=== wedgwood is now known as wedgwood_away === Nisstyre-laptop is now known as Nisstyre [03:57] usb-creator is broken seven ways from sunday [04:10] anyone alive familiar with the dep8 testing suite @ lp:auto-package-testing? [05:22] adam_g_: yolanda is. === fenris- is now known as ejat [11:44] Who broke the builders? Here is a debdiff between pyside 1.1.2-3ubuntu1 and 1.1.2-4 (http://launchpadlibrarian.net/142485559/pyside_1.1.2-3ubuntu1_1.1.2-4.diff.gz) [11:44] ... and the latter ftbfs on both i386 and amd64 while the former built fine... [11:49] am I blind or is there no error message? [11:52] there is no error message :) [11:52] and the build was done up to 100% [12:01] mitya57: make[4]: *** No rule to make target `/usr/lib/i386-linux-gnu/libQtWebKit.so', needed by `PySide/QtWebKit.so'. Stop. [12:02] (trick: look at the make level that fails near the end, namely make[2], then search back for successively increasing make levels until you find the one with the actual error message nearly [12:02] *nearby) [12:03] cjwatson: oh thanks; but why did the mostly identical version build? [12:06] mitya57: dunno, maybe qtwebkit changed? [12:07] oh yes, seb128 sponsored new webkit yesterday [12:07] *qtwebkit [12:15] libqtwebkit-dev_2.3.1-0ubuntu1_i386.deb does contain /usr/lib/i386-linux-gnu/libQtWebKit.so symlink [12:23] * mitya57 found the issue [12:24] what is it? [12:28] hardcoded version in rules [12:28] ln -s libQtWebKit.so.4.10.0 debian/libqtwebkit-dev/usr/lib/$(DEB_HOST_MULTIARCH)/libQtWebKit.so [12:29] the .so is 4.10.1 in the new version [12:49] Grr, who retried llvm-toolchain-3.3 on i386? :( [12:49] Pro tip: if a build is hanging, retrying it over the weekend when no one will be around to kill it is probably silly. [13:08] infinity: probably that was a new upload to sid [13:09] and it hangs again :( [13:09] mitya57: It wasn't. It's already been killed once and someone retried it in the last couple of hours. [13:11] infinity: that wasn't me, but kill it anyway if you're able to [13:11] mitya57: I'm not able, which was the point of my grumpiness. All the people who have access to log in to buildds tend to take their weekends very seriously. Oh well. :P [13:12] Somebody really needs to fix auto-killing... [13:12] Yeah, it's on my TODO. === tkamppeter_ is now known as tkamppeter [16:17] infinity, hi :), jfyi, this linker failure looks like a toolchain problem -- https://launchpadlibrarian.net/142502649/buildlog_ubuntu-saucy-i386.gst-plugins-bad1.0_1.1.1.1%2Bgit20130614.d50625ee-0ubuntu1~13.10~ricotz0_FAILEDTOBUILD.txt.gz [16:19] ricotz: Why on earth is that using -nostdlib and then trying to guess the right thing to do? [16:21] ricotz: Anyhow, without looking too closely, I'd give good odds that the -lfoo bits are at the wrong point in the linker command. [16:22] ricotz: Still, not trusting your compiler or linker, and then manually linking all of crt* -lc, etc, is a bit nuts. [16:22] ricotz: So, I'd be less inclined to call it a toolchain bug and more inclined to assume it's a misuse of the toolchain. [16:23] infinity, thanks for taking a glimpse at it, will point it out to upstream [16:24] ricotz: If you happen to have a failed local build tree around, try re-running that last link command with "-lstdc++ -lm -lc -lgcc_s" at the end of the command line. [16:24] infinity, the amd64 didnt fail which i am have locally too [16:25] so a vanilla git build is fine on amd64 [16:27] I'll see if I can reproduce on i386 in a bit. [16:28] infinity, thanks! [16:28] (Probably going to blame crazy toolchain abuse either way, but there's a slim chance it's a gcc/binutils bug) [16:28] Though, probably not, given that it's whining about stack_chk symbols in libc, and I haven't changed glibc since raring. === fenris- is now known as ejat === fenris- is now known as ejat