/srv/irclogs.ubuntu.com/2011/11/06/#ubuntu-devel.txt

sveinsepenguin42: excellent, thanks. It works using the Linaro PPA. I'll comment that on the bug00:04
penguin42no problem00:05
broderugh. looks like upstart's test suite works fine on the buildds, but errors out on my oneiric laptop: http://paste.ubuntu.com/729586/00:35
penguin42broder: could be a load of differences I guess; don't the buildd's tend to run with fairly old kernels?00:46
broderpenguin42: they do - i believe they're generally the previous lts. i know that libnih and upstart's test suites have caught weird kernel bugs in the past and i'm afraid i'm going to end up chasing a similar rabbit hole or something. trying to get a minimal test case00:47
penguin42broder: Or it could be an interaction of the upstart/packages on the laptop I guess00:47
lacquiHi.  Trying to fix my first bug here (bug #885329).  I've followed the instructions on https://wiki.ubuntu.com/Bugs/HowToFix but I'm getting the following error:03:39
ubottuLaunchpad bug 885329 in eggdrop (Ubuntu) "eggdrop crash on i386" [Undecided,New] https://launchpad.net/bugs/88532903:39
lacquidebian/rules:4: /usr/share/cdbs/1/rules/debhelper.mk: No such file or directory03:39
bjsniderlacqui, install cdbs on your build system03:42
bjsniderapt-get build-dep eggdrop03:42
lacquibjsnider: thanks03:42
lacquilike I said, first bug :P03:42
lacquiIs this the right place to request a review for a bug fix? (#885329)05:38
sladenlacqui: let me have a look05:54
lacquisladen: thanks06:03
=== yofel_ is now known as yofel
pgquilesI'm the maintainer of the 'witty' package. In Oneiric, the package does not work because it was apparently built against a buggy jQuery ( https://bugs.launchpad.net/ubuntu/+source/witty/+bug/886456 ). How can I get an update (actually, a rebuild) included in oneiric-updates :-?13:13
ubottuLaunchpad bug 886456 in witty (Ubuntu) "Built-in httpd does not render any widget" [Undecided,Fix committed]13:13
tumbleweedpgquiles: request a rebuild in the bug and subscribe ubuntu-sponsors13:16
pgquilestumbleweed: thank you13:16
tumbleweedit probably should symlink to libjs-jquery's jquery rather than bundling it, although that's not without caveats either (see the recent discussion in debian-devel)13:17
Chipzzpgquiles: actually it depends what you mean. You didn't specify if you wanted an update of your package or of jquery13:17
pgquilesChipzz: my package13:18
pgquilestumbleweed: witty symlinks to jquery on build, but it converts jquery to C++ and embeds it in the .so13:18
pgquilestumbleweed: I was part of the discussion in debian-devel :-)13:18
tumbleweedconverts jquery to C++? eep13:19
Chipzzpgquiles: exactly how do you intend to convert jquery into C++??13:20
hyperairO_o13:20
ChipzzjQuery is a CLIENT-side javascript library. C++ is (at best) used in your software which is SERVER-side13:21
hyperairmaybe it's something elfrc13:21
hyperairstuff the jquery source code into a .so file that can spit out things on the server side?13:21
* hyperair shrugs13:21
Chipzzhyperair: jQuery will NEVER run server-side13:22
Chipzzbecause it operates on the DOM13:22
hyperairChipzz: read what i said carefully.13:22
pgquilesChipzz: tumbleweed: on build, Wt takes jquery.min.js and generates a C++ file, which is compiled into the .so13:22
hyperairChipzz: elfrc embeds arbitrary files into elf objects. since it's a built-in httpd, it wouldn't be unplausible to have it serve jquery.js from a static string somewhere13:23
pgquilesWt is not like PHP. Webapps using Wt are real programs (binaries), the result of compiling C++. Instead of reading jquery.min.js every time from disk, it is streamed from memory13:23
hyperairright, see, that's what i said.13:23
Chipzzhyperair: that would be a big big BIG facepalm13:23
hyperairChipzz: it's serving jquery.min.js from memory instead of reading from disk.13:23
* Chipzz starts to understand the sentiment against web developers on #debian-devel13:23
pgquilesChipzz: developing web apps with Wt is more like developing a desktop app with Qt13:24
pgquilesC++, a copycat of the Qt API, etc13:24
pgquiles:-)13:24
pgquilesgreat for embedded and/or high-performance13:24
Chipzzpgquiles: as someone who does web development as a full-time job I have 3 words for you: WHAT. THE. FUCK?13:24
pgquilesChipzz: http://webtoolkit.eu13:25
pgquilesI gave up on Rails when I started using Wt13:25
hyperairChipzz: hey it's not so bad, it's like CGI except without the exec() overhead?13:25
* hyperair wouldn't mind doing native code webapps.13:25
hyperairit could turn out to be pretty performant.13:25
pgquilesit's not only native code, it's also the API: forget about pages, etc. It's like a desktop app.13:26
pgquileshyperair: in fact, you can also compile the webapp to a FastCGI module instead of using the embedded HTTP server13:26
hyperairpgquiles: in all my life in web development, i've always looked down upon *anything* that autogenerates my html code for m.e13:26
Chipzzhyperair: for what gain exactly? serving jquery.min.js a couple of ms faster? taking in mind that jquery.min.js is minized already... if performance REALLY matters that much, put a reverse proxy in front of it. but for the love of god, why obscure your web application in such a way???13:26
hyperairChipzz: *shrug* i'm talking about the idea of webtoolkit in general, not the jquery instance.13:27
hyperairChipzz: also a reverse proxy is sometimes a rather overengineered solution.13:28
Chipzzhyperair: not as overengineered as compiling jquery.min.js into a .so :|13:28
pgquilesChipzz: because that way when you want to deploy your webapp, you only need to deploy ONE single executable that embeds everything13:28
hyperairChipzz: the framework's already there. i don't think it's really that overengineered.13:28
hyperairChipzz: it probably ends up even simpler.13:28
hyperairChipzz: especially at the deployment stage13:29
Chipzzpgquiles: which matters exactly nada since you're packaging it. you deploy your webapp by using apt-get install or dpkg -i13:30
pgquilesChipzz: not everybody packages everything. Especially on embedded, Windows, mobile, etc13:31
Chipzzpgquiles: no, but you are asking specifically in the context of packaging it13:32
hyperairpgquiles: does witty generate this one-executable-binary thing for new webapps?13:32
pgquilesChipzz: sorry, I don't follow. The 'witty' package contains the libraries to develop webapplications, not an aplication13:32
hyperairpgquiles: i.e. it generates the single-executable-webapp?13:33
hyperairpgquiles: like a funky linker?13:33
pgquileshyperair: yes, it does. Just like against wthttp and it will embed the HTTP(S) server. If you'd rather generate a FastCGI module, just link to wtfcgi.13:33
hyperairwtfcgi13:33
pgquileshyperair: Wt webapps are developed in C++ and compiled and linked using gcc and ld13:33
hyperairi like that name.13:33
Chipzzpgquiles: same difference really.13:34
pgquilesit's not "link a funky linker" but "using a funky linker" :-)13:34
hyperairChipzz: so there you have it. witty isn't a web server. it provides the tools to make a one-executable webserver.13:34
pgquilesthat's it13:34
hyperairChipzz: with embedded webapp.13:34
Chipzzhyperair: like I just said, same difference. unless witty is some sort of "compiler" instead of a "library"13:34
hyperairChipzz: that's exactly what i'm talking about.13:34
hyperairChipzz: more like a runtime that's statically linked with your webapp.13:35
pgquilesChipzz: witty is a C++ library. You use it like you use Gtk+, Qt, libc, etc13:35
Chipzzstatic linking? this keeps getting better and better13:35
pgquilesChipzz: or dynamic linking, your choice13:35
hyperairChipzz: it's a one-executable webapp-cum-server. wtf else do you expect it to be besides static linking?!13:35
Chipzzhyperair: do I need to point out debians and ubuntus stance about static linking?13:36
hyperairit's something you throw at somebody and tell them to execute to host a web server with said webapp13:36
pgquileslibwt*-dev include the static libraries and the dynamic libraries13:36
hyperairChipzz: it's for *your own* web app.13:36
hyperairChipzz: something you're going to throw at $other_distro to run.13:36
Chipzzhyperair: what happens when it turns out the version of jquery witty is currently shipping turns out to be buggy?13:37
hyperairChipzz: then you binNMU.13:37
pgquilesChipzz: actually, then you run into the discussion from debian-devel13:37
Chipzzall the reverse deps? including all the webapps you compiled against witty and you sent off god-knows-where?13:38
pgquilesChipzz: if you replace jquery, probably the webapp breaks because it's not ready for that new version of  the javascript library13:38
pgquilesChipzz: just dynamic link, man13:38
hyperairChipzz: pgquiles said you can dynamically link your webapp anyways, so...13:38
pgquilesstatic linking is your choice and it's not the default13:38
pgquiles(CMake picks the so's by default)13:38
ChipzzI think I'm going to retract from this discussion in order not to induce self-inflicted harm by repeatedly banging my head against the wall. laters13:39
hyperairChipzz: i don't know much about wt, this being the first time i'm hearing about it, but all your points so far has been off-mark.13:39
hyperairyour strawman is burning very brightly13:39
pgquilesChipzz: try Wt, then tell me what you think. After using it :-)13:40
hyperairpgquiles: he probably won't, due to his set preconceptions against it.13:41
pgquilesI have to admit the first time I read about Wt I thought it was crazy or plain stupid. Turns out it works beautifully and it's extremely performant.13:42
hyperairi'll just say this out loud first: i dislike qt's way of reimplementing half the C++ standard library.13:43
hyperairif wt does the same thing, i won't touch it with a 100-foot pole.13:44
hyperaircase in point QString13:44
pgquileshyperair: Qt needed to do that. Back then, C++ compilers were so broken and/or incomplete there was no other way to support other Unices.13:44
hyperairpgquiles: that's in the past. qt4 could have dropped that.13:45
pgquileshyperair: Wt uses Boost, it does not reimplement13:45
hyperairoh goody.13:45
pgquileshyperair: no, Qt 4 could not do that13:45
pgquilescase in point: Solaris13:45
hyperairpgquiles: typedefs blah blah thin wrappers13:45
pgquilesHP-UX13:45
pgquilesAIX13:45
hyperairpeople use that crap?13:45
* hyperair coughs13:45
pgquilesyou'd be surprised how many companies use Boost13:45
hyperairi'm not talking about boost13:45
hyperairi'm talking about solaris/hp-ux/aix13:46
pgquilessure, less and less, but it's still widely used13:46
hyperairboost is superbly awesome.13:46
pgquilesheck, even SCO still has some clients left! :-D13:46
hyperairyes, why couldn't qt use boost?13:46
pgquilesback then Boost did not provide all it provides now13:46
* hyperair mumbles something about how it should die in a fire already13:46
hyperairpgquiles: qt4 is fairly recent. boost was very mature by then13:46
pgquilesalso, Boost does not promise binary compatibility across versions13:46
pgquilesnot even now13:46
hyperairoh right, that.13:46
* hyperair sighs13:46
pgquilesQt4 was released in 2005, started in 200313:47
pgquilesQt 5 will be released early 201213:47
pgquilesalso, C++ still does not provide introspection capabilities13:47
hyperairnor will it ever.13:47
pgquileswhich is one of the main features moc provides13:47
hyperairi find it a disgusting bastardization of C++13:47
pgquileseven Glib has resorted to invent a mechanism for introspection13:47
hyperairthat's because C has no OO capabilities.13:48
pgquilesthe fact that C does not provide classes does not mean it has no OO capabilities13:48
pgquilesGlib and Gtk are OO with C13:48
hyperaircorrection: with C/GObject.13:49
hyperairit's all part of GObject's API13:49
hyperairit's not pure C.13:49
hyperairfor the record, gobject's reference counting has driven me up the wall many a time13:49
pgquilesgrammatically you are right13:49
* hyperair sighs13:49
hyperairmy definition of standard C is what ##c has.13:50
hyperairand my definition of standard C++ is what ##c++ has.13:50
hyperairby those definitions, i am correct in saying that C has no OO capabilities, period.13:50
hyperairnot "grammatically" or "technically" or any crap like that.13:50
hyperairbut GObject/C is better than Qt/Moc in the sense that it is implemented in standard C, rather than requiring some funky preprocessor to do stuff prior to compilation.13:54
hyperairwhen you need a preprocessor that's not the standard C++ preprocessor, it's no longer C++.13:54
macoWhat is moc?13:57
penguin42hyperair: It's not too unusual to have preprocessed stuff before your lang13:59
pgquilesmaco: the metaobject compoiler13:59
pgquileshyperair: GObject introspection requires the use of the gir scanner, which is essentially the same as moc13:59
hyperairpgquiles: there is a difference.14:00
pgquilesmaco: http://doc.qt.nokia.com/latest/moc.html14:00
hyperairpgquiles: the gir scanner scans the *exact* same thing as the input to gcc.14:00
hyperairpgquiles: the input to moc isn't the same thing that goes into g++14:00
penguin42hyperair: All of these things are just signs that neith C or c++ quite fits into the view of what the GUI programmers need14:04
penguin42hyperair: Which I guess is why you get things like Vala14:05
hyperairpenguin42: or people could use C#! \o/14:05
penguin42hyperair: Yeh14:05
* hyperair holds up his evil mono supporter flag14:05
* penguin42 burns it14:05
hyperair>=O14:05
hyperairpenguin42: gtkmm is standard C++ though.14:06
* penguin42 hasn't played with it14:07
hyperairit's good stuff. unfortunately it doesn't work with gobject introspection14:09
hyperairso C++-implemented widgets can only be used by other gtkmm things14:09
mewerner_arandWhen was the installer modified to use @ + @home subvolumes for btrfs? Did it appear in natty? I'm making a layout section on the wiki: https://help.ubuntu.com/community/btrfs#Ubuntu-specific_subvolume_layout14:40
hyperair@?14:41
udevbot_Error: "?" is not a valid command.14:41
hyperairthat's pretty cool14:41
mewerner_arandhyperair: the subvolume tat holds "/" is named "@", yes.14:42
hlamerMay package in PPA depend in package in another PPA? (Or what is the right channel for this question?)17:19
broderhlamer: at the top-right of the PPA page should be an "Edit PPA dependencies" link. (for reference #launchpad is probably a better channel for discussing PPAs, though)17:21
hlamerthanks, broder17:22
highvoltageWhen will I be able to see my work items in http://status.ubuntu.com/ubuntu-precise/u/jonathan.html ?17:31
highvoltageOr should I be looking somewhere else?17:31
=== emma is now known as Em
dupondjeEmpathy is broken since last upgrade it seems :(19:13
dupondje[   88.095003] type=1400 audit(1320603235.045:21): apparmor="DENIED" operation="open" parent=1 profile="/usr/lib/telepathy/telepathy-*" name="/etc/default/apport" pid=1857 comm="telepathy-butte" requested_mask="r" denied_mask="r" fsuid=1000 ouid=019:13
=== lacqui_ is now known as lacqui
slangaseklifeless: lp:~vorlon/launchpad/sbuild-multiarch-syntax test case> where's the right place for me to add this?20:10
lifelessslangasek: you know, I have no idea :)20:11
lifelessrather than flailing around with grep etc, I will ask some folk that have done a lot in this area, in a couple of hours20:12
lifelesshow soon do you need this change live?20:12
slangaseklifeless: the sooner it can go live, the better; there are a number of packages in the base system that need to be switched to make use of this syntax to be cross-buildable20:20
slangaseklifeless: indeed, 3 out of 5 Prio: required packages that I've looked at so far are affected (acl, attr, bash)20:27
marseillai_hello21:08
marseillai_could someone tell me where dh_make store the packager name information ? i'd to change it21:08
jbichamarseillai_: debian/control21:09
marseillai_jbicha: i didn't explain weel sorry.21:10
marseillai_when dh_make build my package it store a wrong name in debian/control I'd like to change the source of this information21:11
marseillai_it mispel my name21:11
DktrKranzmarseillai_: DEBFULLNAME environment variable21:42
bjsnidermarseillai_, you can store that info in .bashrc, and it may be there21:50
marseillai_TheMuso: DktrKranz and bjsnider21:58

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!