Do we already have a decision on killing svn and moving to git?
18:07 daxim no, but this result serves as feasibility study
18:07 daxim I'm a big proponent of git
18:07 daxim and github
18:07 Sewi If there any study about svn vs git clients installed on people's computers? I'd guess that much more people have svn...
18:08 daxim well duh, confirmation bias much?
18:08 daxim if padre used hg, then most people already would use a hg client...
18:10 Alias_ The problem with git is that the Windows clients still suck
18:11 daxim really?  tell me more
18:11 Alias_ And svn is very much a lowest common denominator platform, it makes it easy to get the code
18:11 Alias_ The recommended way to use git is via cygwin still I think
18:12 Alias_ Which is incompatible with strawberry
18:12 Alias_ There's various ports and clean ups underway
18:12 daxim aha, so you're actually just not up-to-date
18:12 Alias_ But they aren't really mature yet
18:12 Alias_ Alternatively, you can use the command line native ones yes?
18:12 Alias_ Which nobody on Windows has any fucking idea about, because nobody uses the command line
18:13 Alias_ There's also the fact that git people can use svn, but svn people can't use git
18:13 daxim http://code.google.com/p/tortoisegit/
18:13 Alias_ ugh
18:13 Alias_ shell integration is a horrible half-ass way to use version control
18:13 daxim alright, what do you want, then?=
18:14 Alias_ What causes problems with using git-svn?
18:14 daxim I don't understand that question
18:14 Alias_ OK, so 40% of all programmers primarily use svn
18:14 Alias_ And a fairly small percentage are using git
18:15 Alias_ And the git people, like jq demonstrated, can clone off the svn repo, do their thing, then commit back, yes?
18:15 daxim yes
18:15 Alias_ And if we do it the opposite way around, can the svn people checkout of somewhere and commit back to git?
18:15 Alias_ Using normal editor-integrated stuff, and proper gui clients
18:16 daxim in principle yes, but it's a head-ache because git is richer than svn
18:16 Alias_ And everyone can continue to commit to trunk, without someone being forced to pull 50 patches a day?
18:16 Alias_ Presumably more if git allows more granular changes
18:17 daxim do you think that that it is a problem to pull 50 changesets?
18:17 Alias_ Every day
18:18 Alias_ You go away for the weekend, and on monday that's 150
18:18 daxim and?
18:18 Alias_ Would you volunteer to pull everything, and manage all the clashes?
18:18 daxim what clashes?
18:19 Alias_ Well, if everyone is cloning trunk and working independantly, there's going to be patch delays right
18:19 Alias_ Unlike now when there's pretty much instant syncing
18:19 PerlJam Alias_: not that you need validation, but I think you have things the right way around.  Keep the svn repo and it's easy for both git users and svn users to contribute, switching the repo to git will mess that up.
18:19 daxim Alias_, I see you're still very much in the svn mindset
18:19 PerlJam (and I'm an avid git user/fan)
18:19 Alias_ daxim: Don't get me wrong, I like the IDEA of git
18:20 daxim to make that more clear, there are no such clashes
18:20 daxim PerlJam, don't you think that github will increase the network effect to collaboration immensely?
18:20 Alias_ So lets assume you are pulling these 50 or so patches a day
18:20 Alias_ And Sewi and I change the same thing without knowing about it
18:21 Alias_ And you are gone for the weekend
18:21 Alias_ And on Monday, there's 200 patches in your queue
18:21 Alias_ And somewhere in all that Sewi and I and someone else changed the same thing without knowing about it
18:21 Alias_ What happens
18:22 PerlJam daxim: not if there's a significant number of svn people contributing already.
18:22 daxim at that time, the changes are still local to each repo, right?
18:22 PerlJam daxim: switching to github will effectively shut them out unless they learn git.
18:23 daxim indeed, that's the only big downside I see, and this needs to be weighed against the benefits
18:23 Alias_ daxim: Sure, they are all local to each repo
18:24 daxim alright, now the interesting thing comes.  assume we have github as the central/authoritative repo
18:24 daxim you finish your changes, rebase from github, then push out your changesets
18:24 Alias_ Wait, what?
18:24 daxim sewi finishes, rebases, resolves the merge conflict, then pushes out
18:25 Alias_ Define some terms in there
18:25 Alias_ I've been trickling changes all weekend
18:25 daxim what's trickling now?
18:25 Alias_ I'm finishing changes every couple of hours
18:26 Alias_ Over a weekend
18:26 Alias_ I change maybe a dozen various bits and pieces, each a standalone change
18:26 daxim alright, so you can rebase any time you want
18:26 Alias_ Which means what
18:26 Alias_ Something similar to svn update?
18:26 daxim yes, it avoids merge conflicts
18:26 daxim svn has update before commit
18:27 Alias_ Oh wait, so our analogy is already broken
18:27 daxim git has rebase before commit
18:27 Alias_ But continue
18:27 Alias_ Now define "push out your changesets"
18:27 daxim push is git push
18:27 Alias_ Which does waht
18:27 Alias_ what
18:28 daxim transfer changesets from one repo to another (github), like svn commit
18:28 Alias_ And when I do git push, everyone else will get the changes
18:28 Alias_ So now instead of save, save, save, update, commit
18:28 daxim not immediately
18:28 Alias_ it's commit, commit, commit, rebase, push
18:29 Alias_ Effectively..
18:29 daxim and on my side then pull or better pull --rebase
18:29 daxim that's totally analogous to the current svn model
18:29 Alias_ And push changes trunk?
18:29 PerlJam daxim: and why do you say that rebase is better?
18:29 Alias_ Which everyone rebases from?
18:30 daxim yes, push would change trunk.  I'll come back to that a bit later
18:30 daxim PerlJam, read http://blog.woobling.org/2009/05/g​it-rebase-considered-awesome.html
18:31 daxim realistically, on github, everyone would push to his own branch and then give the maintainer of the authoritative repo a pull request
18:32 PerlJam daxim: thanks  (I don't often rebase.  Can you tell? :)
18:32 daxim who then pulls in the changesets
18:32 daxim I meant to say: push to his own *fork*
18:33 Alias_ Wouldn't that mean that if you, as the volunteer for these 50 pull requests a day, are critical?
18:33 Alias_ That nobody else gets anyone else's changes until you process the pull requests?
18:33 Alias_ s/if//
18:33 Alias_ And if you get hit by a bus, developmen basically stops?
18:34 daxim that's not true, because if szabgab falls sick and can't do it, and sewi has already pushed his changesets to his fork, then I can pull from sewi for a while
18:34 daxim distributed, remember?
18:34 Alias_ daxim: We have 25 people commiting monthly
18:34 Alias_ How do I know who they all are this month
18:35 daxim or again, you can use the svn model and trust allow them to push directly into the authoritative repo
18:35 PerlJam daxim: ah ... this line makes me happy "If you run git pull followed by git pull --rebase the merge commit created by the first pull will be omitted from the results."
18:35 Alias_ At the pace we move, that is pretty much essential
18:35 Alias_ 8 hours while a central maintainer is sleeping is along time
18:35 PerlJam daxim: that's the only annoyance I've had with merging in git
18:36 daxim oh boy, if that already annoys you you should try merging something complicated in svn :/   you'll pull out your hair
18:36 Alias_ It's easy
18:36 daxim anyway, rebase avoids merges and creates a linear history
18:36 Alias_ You check out one branch, you check out another brance, and you launch a proper merge tool :)
18:37 PerlJam daxim: I've done complicated merges in svn.  That's why I dropped it so quick when I found out about git.  :)
18:37 daxim alright
18:37 * Alias_ uses http://www.araxis.com/merge/ for merging with svn
18:38 daxim I use git mergetool (which has a built-in list) and then it launches kdiff3 which rules supreme
18:38 Alias_ I hate this idea that the VC tool will just smash code together and only tell me about the files that DON'T wortk
18:38 Alias_ Araxis is better than kdiff3 :)
18:38 Alias_ I've used both
18:39 Alias_ It is, however, Windows only
18:39 Alias_ And commercial
18:39 Alias_ Which means I don't bother promoting it much
18:39 daxim any more git discussion?
18:39 Alias_ Not really
18:39 Alias_ Sounds like we could fairly easily replicate the current setup
18:40 daxim indeed
18:40 Alias_ I'd just lose all the useful gui-heavy tooling
18:40 Alias_ And have to revert to command line or shell integration
18:40 daxim try out tortoise, it's more than that
18:40 Alias_ And I'm not sure yet what we'd gain
18:40 Alias_ I have, and I agree it's a bit better
18:40 PerlJam Alias_: there's a reason parrot is still using svn instead of git (though it's come up several times)
18:41 daxim a much increased ease of collaboration, and if that isn't worth it, then I don't know ;)
18:41 Alias_ oh?
18:41 Alias_ So far, we get the same thing, just with more commands needed
18:41 Alias_ We give out commit to anyone that asks, so that's not a bit deal
18:42 daxim ah, but see, if it's on github
18:42 daxim a developer doesn't need to ask!
18:42 Alias_ Sure he does
18:42 daxim nay
18:42 Alias_ The changes just sit in his repo
18:42 Alias_ And he eventually asks via a pull request
18:43 daxim yes
18:43 Alias_ Unless someone magically knows he exists
18:43 Alias_ And someone like you, with a git-svn, can accept those pulls, right?
18:43 daxim it's the pull request thing.  like creative commons, it changes the model from "may I?" to "go right ahead"
18:43 PerlJam Alias_: I'm perlpilot on github (for that magic future so that I don't have to ask  ;)
18:44 Alias_ may you want?
18:44 Alias_ "May I add code to trunk" is needed in both cases
18:44 Alias_ It's just via a different mechanism
18:44 Alias_ One being once per user, the other being once per change
18:45 daxim and that exactly attracts one-off patchers
18:45 Alias_ Because sending patches with svn checkouts is too hard?
18:45 daxim in fact it is
18:45 Alias_ Right!
18:46 Alias_ Mental Note: Add "send diff to project author" automation to Padre
18:46 daxim even giving out commit bits liberally like you do here and on svn.ali.as is still worse
18:46 Alias_ (We actually already have at least half the stuff needed to make that work)
18:47 Alias_ daxim: I'm still waiting for study results on that one
18:47 Alias_ Although anecdotal evidence seems to point to the people that DO make those one-off changes liking it more
18:48 Alias_ Still waiting for metrics on if it actually results in improved volumes of change
18:48 Alias_ Anyways
18:48 Alias_ Padre will almost certainly eventually get big enough that git is needed
18:49 bruck joined #padre
18:49 Alias_ But on the balance, I don't see we're there yet
18:49 daxim ah bugger
18:49 Alias_ But in the mean time, it makes sense to make sending in one-off patches much easier
18:49 daxim okay, I go home now and upload the result repo
18:49 Alias_ So that at least we remove the biggest problem
18:49 Alias_ daxim: Proof of concept repo absolutely can't hurt though
18:50 daxim please have a look and get a feel for it when I publish the link
18:50 Alias_ It keeps the idea moving forwards
18:50 Alias_ 5:50am, I should sleep
18:52 szbalint or you should wake up :)
19:12 jq Alias_: ping
19:52 robn joined #padre
20:29 code4pay joined #padre
20:54 cognominal joined #padre
21:03 daxim joined #padre
21:23 CSJewell joined #padre
23:05 cognominal joined #padre

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