=== jamesh_ is now known as jamesh === maclin1 is now known as maclin [04:34] hi all [04:34] https://launchpad.net/ubuntu/+source/gsequencer [04:34] ^^ does proposed mean, it is going to be synced? === wgrant changed the topic of #ubuntu-devel to: Launchpad build farm capacity limited, and non-x86 builds suspended | Artful Released + taken down LP#1734147 | Archive: open | Devel of Ubuntu (not support or app devel) | build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of trusty-artful | #ubuntu-app-devel for app development on Ubuntu http://wiki.ubuntu.com/UbuntuDevelopment | Patch Pilots: [04:44] joelkraehemann: has been auto-sync'd, not yet built though [04:50] ginggs: thank you. [04:50] I can strongly recommend this version [04:51] since it brings some real improvements === mvo_ is now known as mvo [08:16] Hello! I saw linux-generic linux-headers-generic linux-image-generic linux-tools-generic with linux-meta (3.13.0.139.148) trusty; urgency=medium // * Bump ABI 3.13.0-139. Ubuntu wesite tells me https://launchpad.net/ubuntu/+source/linux/3.13.0-139.188 its KAISER / KPTI patches for Linux. [08:17] My question is the sage as stated in https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1741609/comments/4 [08:17] Launchpad bug 1741609 in Kernel SRU Workflow security-signoff "linux: 3.13.0-139.188 -proposed tracker" [Medium,Fix released] [08:17] Can someone please tell me if thats feasible? === Elimin8r is now known as Elimin8er === caribou_ is now known as Caribou [08:29] Ah I found an answer to the KAISER problem (https://askubuntu.com/questions/991874/how-to-disable-page-table-isolation-to-regain-performance-lost-due-to-intel-cpu). Thanks anyways! :-) [09:58] rbalint, hey, is https://bugs.launchpad.net/ubuntu/+source/lightdm/+bug/1663157 still on your list? we would like to re-enable guest session for the LTS but the apparmor/systemd user issue needs to be sorted out ... and it seems Steve pinged you about that? (it has been a while though so not sure if that's still current) [09:58] Launchpad bug 1663157 in Light Display Manager "Guest session processes are not confined in 16.10 and newer releases" [Undecided,New] [10:43] slangasek, will you merge klbic from Debian unstable? [10:52] seb128: it is still on my list but had higher priority issues :-( [12:28] rbalint, ok, and do you have any idea if/when you might get to that item? [12:38] seb128: speaking of still on people's list, I made some progress on debconf gtk3; just trying to sort out some complexity regarding the startup sequence [12:39] it mostly works but is producing weird warnings at the moment, so I want to clear that up [12:58] xnox, slangasek, can you have a look at bug 1721839? It is a severe regression which we should not have in LTS and upstream is not answering. [12:58] bug 1721839 in systemd (Ubuntu) "[REGRESSION] Services asked for by UDEV do not get triggered" [Critical,Confirmed] https://launchpad.net/bugs/1721839 [13:31] seb128: i need to discuss that with slangasek, we did not talk about that recently [13:31] seb128: but i agree that we should enable it [13:38] cjwatson, ah, that's good new, thanks [13:39] rbalint, can you bring the topic with Steve and come back to me then, we need to know what to do on our side and the feature is sort of blocked on that item to be resolved [13:51] seb128: sure [14:18] rbalint, thanks [14:22] didrocks, slangasek - i want new plymouth, and i want initramfs hooks to correctly set force-scale, and i want high-dpi plymouth themes by default. [14:22] didrocks, slangasek - are you planning on merging plymouth, or shall i? [14:29] xnox: please go ahead ;) I won't really have time for handling it [15:34] xnox, slangasek, can you have a look at bug 1721839? It is a severe regression which we should not have in LTS and upstream is not answering. [15:34] bug 1721839 in systemd (Ubuntu) "[REGRESSION] Services asked for by UDEV do not get triggered" [Critical,Confirmed] https://launchpad.net/bugs/1721839 [15:37] * juliank feels like freenode is being ddos'ed or something - everything is really slow [15:38] 30% slower, you say? [15:44] Seems I hit a bug in znc-clientbuffer [15:52] It might sound stupid, but searches weren't kind to me - is there a Ubuntu specific counterpart to README.Debian? [15:52] I mostly see us adding to it which is fine [15:53] just thought to ask for once to be sure there is no hidden "if ubuntu only add to X instead" [15:56] cpaelzer: not that I'm aware of [15:59] Odd_Bloke, hahahahahaha [15:59] i'm bouncing too [16:00] cpaelzer: if README.Debian is misleading (because of an Ubuntu delta), I may patch README.Debian rather than create a separate file. Being careful to avoid misattributing the change to Debian. [16:00] As I see README.Debian as more of a README.packaging as opposed to a README which I'd normally attribute to upstream. [16:00] That's just IMHO. I'm not sure what other Ubuntu devs do. [16:02] thanks rbasak / mdeslaur [16:03] I know what I want to write and have no fear to be misleading [16:03] just wanted to make sure to not miss something all the time [17:19] is this a valid C program? https://paste.debian.net/1004679/ [17:20] on i386 it prints 42123455, which is not what I would expect [17:24] is it a valid C program? sure. Are you right to expect it to give you the obvious number as an answer? not without some compiler flags [17:24] Laney: unfortunately I don't remember what the flags are or have a good way to look them up without highlighting doko [17:25] Laney: it might be -ffloat-store [17:28] Laney: welcome to the world of floating point. You can play with #include fesetround(FE_UPWARD); etc [17:29] Laney: https://godbolt.org/g/E8d6sH might help you figure it out [17:30] 64 has a jump happening in the multiplication somehow, is curious [17:31] It's interesting because if I multiply and then truncate in separate steps I get the "right" answer [17:34] but if I annotate the intermediate double as a register variable it "works" [17:34] and ffloat-store indeed "fixes" it [17:35] s/works/breaks/ sorry(!) [17:40] Laney: add to that, that Intel has internal 80-bit float registers, but only 64-bit when written back to memory [17:42] bloop [17:46] Laney: if you more carefully pick your example numbers to be ones that are accurately representable in floating point (ie. power-two integer fractions), you should be able to get "accurate" answers [17:50] sladen: I would understand it more if the intermediate multiplication were wrong (https://paste.debian.net/1004686/) [17:51] Laney: one has to first swallow the blue pill and accept that it is _not wrong_ [17:53] It seems wrong for it to differ based on whether I do it in one or two steps [17:53] Laney: you're trying to view a decimal (base-10) approximation of mulitplying two binary fractions (1/N) * (1/N), which have been created from representable approximations of decimal (base-10) inputs [17:54] The link I just pasted shows the expected answer [17:56] in one case the printf() routine is rendering a base-10 approximation of a floating point value. In the other case the printf routinue is rendering a precise representation of an integer value [17:58] ...but that integer value is a truncated approximation of multiplying two approximations of base-10 numbers together, in 1/base-2 space [19:21] Laney: did you try long double and __float128 to get the full menu of outputs: https://stackoverflow.com/questions/13516476/long-double-gcc-specific-and-float128