>= [15:55] +static int auto_disabled() { [15:55] + static int initialized = 0; [15:55] + static int auto_up_disabled = 0; [16:00] Shouldn't "static int auto_up_disabled = 0;" mean it defaults to working unless specifically set to be broken? [16:01] It *should* be broken for systems with network manager installed, or it exposes bug #256054 [16:01] Launchpad bug 256054 in network-manager "[intrepid] new 0.7 branch ignores /etc/network/interfaces" [High,Fix released] https://launchpad.net/bugs/256054 [16:02] Wait, no, I'm an idiot, and got caught by the same issue that probably caught the author. 0 is true. [16:03] Or, no, that's not it. Ignore that last line [16:04] + int managed = iniparser_getboolean (ini_dict, "ifupdown:managed", -1); [16:05] Is anyone familiar with how bugzilla in intrepid is packaged? It extracts to individual tar.gz (seems like different languages). How do patches get applied in this case? [16:05] the -1 is what to return if not found, e.g. the file is not there presumably [16:05] isn't -1 == TRUE? [16:05] persia, :/ it looks like it wants network-mager files - I don't have network-manager [16:07] james_w, That's one of the things that is language dependent. In C, if I remember correctly, 0 is false, and non-zero (including -1, 17, 2432349827, etc.) are true. [16:08] Yagisan, Yeah. I think maybe that section needs to check for the existence of the file, and if the file is absent, presume that it should just work. [16:08] yeah, just checked, if it's not found "managed" will evaluate to true, so it will set auto_up_disabled to 1 [16:09] I just want it to not break the servers - network-manager is not in ubuntu-minimal, and if I get shockers like this - I'm less comfortable even testing new releases [16:09] Yagisan: I'll put together a fix, would you be willing to test it? [16:10] james_w, sure. server is amd64. when do you want a report back ? [16:11] Yagisan: actually, can you tell me if you have /etc/NetworkManager/nm-system-settings.conf [16:11] Soonest is best (despite the hour). With luck, we can fix it before people in NZ upgrade on Monday morning ... [16:11] * Yagisan is .au [16:12] Yes, so you know exactly how soon some sysadmin is going to come in early to work on Monday and break the world :) [16:12] (about 4 hours by my count) [16:13] james_w, no [16:14] Yagisan: then that's not the problem [16:14] persia, that was me ~14 hours ago [16:15] james_w, Hrm. So why would "managed" end up true on a server? I'm sure there's a logic error somewhere, but don't see it. [16:15] http://paste.ubuntu.com/56681/ [16:16] that's the code I'm looking at, just to check we're looking at the same thing [16:18] james_w, when I removed that file you asked for (this morning) I got networking back. I just did a quick check by touching a blank file there - and networiking went down again [16:18] Which file? [16:18] /etc/NetworkManager/nm-system-settings.conf [16:18] That should only be there if network manager is installed. [16:19] If `touch /etc/NetworkManager/nm-system-settings.conf` is enough to break it, the check is clearly wrong. [16:19] it was at one stage. long long ago. [16:19] Yagisan: ok, so it is the issue [16:19] Yagisan: are you happy to test from a PPA, or is a debdiff better? [16:19] james_w, a ppa is fine [16:20] Yagisan: cool, it will take me a few minutes [16:21] Yagisan: would you adding the information that you have that file to your bug report? [16:21] I gave up on network-manager long ago when it finished on doing dhcp for static connections [16:21] james_w, editing it now [16:21] Yagisan: and I would like it if you would remove some of the ALL CAPS parts as well, I don't think they are constructive. [16:22] james_w, I think I found it. I think "int managed = int managed = iniparser_getboolean (ini_dict, "ifupdown:managed", 0);"iniparser_getboolean (ini_dict, "ifupdown:managed", -1);" should be " [16:22] * persia tries that again [16:22] I think "int managed = iniparser_getboolean (ini_dict, "ifupdown:managed", -1);" shoud be "int managed = iniparser_getboolean (ini_dict, "ifupdown:managed", 0);" [16:23] persia: yeah, that's the change I'm making [16:23] neither was getting to spend sunday morning fixing the dammed servers instead of the other things I had to do. I'd have been roasted alive if I were to try and send a package like that to revu. [16:24] Yagisan: yes, I understand your frustration. [16:25] someone may want to adjust the Depends line so it pulls in network-manager as well considering it depends on it. [16:26] No, it specifically *shouldn't* depend on network manager. [16:26] If it does, the code should be fixed to not do that. [16:27] The problem looks like a bad default setting, but it could also be an issue with how dictionary.c handles a missing file. [16:27] no, it does the right thing if the file is missing [16:27] persia, the message it states is that ifup -a is specifically "disabled in favour of NetworkManager." [16:28] so a normal apt-get remove - not apt-get purge - will fail and hit this bug [16:28] james_w, OK. Hadn't read iniparser_load yet. Thanks for the confirmation. [16:28] (?? when did motu start fixing main ??) [16:29] Yagisan: the problem is that the current code "thinks" that NM is enabled for your device and then disables ifup as you don't want both ifup and NM messing with your NIC at the same time [16:29] When you started complaining about it here :p [16:30] persia, I should do that more often - it's so much more productive ;) [16:31] stgraber, perhaps it should check for the network-manager binary as well as the config files ? [16:32] Yagisan, You could have said something in #ubuntu-devel, which is arguably the more correct forum. [16:32] Oooh. That's a good idea. Make sure the binary is installed, and *then* check the configuration. [16:32] aaasaazaaza [16:32] oops [16:32] lucas, wrong terminal, or escaped binary snippet? [16:33] hanged terminal, but apparently what I was typing was still being sent [16:35] Yagisan: you can see it building on https://edge.launchpad.net/~james-w/+archive [16:36] Yagisan: please test once it is published, starting from a broken configuration [16:36] james_w, is that just -1 -> 0 ? [16:36] I've replicated the issue here, and tested the fix [16:36] persia: yeah [16:36] persia, I could but last time I reported a bug, I got blown off to launchpad and #ubuntu [16:36] What do you think about Yagisan's suggested guard against apt-get remove? [16:37] persia: makes sense to fix the immediate problem now, we can have a discussion on making it more robust when people networking isn't at risk [16:37] Yagisan, Depends on the bug. This is fairly core, and easy to fix. Ask something less important, and even here you'll get blown off. [16:37] persia: it makes sense, but in theory this code should be robust enough to not need it. Defence in depth is reasonable though. [16:38] james_w, Well, considering that the default answer to "Network Mangler is broken" was "remove it" for quite a while, I wonder if it's really that uncommon that the config file is present without the binaries. [16:39] persia: it actually breaks for those who upgrade ifupdown before network-manager currently, so we don't even need to look that far to find failure cases [16:39] it's not. When most of my hardy boxes go to intrepid, all of them will have the network-manager config files, but not the binary [16:39] persia: but now I see what you are getting at, and so yes, I think we should make that change [16:40] Yagisan, Yes, but you manage all those boxes, so it's only one data point. james_w point that it will break for *everyone* is probably a better defense to get this sponsored. [16:41] I don't think we'll have trouble finding a sponsor for this fix :-) [16:41] persia, if I have the wife complain - does that count as two data points ;) [16:51] huzzah - it built [16:51] just waiting for it to be published [16:58] * Yagisan twiddles thumbs [17:01] james_w, it's installed. I'll be back in a minute or two once the server reboots. [17:02] Yagisan: cool, you are starting from a broken configuration? [17:02] /etc/NetworkManager/nm-system-settings.conf now exists. [17:02] hopefully eth0, tap0 and br0 all come up. [17:03] hmm - better idea [17:05] james_w, sudo /etc/init.d/networking restart shows all interfaces up. [17:09] james_w, it worked fine on boot [17:09] Yagisan: thanks for the confirmation [17:10] thank you james_w and persia for looking at the problem [17:10] * Yagisan won't push his luck and complain about kernel bugs yet ;) [17:12] For those, you'll want to ask in #ubuntu-kernel : the likelihood any of us can solve them is a lot lower. [17:14] persia, I wish I knew how to rebuild the intrepid kernel- I strongly suspect it's a libata problem - as the dvd-rw doesn't like the commands it's getting. [17:17] Yagisan, apt-get source linux-image-2.6.27-7-generic should get you most of the way there. [17:20] Yagisan, Also, please be more careful when filing bugs. When finding a sponsor for the fix, th following comment was generated "the bugreport is a ranting chaos ..." [17:20] It's a very important bug, and it's great to have it fixed, but better to avoid that :) [17:24] * Yagisan shrugs. my ranting bugs get fixed - my non-ranting bugs don't. I've taken to fixing whatever I can and bypassing the bug reporting system. [17:25] Yagisan, I guess, although in this case the rant *delayed* the fix. === coppro is now known as coppro|breakfast [17:26] james_w, On another note : thanks for chasing jd. I'm curious though: do you read Japanese? [17:26] persia: nope [17:28] Ah. I had initially assigned that to myself, just because I figured I'd have an easier time testing, although on investgiation it turned out you had fixed it. Fairly impressive, considering that there's a bug against it in Debian that the internals are in Japanese. [17:28] yeah, I grabbed the wrong source package, so the ancestry is wrong in the changelog. [17:28] luckily it didn't involve reading the comments :-) [17:29] * Yagisan can read some, but not more than a 4 year old can. [17:29] Oh, is some of that Heinrich's changes in Debian then? [17:30] no, it just no longer has the build1 entry [17:30] there is an open sync request though [17:31] I saw that. I'm planning to poke shibata about it when I next see traffic, but it's a holiday weekend, so it's probably not until Tuesday at the earliest. [17:32] * james_w notes that there are a couple of jp-related sponsor requests open, perhaps good for someone with experience in that area [17:32] Yeah. I'll probably hit the sbackup one in the morning. Any others worthy of specific note? [17:33] timidity was the other I was looking at [17:33] And that's even a source package I've touched before :) [17:35] Hrm. I thought I fixed the init script already. There's a fair bit of timidity work. Thanks for the pointer. === coppro|breakfast is now known as coppro [18:17] I've adapted a workaround for an X server bug from kpowersave to xprintidle. [18:17] The debian/control file lists Milan Zamazal as the debian maintainer. [18:18] I'll submit a patch to the author of xprintidle, Magnus Henoch , once I've tested it for a while. [18:18] emh, Is there a bug in launchpad also? [18:19] persia: I don't know. [18:19] there is a bug here: http://bugs.freedesktop.org/show_bug.cgi?id=6439 [18:19] Freedesktop bug 6439 in Lib/other "Idle time will get reset to 0 after 20 minutes" [Normal,New] [18:20] emh, OK. So there are three places you can fix the bug to get it fixed in Ubuntu. [18:21] You can fix it upstream, and Debian will get the new version, and then Ubuntu will get the new version. It sounds like you're already planning to work with Magnus to do that. [18:21] Yes, that seems like a good plan. [18:22] You can get it fixed in Debian, in which case you want to open a bug in the Debian BTS, and submit the patch, so Milan can look at it. It may or may not be suitable for Lenny, but at least Milan can make that decision. It probably won't get seen if it only goes to Magnus until after the next release. [18:23] You can get it fixed in Ubuntu. Submitting the patch (or preferably a debdiff including the patch) in a bug in Launchpad is the best way to request this. Someone will evaluate, and maybe apply. [18:23] If you do open a bug in Launchpad, linking to the freedesktop.org and Debian bugs would be best. [18:24] So, the path to take depends on the urgency of the bug, and the amount of energy you have to spend on it. [18:27] Ok. Thank you. I currently have plenty of time and energy, so perhaps I'll pursue all 3 paths once I'm more confident that my changes work. It will be nice for learning about the processes involved. [18:31] Anyone know any caveats about replacing libcgi-pm-perl with perl-modules? [18:34] emh, Once you have something that works, you can ask here for help with all three processes. We're going to be best at explaining about getting it straight into Ubuntu, but since we also send patches upstream and to Debian, we can provide some advice with that. [18:35] persia: Ok. Another thing is that the code I adapted from kpowersave is GPL but xprintidle is BSD, so xprintidle will probably have to be relicensed under GPL? [18:37] emh, That would be Magnus' decision. It's unlikely Milan or anyone in Ubuntu would be willing to relicense. [18:37] emh, I'd advise finding the solution, and letting Magnus know you have a GPL patch. Ask if he wants the patch, or a description of the behaviour. [18:38] If he doesn't want to relicense, he may want you to describe the solution in words, so he can reimplement it without having ever seen the other code. [18:38] That way there's no license question. [18:38] Clever :) [18:43] anyone care to sponser - bug #280641 - need to fix this before i can put a CVE into it. [18:43] Launchpad bug 280641 in bugzilla "Broken package" [Undecided,Confirmed] https://launchpad.net/bugs/280641 [18:50] stefanlsd: why is that stopping you from doing the update for the CVE? [18:51] james_w: i have the debdiff for the CVE already. just it will need to be based on the updated version. [18:51] why not roll the fixes together? [18:53] as far as i know, security fixes are just the security fix, no other changes. [18:53] stefanlsd: only for stable releases [18:54] stefanlsd: and a bug that prevented a security update from being installed would kind of nullify the point of the security update [18:57] james_w: ok, if i can do it in both, i'll fix the diff. thanks [18:58] stefanlsd: do you know when libcgi-pm-perl was removed? [18:58] what source package was it built from? [18:58] james_w: unsure. but i suspect it was some time ago, from some of the other changelogs i've seen [18:58] james_w: it's probably in perl-modules now.. [18:59] laga_: yeah, it is [19:12] james_w: fixed the diff to include the CVE [19:13] stefanlsd: what needs fixing in Debian? [19:13] james_w: sry, I was ill the last days. will prepare ssmtp updates back to dapper tomorrow :) [19:14] sebner: cool, thanks [19:14] get well soon [19:14] james_w: thx [19:14] james_w: what do u mean? [19:15] stefanlsd: I don't see links to Debian bugs for either of the changes [19:23] james_w: seems like libcgi-pm-perl is still in debian, so not an issue. http://security-tracker.debian.net/tracker/CVE-2008-4437 - best i can find on the cve. [19:23] stefanlsd: it might still be worth reporting [19:25] the security issue is marked as fixed in unstable [19:25] james_w: will pass patches up to debian. yeah. [19:25] they've got the fix then, so I don't think there is a need to forward it [19:25] my dinner is ready now, I'll take a look after that [19:25] james_w: kk. thanks :) [19:55] hey guys [19:55] can I still fix bugs in a package or is it too late now [19:55] sorry I've been a little out of touch lately [19:55] fix only is fine, if it requires a new upstream that also introduces new features you'll need a FFe though [19:56] ok [19:56] it's only a tiny app [19:56] gisomount [19:56] I saw that some bugs were raised that I should ought to fix [19:57] I don't intend to add any new features [19:57] but i would be fixing these bugs upstream [19:57] if the new upstream is bugfix only, that should be fine [19:57] ok [19:58] how long do I have to fix these bugs? [19:58] we'll enter RC freeze on Thursday IIRC, so best to have that fixed before then [19:59] but as I guess it's an universe package, it should be possible to send bugfix only even after that date (I don't upload in universe so I'm not quite sure about that though) [19:59] hehe [19:59] I can upload to universe [19:59] I think ;) [20:20] stefanlsd: "debian/maintenance"? bonkers [20:20] not your fault of course [20:20] yeah. first time i've seen that system also. [20:33] stefanlsd: why ubuntu2.1? [20:35] james_w: yeah. wanted to check about that. was either 1.1 or 2.1. i was under the impression that .1 would be just a security update. wasnt sure what happened if it had another fix with it. [20:36] hello [20:37] stefanlsd: that only applies for updates the stable releases. The reason it is done is so that it doesn't use a version number where the same version number would be in the next release, which would mean there wasn't an upload path. That doesn't apply here, so I've made it just ubuntu2 [20:37] hey crevette [20:37] hello james_w [20:38] james_w: thanks. understood. [20:44] I'd be interested if someone can review https://bugs.launchpad.net/ubuntu/+source/nemiver/+bug/262246 [20:44] Launchpad bug 262246 in nemiver "Sponsor request to update Nemiver to version 0.6.3" [Wishlist,Confirmed] [20:44] thanks [20:47] stefanlsd: doesn't work for me [20:48] Checking for CGI.pm (v3.33) found v3.29 [20:49] crevette: ping [20:49] slytherin: hello [20:50] crevette: nautilus-sendto seems to be broken for some people, showing error obex not supported. Any idea? [20:50] hmmm no [20:50] It tested with my deviceds ant it was working [20:51] * crevette tries on its new phone [20:51] crevette: It shows error for me too but then nothing is working on my side so Ican not say only nautilus-sendto is broken. [20:52] crevette: does nautilus-sendto have depends on libbtctl? [20:58] james_w: what doesnt work for you? [20:59] Checking for CGI.pm (v3.33) found v3.29 [20:59] installing the package [21:00] http://paste.ubuntu.com/56776/ [21:04] slytherin: do ypu have any error when you run the command from the terminal ? [21:05] crevette: as I said bluetooth is not at all working for me. Probably hardware problem. So you will get no useful information from me. [21:10] slytherin: it seems nautilus-sendto just use plain bluez dbus APoi [21:10] API [21:12] crevette: Ok. I will let you know the bug tomorrow if I find it. Too late for me now. Got to go. [21:12] okay good night slytherin [21:17] wgrant: mm. bleh. just did it in vm and get the same error. sorry. from what i've found. bugzilla does want 3.33. perl-modules doesnt have it. debian recommends - libcgi-pm-perl which replaces perl-modules CGI.pm with a later version (we dont have it) [21:17] evening [21:17] emgent: \o/ [21:17] sebner: :) [21:18] bleh. james_w ^ [21:18] emgent: did you also participate at the ubuntu-it meeting? [21:18] stefanlsd: yeah, you might want to find out why we don't have it [21:19] sebner: no [21:20] james_w: kk. thanks. will try find out. [21:21] emgent: what a pitty. Luca showed me the "I'm drunken because of the good wine" photos :P [21:21] hahah [21:22] i saw that. [21:22] ^^ === rn114_ is now known as rn114 [22:20] hi, i'm having a play with creating .deb packages, and have successfully created a couple. however, having updated the sources i'm working from and done a few tweaks of the packaging no binaries actually end up in the .deb following running pbuilder on the .dsc [22:20] could anyone think of an obvious reason? [22:23] rn114: cleaning up the debian/files during the process was an obvious one for me the other day [22:24] rn114: do you try to build multiple binaries from one source? [22:25] Elbrus, im certainly not trying to [22:25] rn114: maybe you can show your rules file in a pastbin [22:25] !pastbin [22:25] Sorry, I don't know anything about pastbin [22:25] yeah ill paste it up thanks [22:25] !pastebin [22:25] pastebin is a service to post multiple-lined texts so you don't flood the channel. The Ubuntu pastebin is at http://paste.ubuntu.com (make sure you give us the URL for your paste - see also the channel topic) [22:26] thats what I meant... :( [22:26] yeah :) here it is: http://paste.ubuntu.com/56800/ [22:27] its quite simple, based off a dhbuild's production [22:29] I think you commented out the dh_install was that intensionally? [22:31] Elbrus, around line 79? [22:31] exactly [22:32] hmm, its a good point [22:32] i'm just going to diff with my old version to check if i missed that [22:34] thats the same as before, i think it should be handled by the dependency on install shouldnt it? that calls $(MAKE) install INSTALL_ROOT=$(CURDIR)/debian/mythtv and i think that should work [22:35] its strange because when i run pbuilder it does all the compilation and everything (and seems to do the install step as well) but then the .deb just doesn't contain any binaries! [22:38] rn114: I might be completely wrong, but either the install should be in DEBIAN instead of debian, or something like dh_install --sourcedir=debian/mythtv is needed (mind you I don't know the details of building yet). [22:40] rn114: maybe have a look at the man-page of dh_install [22:44] Elbrus, i think dh_install just copies a list of files and the make install command should handle that for me [22:47] * Elbrus just tries to find out where the files are copied to for the build.. is that in debian/? [22:49] heres an example line from the build log file: cp -f "mythfrontend" "/tmp/buildd/mythtv-robin-18380/debian/mythtv/usr/local/bin/mythfrontend" [22:49] this is were it copies it [22:58] rn114: maybe you can also pastebin the build log? Just to see if I see something obvious... [22:58] its almost a megabyte! [22:58] i just cant really see much of a difference annoyingly [23:04] rn114: sorry, I am out of ideas... Somebody else??? [23:07] one thing, shouldn't the INSTALL_ROOT not be DESTDIR? See: http://www.gnu.org/software/automake/manual/html_node/DESTDIR.html [23:07] s/not// [23:11] norsetto_: o/ [23:11] emgent: \o/ [23:14] hihu norsetto_ [23:15] sebner, \\o// [23:16] * sebner is scared. Why does norsetto has 4 hands ^^ [23:16] sebner its the 2 heads which should be worrying [23:17] sebner: that's the legs flying up as he leaps in the air [23:17] norsetto: ^^ [23:17] ajmitch: flying ^^ [23:23] thanks for your thoughts Elbrus