Camelia, the Perl 6 bug

IRC log for #padre, 2010-08-18

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

All times shown according to UTC.

Time Nick Message
00:29 Hyppolit svn: r12277 | submersible_toaster++ | http://padre.perlide.org/trac/changeset/12277
00:29 Hyppolit [Swarm] migrate to new task API
00:29 Hyppolit trunk/Padre-Plugin-Swarm/lib/Pad​re/Plugin/Swarm/Transport/Local/ trunk/Padre-Plugin-Swarm/lib/Padre/Pl​ugin/Swarm/Transport/Local/Multicast/
00:39 Alias oooh, neato
00:39 Alias Was it hard?
00:39 * Alias heads over the road
00:58 jazzanova joined #padre
01:11 PerlJam joined #padre
03:21 ppant joined #padre
03:23 ppant_ joined #padre
03:23 ppant left #padre
04:00 Sno| joined #padre
05:08 [Sno] joined #padre
05:28 kaare joined #padre
05:49 danlucraft joined #padre
05:55 pece joined #padre
07:05 waxhead Sewi, the file changed dialog still seems broken
07:05 waxhead I deleted a file and when I said "reload", it crashed Padre
07:06 mattp_ joined #padre
07:17 danlucraft joined #padre
07:17 Steffen joined #padre
07:27 yorfeix joined #padre
07:43 danlucraft joined #padre
07:49 Alias_ We REALLY need to fix that shit
07:50 Steffen Context for a late comer?
07:50 Steffen Ah, nevermind. Logs in topic.
07:51 Sewi waxhead: Which OS? Padre usually detects deleted files here...
08:00 Sewi waxhead: Confirmed, my latest fixes to the dialog are gone and there is a new crash while reloading
08:06 Hyppolit svn: r12278 | Sewi++ | http://padre.perlide.org/trac/changeset/12278
08:06 Hyppolit Reload-on-change crash fix
08:06 Hyppolit trunk/Padre/lib/Padre/Wx/
08:08 Alias_ We should switch back to file at a time
08:09 Alias_ I wonder if we could dig out the old code
08:09 Sewi Alias: Please don't, it's bad if you use svn or something like this.
08:10 Sewi svn update and then make Padre reload each file individually
08:10 yorfeix hello
08:10 Sewi hi yorfeix
08:11 Sewi Alias: Make it a preference if you want
08:11 Alias_ Sewi: It's worse if you use the thing we have now
08:11 Alias_ And it's pretty easy to deal with one at a time
08:11 Sewi I'm using it and it works fine (usually)
08:11 Alias_ You wait until you actually focus on a changed file
08:12 Alias_ Then you ask "Update to disk?"
08:12 Alias_ And you can answer "Yes", "No", "Yes to all" etc
08:12 Sewi And you don't know if it was changed in the editor.
08:12 Alias_ If you aren't focused on a file, you don't ask the question
08:12 Alias_ Different question
08:12 Alias_ There's four cases here
08:13 Sewi And "Yes to all" kills all changes to other files
08:13 Alias_ 1. Padre == disk ( Update from disk, yes, no, all, none )
08:13 yorfeix How can I change the project path in the left pane? (padre 0.69; win7)
08:13 Alias_ 2. Padre != disk ( Update or keep?, update, keep, update all, keep all )
08:13 Sewi yorfeix: Add a Makefile.PL, or padre.yml to the project dir.
08:13 Alias_ 3. Deleted from disk ( Close or keep open? Close, keep, close all, keep all )
08:14 Sewi Alias: Seems we don't agree on this topic, so make it a preference and everybody will be happy.
08:14 Alias_ False logic
08:14 Alias_ If we make it a preference, Padre bloats and slows
08:14 Alias_ Especially if the implementations are completely different
08:15 Alias_ The current method isn't even logically valid
08:15 Sewi There have been people on the mailinglist clearly voting against the one-file solution.
08:16 Alias_ I'm sure that's just because they don't know what they are voting against
08:16 Sewi If you want it, you need to create a preference option.
08:16 Alias_ If I have one file of each of update, changed, and deleted, the dialog is completely invalid
08:16 Alias_ There's no option for keep, there's no option for close
08:16 Sewi It would add just one if to the procedure, because change-detection of other files isn't done in Main.pm
08:16 Alias_ You can't mix choices
08:16 Alias_ You have to answer for every single file in advance
08:16 Alias_ It doesn't close
08:16 Alias_ You close it, it just pops up again a second later
08:17 Alias_ Whether or not a single dialog is conceptually valid, the current version is a hyper-obtrusive fail
08:17 Sewi This is a bug, but no reason to remove the whole thing people like
08:17 Sewi Sorry, but I don't agree. It's working with small bugs - like many things in Padre.
08:19 Alias_ If there's gunna be an argument over this, maybe I should delay fixing it then
08:19 yorfeix I changed the path in "Tools -> Preferences -> Bahaviour -> Default projects directory" but nothing changed.
08:19 Alias_ And just update the file scanning code instead
08:19 Alias_ yorfeix: Wrong setting
08:19 Alias_ (For now)
08:20 Alias_ That's just there Padre looks for projects
08:20 Alias_ Not the default directory tree root
08:20 Sewi yorfeix: The project dir detection is automatic (at the moment), it looks for a typical project-root file (like Makefile.PL or padre.yml and others)
08:20 Alias_ Tools -> Preferences -> Advanced -> main_directory_root
08:21 Alias_ Sewi: He means I think the default tree
08:21 Alias_ When there's no active project
08:21 Sewi ok
08:21 * Alias_ really wishes that we'd never merged the ideas of directory tree and project tree
08:21 Alias_ But oh well
08:22 Alias_ I'll push on the with improved tasks for the directory tree
08:22 yorfeix thanks Alias_
08:22 Sewi We need to make Padre more lightweight in terms of memory
08:22 Alias_ Sewi: It's as light as it's going to get for now
08:22 Alias_ 100meg is reasonable
08:22 Alias_ And it's holding bloat beyond that point to a reasonable level now
08:23 Alias_ We can't improve much more until we get an improved Wx bootstrapper
08:23 Alias_ So that all of Wx isn't being loaded in the background threads
08:23 Alias_ We already load as much of the gui code as we can on the fly
08:23 Sewi It was something I noticed while looking at my workstation, nothing really new or burning :-)
08:24 Alias_ Trust me, I've already found all the easy wins :)
08:24 Alias_ And a few of the hard ones
08:24 Sewi So you want to go to the file scanning code now?
08:24 Alias_ The only ones left are the bloody high fruit
08:24 Alias_ I want to reimplment it
08:24 Alias_ We should implement the file scanning as a task
08:24 Alias_ Which updates the document objects
08:24 Alias_ To flag "changed" or "deleted"
08:25 Sewi Oh, I though you were talking about the Directory browser scanning...
08:25 Alias_ oh, I was
08:25 Alias_ But WRt that one, see the new Padre::Directory::Search
08:25 Alias_ That should do the incremental search
08:25 Sewi Do you see STDERR of your Padre?
08:25 Alias_ Normally?
08:25 Alias_ No
08:25 Alias_ I run from a clicky icon
08:25 Alias_ So no console output
08:26 Alias_ I only see it when I run perl dev.pl
08:27 Sewi I'd like to commit a small debug output fix which shows the directory browser scanner runs.
08:27 Alias_ hrm
08:27 Sewi It feels like there are too many scans, but I don't know where they come from.
08:27 Alias_ I have a better option
08:27 Alias_ Wanna see something funky?
08:28 Alias_ Oh wait, you want the source of the task calls?
08:28 Alias_ If you just want to see what the scan is looking at, at any given point in time, use state messages
08:28 Alias_ In the task ->run method put the following
08:28 Sewi It's scanning while there is no reason to do so.
08:28 Alias_ $self->handler->message( STATUS => $directory->unix )
08:29 Alias_ Ah, you want to see where it's commissioned?
08:29 Alias_ It should already be there
08:29 Sewi I don't know if I understand enough of your work to change something anyway.
08:29 Alias_ oh wait, or not
08:30 Alias_ What is your goal?
08:30 Sewi But it starts scanning processes while in background even if there is no action at all.
08:30 Alias_ To find where the refresh calls are being triggered from?
08:30 Alias_ Try this
08:30 Alias_ perl dev.pl -d
08:30 Alias_ use Padre::Wx::Directory::Task;
08:31 Alias_ b Padre::Wx::Directory::Task::new
08:31 Alias_ c
08:32 Alias_ The only place in the code they get made is in ::Directory line 261ish
08:32 Sewi Cool, I din't know that the debugger could do this :)
08:32 Alias_ As part of ::Directory->refresh
08:32 Alias_ To my knowledge, it only gets called three times
08:32 Alias_ Actually, twice
08:33 Alias_ Once when you open a file in a project, against if you close one
08:33 Alias_ oh
08:33 Alias_ Third if you change file tabs, and those tabs are in different projects
08:33 Alias_ If you change tabs, and the files are in the same project, it shouldn't refresh
08:33 Sewi It does
08:34 Alias_ You sure?
08:34 Sewi The debugger stopped, but I'll add some output and verify this.
08:34 Alias_ T
08:34 Alias_ gives you the stack trace
08:35 Alias_ I think I'm going to dump some TRACE hooks into the ::Directory module anyway
08:35 Alias_ It's a major load focus point
08:40 Alias_ I've started to document the task API now it's getting more complete
08:41 Sewi Wx/Directory.pm line 260: Shouldn't this block go into the unless { } before?
08:41 Alias_ Not much done yet though
08:42 Alias_ Which line?
08:42 Alias_ (content)
08:42 Alias_ I've made some changes
08:42 Alias_ You mean should the task_request go into the unless?
08:42 Sewi # Trigger the refersh task...
08:43 Alias_ Weird... quite possibly
08:43 Sewi ok
08:43 Alias_ you're right, it does rescan
08:44 Alias_ I wonder if I was doing it to pick up any new files added to the project
08:44 Alias_ I've had that problem quite a bit
08:44 Alias_ You svn update, and the directory tree doesn't reflect it
08:44 Hyppolit svn: r12279 | Sewi++ | http://padre.perlide.org/trac/changeset/12279
08:44 Hyppolit Trigger a directory refresh only if the root dir changes (was a typo: } at wrong line)
08:44 Hyppolit trunk/Padre/lib/Padre/Wx/
08:44 Alias_ But that might be a teeeensy bit expensive under the current super-heavy scans
08:45 Alias_ What we probably want is to re-enable it later, once we're only rescanning the parts of the tree that are visible
08:45 Alias_ Or to switch to doing it periodically or something
08:45 Alias_ There's some other sub-optimal stuff to
08:45 Alias_ Right now, you can't QUITE cancel on-the-fly tasks that are no longer useful
08:45 Sewi We should trigger a new scan of a directory after a save-as or save-unsaved call but withour sub-dirs
08:46 Alias_ The scan code supports it
08:46 Alias_ Oh wait
08:46 Alias_ No it doesn't, it only supports it in the implementation in my head :)
08:46 Alias_ Yet another feature I discarded to get it onto trunk before it went stale
08:47 Sewi No problem, so just add a call to Alias::Head::Padre::Wx::Directory::Task :-)
08:47 Alias_ :)
08:47 Alias_ My old PPI talk describes three main versions of Perl the language
08:47 Alias_ The one the interpreter understands, the one PPI understands, and the one inside Larry's head
08:48 Alias_ And both of the former try to approach the latter independantly
08:48 Sewi :-)
08:48 Alias_ There's also other versions in every single other developer's heads, but those aren't as important
08:50 Alias_ Anyways
08:50 Alias_ So here's the plan from here
08:50 Alias_ Under the old incremental scan model, the directory search was useless
08:50 Alias_ Under the new model, it's usable now but the cost of feeding it is too high
08:51 Alias_ So we need to
08:51 Alias_ 1. Re-implement the directory tree search mode to search incrementally, and display the results incrementally
08:51 Alias_ (See Komodo with a big project open for what it looks like when it's done right)
08:52 Alias_ With search working properly with it's own dedicated ::Search task, that lets us move the normal tree back to opportunistic shallow scanning
08:53 Alias_ I've cut and pasted ::Task to ::Search to create what I think it a fairly reasonable task
08:53 Alias_ Now we need to change the search code in ::Directory and ::TreeCtrl to support the incremental messages flowing from that task
08:55 Sewi That last commit removed many scan tasks.
08:55 Alias_ Some of which involves some final tweaks to the task model to get a couple of things like abort messages working
08:55 Alias_ I bet it did :)
08:55 Alias_ Probably stopped the directory tree snapping back to the top
08:55 Alias_ If you scrolled down
08:56 Hyppolit svn: r12280 | adamk++ | http://padre.perlide.org/trac/changeset/12280
08:56 Hyppolit Adding TRACE hooks for the main methods in the directory tree
08:56 Hyppolit trunk/Padre/lib/Padre/Wx/
08:56 Sewi Maybe, will try this later
08:56 Hyppolit svn: r12281 | adamk++ | http://padre.perlide.org/trac/changeset/12281
08:56 Hyppolit Adding TRACE hooks for the task owner role
08:56 Hyppolit trunk/Padre/lib/Padre/Role/
08:57 Sewi hmm, seems I found that reload-window-comes-back - bug. It was fixed long long ago but there seems to be a mind error in the fix
08:59 Alias_ Fix massively welcome
09:03 Alias joined #padre
09:07 Hyppolit svn: r12282 | adamk++ | http://padre.perlide.org/trac/changeset/12282
09:07 Hyppolit Dev, test and prod instances of Padre use different single instance ports, so they don't accidentally call the wrong Padre.
09:07 Hyppolit trunk/Padre/lib/Padre/
09:09 Alias_ grr
09:09 Alias_ OK, we need to change our versioning scheme
09:09 Alias_ I need to break API compatibility
09:09 Alias_ And I can't
09:10 Alias_ If I break the API, but don't update $COMPATIBLE, I'll forget and some plugin that needs tasks won't get disabled properly
09:10 Alias_ If I update it to 0.70, then it won't load the plugins that needs tasks at all, because the version won't exist
09:10 Alias_ If I update it to 0.69, it's still wrong and I get the same forgetfulness problem anyway
09:11 Alias_ Actually
09:11 Alias_ I wonder what happens if I update it to 0.70
09:11 Alias_ (for the actual code, not just the way I intended it to work)
09:11 Alias_ hrm, no won't work
09:11 Hyppolit svn: r12283 | Sewi++ | http://padre.perlide.org/trac/changeset/12283
09:11 Hyppolit Show the reload-dialog only once per change, update Changes
09:11 Hyppolit trunk/Padre/ trunk/Padre/lib/Padre/ trunk/Padre/lib/Padre/Wx/
09:12 Alias_ oh wait, I don't need to change Task
09:12 Alias_ Just the role
09:14 Alias_ Anyone remember why we don't bump the version at the beginning of the development cycle?
09:14 Alias_ (I think we probably need to move to even/odd)
09:14 Alias_ (If we can't handle bump in advance)
09:15 Alias_ OK, screw it, I'm just incompatibly changing the API thne
09:15 Sewi What about introducing a plugin-api-version which doesn't change with the main one?
09:16 Alias_ doesnt' work
09:16 Alias_ We let plugins touch any part of the Padre internals directly
09:16 Alias_ So plugins declare which parts of Padre they are going to be touching, and what the last version of Padre was when they know the use of that part of Padre worked
09:17 Alias_ And in Padre, each module that plugins are going to be calling specify a $BACKCOMPATIBLE as well as $VERSION
09:18 Alias_ When we load the plugin, we take each declared interface $needed and check that $BACKCOMPATIBLE <= $needed <= $VERSION
09:18 Alias_ If every required class fits inside the compat range, the plugin as a whole is compat
09:18 Alias_ This way, if I break Padre::Task only the plugins that use tasks will be disabled
09:18 Alias_ The rest keep working
09:19 Alias_ For something like Firefox, every plugin stops working for each major release
09:19 Alias_ And each plugin author has to decertify that it works again
09:19 Alias_ For us, that's too much work
09:19 Alias_ Hence the more incremental approach
09:19 Alias_ It's not perfect, but it's enough for now
09:20 Alias_ Pending a tenfold increase in plugin maintainers :)
09:24 Alias_ Sewi: I'm about to drop a big chunk of task rejiggering, you aren't changing task_request type code are you?
09:26 Hyppolit svn: r12284 | adamk++ | http://padre.perlide.org/trac/changeset/12284
09:26 Hyppolit Moved task_response to task_finish so it will be symmetric with task_message
09:26 Hyppolit trunk/Padre/lib/Padre/ trunk/Padre/lib/Padre/Role/ trunk/Padre/lib/Padre/Wx/ trunk/Padre/lib/Padre/Wx/Dialog/
09:31 Steffen Alias_, since you seem to care about Class::XSAccessor. I just uploaded an experimental version (1.07_01) that implements efficient lvalue accessors:
09:31 Steffen $foo->bar = 12
09:32 Sewi Alias: All changes are committed
09:32 Steffen Very experimental, but apparently just as fast as the regular C::XSA accessors for setting and slightly slower for getting the attributes.
09:45 Alias_ Steffen: Actually, I really don't care at all about lvalues :)
09:45 Alias_ I think they are evil and an unnecesary feature
09:45 Steffen Yeah, I was expecting that answer.
09:45 Alias_ I wish they'd get removed from the language, shortly followed by ties :)
09:45 Alias_ Especially if removing lvalues would make methods faster
09:45 Steffen I don't care much for them either, but there was a feature request and I used the opportunity to learn more about Perl magic.
09:45 Sewi Alias: You don't like $foo = 1; ? :-)
09:45 Alias_ (But I'm not sure if that would be the case)
09:46 Steffen Me neither. They get compiled differently.
09:46 Alias_ Sewi: Quite you
09:46 Alias_ I wonder if anyone still uses them
09:46 Alias_ They were mostly a fad weren't they?
09:46 Steffen http://bugs.debian.org/cgi-b​in/bugreport.cgi?bug=536097
09:47 Steffen I do wonder why anyone would submit a feature request to a distribution bug tracker.
09:47 Steffen That's pretty retarded.
09:47 Alias_ Because that's the way they think of the world
09:47 Steffen What I said.
09:47 Alias_ Because it's easier than doing it via rt.cpan
09:47 Alias_ who knows...
09:47 Steffen Anyway, the good debian people forwarded it.
09:48 Alias_ Yup
09:48 Steffen debian++
09:48 Steffen And (debian-perl)++ # cool gang
09:49 Alias_ Steffen: Also, I've got most of bidirectional communication with tasks working now
09:49 Alias_ Mostly just tweaking for API for style and legibility
09:49 Alias_ Swarm has been ported over
09:49 Steffen I saw that, yes.
09:50 Alias_ Bizarely, on submersible's machine he had an old copy of Padre installed
09:50 Alias_ And he installed 0.69 over the top of it
09:50 Alias_ Then he loaded Swarm $old, and it just started up and worked
09:50 Alias_ Because it loaded both the new modules, and a couple of old modules
09:50 Steffen That is slightly bizarre, yes.
09:50 Alias_ And somehow it all contrived to work
09:50 Steffen Did you rip out the old tasks yet?
09:50 Alias_ Using half the new API and half the old API
09:50 Alias_ Yep
09:51 Steffen Hmm.
09:51 Alias_ The old task code is gone
09:51 Alias_ But ::Service was there from the old tree on his disk
09:51 Alias_ And that implemented it's own private upwards messaging remember
09:51 Alias_ We were both rather impressed it worked
09:51 Alias_ Because it REALLY didn't deserve to
09:52 Steffen Heh. Agreed.
09:52 Alias_ Downside, we because main_thread_only doesn't get special cased any more, it may have been serialising half the editor down into the children :)
09:53 Steffen Err.
09:53 Steffen Apart from being slow, that *could* produce serious problems with some of the Wx stuff, couldn't it?
09:53 Alias_ I guess not if no code in the child tries to actually use it
09:53 Alias_ Because there's no Wx loop running in the child
09:54 Alias_ Didn't seem to cause any problem though :)
09:54 Steffen *If* Mattia got all the thread-safety right for all of the Wx stuff.
09:55 Alias_ Anyways, I think I'm about to break swarm again by changing the task aPI
09:55 Alias_ So it needs another week before it's worth playing again
09:55 Alias_ But he's got the global swarm server online again
09:55 Steffen Well, I don't have the time to get involved anyway. Unfortunately.
09:55 Alias_ Of course
09:55 Alias_ Just placing temptation in your path
09:56 Steffen Fair enough, worth a try!
10:07 hits joined #padre
10:31 Hyppolit svn: r12285 | adamk++ | http://padre.perlide.org/trac/changeset/12285
10:31 Hyppolit I think I've broken tasks badly, but I'm committing so I can go home and fix them. Apologies for temporary break.
10:31 Hyppolit trunk/Padre/lib/Padre/ trunk/Padre/lib/Padre/Document/
12:40 tomko joined #padre
13:09 sjn joined #padre
13:10 sjn heya
13:10 sjn anyone here using Padre for hacking on projects tracked by git?
13:20 Sewi plugin?
13:20 Hyppolit plugin is http://padre.perlide.org/trac/wiki/Plugins
13:20 Sewi The git plugin maintainer might do so.
13:21 sjn that plugin hasn't been updated in more than a year
13:21 sjn and last time a tried it (a few months back) it didn't even work
13:22 sjn (it's not the plugin I'm asking about, though)
13:23 kanishka joined #padre
13:30 dapatrick joined #padre
13:44 yorfeix joined #padre
14:00 Alias sjn: I'd LIKE to use git, but until someone integrates a pure perl client, I can't
14:00 Alias There is, however, now an actual PP client
14:00 Alias So maybe it's worth trying to update it
14:12 Hyppolit svn: r12286 | adamk++ | http://padre.perlide.org/trac/changeset/12286
14:12 Hyppolit Correcting an overlooked callback
14:12 Hyppolit trunk/Padre/lib/Padre/Wx/
14:14 sjn Alias: well, I've tried to use Padre with a Git-tracked project, and found that it can be rather annoying
14:14 sjn especially when switching branches and such
14:14 Alias You could make it better? :)
14:14 sjn :-/
14:15 sjn I'd love to do that, but it would be a big distraction :-/
14:17 sjn still, is there a way for the "check for file updates" feature to not ask for closing the files? (reloading would be ok)
14:18 Sewi simply svn up
14:19 * sjn uses git
14:24 Sewi For the Padre repository?
14:26 sjn no
14:27 * sjn suspects Sewi and sjn are talking about different things
14:27 Sewi Maybe, I replied to "16:18:05) sjn: still, is there a way for the "check for file updates" feature to not ask for closing the files? (reloading would be ok)"
14:28 sjn the feature I'm talking about there is the one that checks locally if any open files have been modified by something else
14:28 Alias__ joined #padre
14:28 kthakore sjn++
14:28 kthakore hi
14:28 kthakore hi Alias__
14:28 sjn a "svn up" wouldn't help there at all
14:28 sjn kthakore: heya
14:29 kthakore Alias__: ruoso has started work on a game framework :D
14:29 Alias__ cool
14:29 Sewi sjn: Sure, the file-changed-on-disk check was fixed yesterday and today :-)
14:29 sjn ah
14:29 kthakore Alias__: he likes to make things complicated though :p
14:30 kthakore Alias__: but then again I don't have to code it. :D
14:32 Sewi sjn: Sorry for being confusing :-)
14:34 Alias__ kthakore, bad...
14:34 Alias__ That kind of thing will catch up with you
14:34 kthakore Alias__: what?
14:34 Alias__ Complicated
14:34 kthakore Alias__: I know
14:35 kthakore Alias__: I will strip it down before he relases :D
14:36 Alias__ Tell him he needs to road test the APIs with 12 year olds
14:36 * sjn tries to figure out where the svn repository is mentioned on padre.perlide.org
14:37 Sewi svn?
14:37 Hyppolit svn is http://padre.perlide.org/tr​ac/wiki/DownloadRepository and also http://svn.perlide.org/padre/trunk/ and also http://padre.perlide.org/trac/browser
14:39 sjn is there a reason why the svn repos isn't mentioned (somewhere visible) on the main website?
14:39 yorfeix joined #padre
14:39 Sewi It's some kind hidden because svn might be unstable or non-working and we want "plain" users to use a running version.
14:40 Sewi It's linked from the download page (last link on the right side, I think)
14:40 sjn it's linked on the CPAN page, but I can't find it on the download page
14:41 Sewi download?
14:41 Hyppolit download is http://padre.perlide.org/download.html
14:41 sjn that's where I'm looking, yes
14:41 Sewi Sorry, it was the quickest way for me to go there :-)
14:42 sjn it's on the trac page
14:42 sjn ah well
14:42 sjn maybe it would be cool to mention "SVN" and "Repository" somewhere in the "Source Code" box? :)
14:42 Sewi It's linked from the "Source code" block and below "Other Linux and *BSD", two clicks from each link
14:43 sjn yep, it's linked
14:43 sjn but not with the words "repository" or "svn" nearby, which I expected :)
14:43 sjn it's ok though
14:44 mib_ydg7qp joined #padre
14:44 sjn ("source code download" in my mind ususally means "get out tarball here")
14:44 sjn our*
14:44 sjn Sewi: thanks anyway :)
14:44 mib_ydg7qp hey there, someone
14:45 Sewi hi
14:45 mib_ydg7qp why do I get Warning: Use of "-s" without parentheses is ambiguous at ..
14:45 mib_ydg7qp when I run this line of code
14:45 mib_ydg7qp my @smaller_than_1000 = grep {-s < 1000} @ARGV;
14:45 mib_ydg7qp in padre?
14:46 Sewi Is it a popup warning?
14:46 mib_ydg7qp no
14:46 mib_ydg7qp I hit F5 and I get that on the result page
14:48 Sewi You use warnings; I think...
14:48 mib_ydg7qp anyone???
14:51 sjn t/23_task_chain.t .......... 3/21 Can't call method "is_running" on an undefined value at /home/sjn/src/Padre/blib/lib/Padre/TaskThread.pm line 132.
14:52 sjn # Looks like you planned 21 tests but ran 12.
14:52 Sewi This is why it's called "dev trunk" :-)
14:53 sjn yay \o/
14:54 * sjn knows it's difficult to always keep trunk nice and tidy (and the tests happy)
14:54 sjn so that's why prod about failing tests here :)
14:54 Steffen sjn: Quoting Alias' commit from earlier today: "I think I've broken tasks badly, but I'm committing so I can go home and fix them. Apologies for temporary break."
14:55 Sewi You should go before  r12285
14:55 Hyppolit Changeset #12285 http://padre.perlide.org/trac/changeset/12285
14:55 sjn ok
15:11 gabiruh joined #padre
16:23 danlucraft joined #padre
16:26 Hyppolit svn: r12287 | adamk++ | http://padre.perlide.org/trac/changeset/12287
16:26 Hyppolit Search task integration with the wx directory - Part 1
16:26 Hyppolit (Yes, things could be a bit broken for a short time)
16:26 Hyppolit trunk/Padre/lib/Padre/ trunk/Padre/lib/Padre/Wx/
16:28 PerlJam Hmm.
16:29 PerlJam Padre appears to have a slow memory leak
17:39 Alias__ That could well be my fault
17:43 Alias__ PerlJam: There's a few small known leaks
17:43 Alias__ One that comes to mind is to do with Undo/Redo
17:43 Alias__ Or something that might look like a leak
17:51 PerlJam Alias__: I had a padre instance with two files open sitting there.  I wasn't interacting with it.  After some time (20 min. maybe?)  the memory usage started to gradually climb.
17:51 Alias__ Might be the file polling
17:51 Alias__ In fact, it would have to be
17:51 Alias__ It's the only thing that runs on a regular timer
17:51 Alias__ So Padre probably had some free memory after startup
17:51 Alias__ And being Perl, didn't free it
17:52 Alias__ So then the once per 2 second file poll gradually leaked through it until it ran out
17:52 Alias__ And then headed into new memory space
18:03 danlucraft joined #padre
18:08 Steffen joined #padre
18:17 Jagdwurst joined #padre
19:03 user_2304 joined #padre
19:04 user_2304 Question for anybody using Padre: How do I give a script command line arguments?
20:25 CSJewell joined #padre
20:37 CSJewell2 joined #padre
21:09 kaare left #padre
21:19 danlucraft joined #padre
22:32 chorny joined #padre
22:55 chorny IMHO, if module is removed, a dying stub should used instead of fully removing it - this way if upgrading from old dist, it would be replaced with stub
23:01 chorny best - stub with migration docs
23:01 chorny Was Padre::Action replaced with other module?
23:23 Hyppolit svn: r12288 | chorny++ | http://padre.perlide.org/trac/changeset/12288
23:23 Hyppolit Compatibility with new Padre
23:23 Hyppolit trunk/Padre-Plugin-Git/ trunk/Padre-Plugin-Git/lib/Padre/Plugin/
23:29 Hyppolit svn: r12289 | chorny++ | http://padre.perlide.org/trac/changeset/12289
23:29 Hyppolit chdir to file directory before commit
23:29 Hyppolit trunk/Padre-Plugin-Git/ trunk/Padre-Plugin-Git/lib/Padre/Plugin/

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