slangasekwell, the porter chroots I have access to are broken, and I can't reproduce the sagetex autopkgtest failures (uninstallabilities) w/ chdist00:00
slangasekoh, I /can/ reproduce the armhf one00:03
slangasekaha sagemath00:04
tsimonq2slangasek: Speaking of that, I would probably have a fix for this telepathy-logger-qt issue if I had access to an Ubuntu-based porterbox... :P00:19
tsimonq2Maybe spin up an LXD container or something? ;)00:20
-queuebot:#ubuntu-release- New binary: libreoffice [armhf] (cosmic-proposed/main) [1:6.1.0-0ubuntu1] (kubuntu, ubuntu-desktop)01:31
-queuebot:#ubuntu-release- New binary: libreoffice [arm64] (cosmic-proposed/main) [1:6.1.0-0ubuntu1] (kubuntu, ubuntu-desktop)02:37
tsimonq2slangasek: Could you please kick the can on python-fluids/0.1.72-2? Regressed autopkgtests in release, and it'll make sphinx (plus a few friends) migrate.03:01
tsimonq2Hah, so it seems like imexam/0.8.0-2's s390x autopkgtest fails with the same error as the FTBFS for the no-change rebuild...03:06
* tsimonq2 retries against itself in release03:06
* tsimonq2 bets we can kick the can on python-scipy as well.03:09
tsimonq2Same situation with skimage/0.14.0-0ubuntu2/armhf and skimage/0.14.0-0ubuntu2/i386... autopkgtests fail on the same architectures with the same error that it's FTBFS with after a no-change rebuild. Retrying against release...03:11
tsimonq2After that, python-scipy *should* migrate.03:12
tsimonq2...wat? Why does imexam/0.8.0-2/s390x pass? O_o03:23
mwhudsonuh wat03:26
* tsimonq2 is confused at the fact that both my skimage retry and my imexam retry have higher versions than their triggers...03:27
mwhudsontsimonq2: i *think* there's an endian bug in scipy but as so many things ignore the results of tests during build (angryface) it's hard to know what's going on03:28
tsimonq2mwhudson: Maybe if I could get the autopkgtest infra to play nice I'd be able to confirm that. ;P03:28
tsimonq2mwhudson: Oh, right, python-scipy won't be done for a while npw.03:29
tsimonq212:52:46 AM < tsimonq2> slangasek: fflas-ffpack badtest on armhf? Looks like you retried against release: http://autopkgtest.ubuntu.com/packages/f/fflas-ffpack/cosmic/armhf03:30
tsimonq2Perhaps a UK/European/other side of the pond ~ubuntu-release member like apw can look ^03:31
slangasektsimonq2: not sure how an Ubuntu-based ppc64el porterbox would help you, given that I already said it wasn't reproducible for me...03:41
slangasektsimonq2: python-fluids> if by 'kick the can' you mean 'badtest', no; that's a python3.7+scipy-related test failure, I'm retriggering the tests with --all-proposed03:41
tsimonq2Oh hey.03:41
slangasektsimonq2: skimage> mwhudson was already looking at this yesterday and believes this is a regression in scipy.03:42
mwhudsonand i've looked a bit today and am absolutely no further into understanding it btw03:43
tsimonq2slangasek: ppc64el> ack03:43
slangasekfflas, let's see03:43
tsimonq2slangasek: python-fluids, skimage> ack as well.03:43
mwhudsoni am currently of the opinion that scientists should not write software but i'll probably get over that03:43
slangasekyeah, hinting the ffpack03:44
slangasekmwhudson: I return to that assessment every single time I touch a Debian Med or Debian Science package, then I have to remind myself that their goals are different than mine03:45
slangasekand also, that the real bug is for almost any modern scientific software to be built for architectures with a 32-bit address space03:45
mwhudsonbut the tests fail -> disable the tests mentality is ... surely contrary to their goals too03:46
mwhudsonthat's more of a maintainer thing than an upstream thing of course03:46
tsimonq2slangasek: telepathy-logger-qt> If I could figure out what exactly error 25 even *is*, that'd be a good first step, heh...03:46
tsimonq2Just had to look further up, jeez:03:51
tsimonq2dh_acc: abi-compliance-checker -q -l libtelepathy-logger-qt-dev -v1 17.08.0-2build1 -dump debian/libtelepathy-logger-qt-dev.acc -dump-path debian/libtelepathy-logger-qt-dev/usr/lib/powerpc64le-linux-gnu/dh-acc/libtelepathy-logger-qt-dev_17.08.0-2build1.abi.tar.gz returned exit code 203:51
tsimonq2slangasek: How far did your debugging attempts go down the acc rabbit hole? :P03:52
slangasektsimonq2: I ran the autopkgtest and it didn't exit non-zero.03:53
slangasekso there was no rabbit hole03:53
tsimonq2I have a feeling if I passed -debug to acc it'll point directly a problem... if I had a porterbox I could just do it via shell, but I don't have a PPA big enough to copy over qtbase (bug 1789361) so I either need to wait for my PPA expansion request to be processed or use Bileto (which it isn't meant for)...03:55
ubot5bug 1789361 in Auto Package Testing "When queueing tests against a PPA, if a trigger isn't found it should fall back to devel-proposed" [Undecided,New] https://launchpad.net/bugs/178936103:55
tsimonq2So looks like the error can be easily found when retrying against itself, due to my no-change rebuild...03:56
tsimonq2Mkay, so it *is* doable with a PPA then...03:56
tsimonq2Hmm, this is a nice little TODO in the dh_acc code :P "# TODO if next command fails, should output debug/log info?"03:59
acheronuktsimonq2: you can get artifacts output. e.g. https://git.launchpad.net/~kubuntu-packagers/kubuntu-packaging/+git/khtml/tree/debian/tests/acc?h=kubuntu_cosmic_archive04:25
tsimonq2acheronuk: TIL, thanks, but first I want to see if this acc hack works :P04:26
slangasekfigured out why armhf autopkgtest runners seemed to be going awol; we'd had a cowboyed edit to the cronjob to restart the runner systemd units hourly (and why do I not now find the related bug report?), and the lxd worker was recently reprovisioned so lost that bit.04:26
tsimonq2acheronuk: Darn, just looked and it didn't... I'll try that.04:26
slangasekso we should be back to not having a 3h window where armhf queues are not progressing.04:26
tsimonq2acheronuk: This is actually super sweet... thanks.04:27
slangasekhmmm that was https://bugs.launchpad.net/auto-package-testing/+bug/1735878 and it was reported closed.04:27
ubot5Ubuntu bug 1735878 in Auto Package Testing "worker auto restarting should be handled by systemd unit, not by cronjob" [Medium,Fix released]04:27
acheronuktsimonq2: say that if if gives useful output for this ;)04:28
tsimonq2acheronuk: hehe04:28
tsimonq2acheronuk: It very likely will, it seems.04:29
slangasek... so the pkill was removed but the crontabs were never actually redeployed04:29
slangasekI think this actually means the lxd worker wasn't actually redeployed recently... it was redeployed in May which means this was broken for a while until it got noticed04:33
tsimonq2slangasek: So I found our error...04:49
tsimonq2Creating library ABI dump ...04:49
tsimonq2ERROR: can't pack the ABI dump: Cannot allocate memory04:49
tsimonq2See https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-cosmic-tsimonq2-upload-testing/cosmic/ppc64el/t/telepathy-logger-qt/20180829_044637_96ed1@/artifacts.tar.gz04:49
tsimonq2If it's truly an OOM issue, I'm not sure what the next step is here.04:53
tsimonq2...can someone with access to the infra cowboy a resource bump or something weird like that?04:53
slangasektsimonq2: well, in theory that gives me a path to reproduce it, then I can look at what's taking memory and if it's fixable05:27
tsimonq2slangasek: Awesome.05:30
slangasekfwiw the ppc64el runners have 1.5GiB of total memory05:42
slangasekacc is implemented in perl and it uses this much memory? sad05:50
tsimonq2I'm willing to bet it's glibc's fault, fwiw. :P05:54
slangasekhighwatermark of memory usage for a-c-c on ppc64el is 768960k according to /usr/bin/time05:54
slangasektsimonq2: I've run the tests every which way; -proposed glibc is not a requirement for reproducing05:55
tsimonq2slangasek: But if there's 1.5 GB *total*, that means if there's any other test running it'll oom?05:56
slangasekit's 1 autopkgtest per VM05:56
slangasekbut other processes take up memory, so I'm not sure what the effective available memory is05:56
slangasektsimonq2: I just traced a running instance of the autopkgtest and there was never less than 840MB free on the node06:04
tsimonq2I'm out of ideas if oom isn't it.06:05
tsimonq2Maybe an acc bug if it's not that..06:05
slangasekso I *can* make a-c-c OOM, but not in an autopkgtest runner context06:07
slangasekthe actual OOM happens when calling 'zip', hmm06:09
tsimonq2Sounds about right: https://github.com/lvc/abi-compliance-checker/blob/master/abi-compliance-checker.pl#L936606:10
slangasekthat's what I was going by06:10
tsimonq2tar does seem more likely, which the code below has.06:10
slangasekah, zip is only used on windows, ok06:11
tsimonq2Because that code's ran on Windows.06:11
slangasekah of course06:16
slangasekclone(child_stack=NULL, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x75b7f8544520) = -1 ENOMEM (Cannot allocate memory)06:16
slangasekbecause forking tar means creating a copy of the process first and that process is oversized :P06:16
slangasekfreeing memory before invoking tar would fix it06:16
slangasekand the failure should be generically reproducible with a cgroup (but not with ulimits, which was my first attempt)06:17
acheronukslangasek: suddenly getting some armhf tests fail with "mkdir: cannot create directory ‘//.config’: Permission denied"06:20
acheronukany issue you know of?06:20
acheronukslangasek: ok. thanks. retried 1, and it passed. retrying another but takes a while.06:22
=== ogra_ is now known as ogra
-queuebot:#ubuntu-release- Unapproved: brltty (bionic-proposed/main) [5.5-4ubuntu2 => 5.5-4ubuntu2.0.1] (core)07:59
-queuebot:#ubuntu-release- New binary: linux-signed-oem [amd64] (bionic-proposed/main) [4.15.0-1018.21] (kernel)08:50
-queuebot:#ubuntu-release- New binary: linux-signed [amd64] (xenial-proposed/main) [4.4.0-135.161] (core, kernel)08:59
-queuebot:#ubuntu-release- New: accepted linux-signed-oem [amd64] (bionic-proposed) [4.15.0-1018.21]09:00
-queuebot:#ubuntu-release- New: accepted linux-signed [amd64] (xenial-proposed) [4.4.0-135.161]09:01
-queuebot:#ubuntu-release- Unapproved: avahi (bionic-proposed/main) [0.7-3.1ubuntu1 => 0.7-3.1ubuntu1.1] (core)09:26
=== tomwardill_ is now known as tomwardill
-queuebot:#ubuntu-release- New binary: python-cffi [amd64] (cosmic-proposed/main) [1.11.5-2] (kubuntu, ubuntu-desktop, ubuntu-server)10:38
rbalintLocutusOfBorg, it seems erlang now can be a sync, how do you feel about updating it for cosmic? Debian changes are mostly bugfix releases11:56
LocutusOfBorgrbalint, ack12:30
LocutusOfBorgcan you please double check why is it a sync?12:31
-queuebot:#ubuntu-release- Unapproved: glib2.0 (bionic-proposed/main) [2.56.1-2ubuntu1 => 2.56.2-0ubuntu0.18.04.1] (core) (sync)13:31
-queuebot:#ubuntu-release- New binary: linux-signed-lts-xenial [amd64] (trusty-proposed/main) [4.4.0-135.161~14.04.1] (kernel)14:23
-queuebot:#ubuntu-release- New: accepted linux-signed-lts-xenial [amd64] (trusty-proposed) [4.4.0-135.161~14.04.1]14:42
-queuebot:#ubuntu-release- New: accepted cmake-vala [sync] (cosmic-proposed) [1-1]14:52
-queuebot:#ubuntu-release- New: accepted ironic [amd64] (cosmic-proposed) [1:11.1.0-0ubuntu3]14:53
-queuebot:#ubuntu-release- New: accepted libminc [arm64] (cosmic-proposed) [2.4.03-1]14:53
-queuebot:#ubuntu-release- New: accepted libminc [i386] (cosmic-proposed) [2.4.03-1]14:53
-queuebot:#ubuntu-release- New: accepted libminc [s390x] (cosmic-proposed) [2.4.03-1]14:53
-queuebot:#ubuntu-release- New: accepted libreoffice [arm64] (cosmic-proposed) [1:6.1.0-0ubuntu1]14:53
-queuebot:#ubuntu-release- New: accepted libreoffice [i386] (cosmic-proposed) [1:6.1.0-0ubuntu1]14:53
-queuebot:#ubuntu-release- New: accepted libreoffice [s390x] (cosmic-proposed) [1:6.1.0-0ubuntu1]14:53
-queuebot:#ubuntu-release- New: accepted murano-agent [amd64] (cosmic-proposed) [1:3.5.1-0ubuntu2]14:53
-queuebot:#ubuntu-release- New: accepted panko [amd64] (cosmic-proposed) [5.0.0-0ubuntu2]14:53
-queuebot:#ubuntu-release- New: accepted ironic [amd64] (cosmic-proposed) [1:11.1.0-0ubuntu2]14:53
-queuebot:#ubuntu-release- New: accepted libminc [armhf] (cosmic-proposed) [2.4.03-1]14:53
-queuebot:#ubuntu-release- New: accepted libreoffice [amd64] (cosmic-proposed) [1:6.1.0-0ubuntu1]14:53
-queuebot:#ubuntu-release- New: accepted libreoffice [ppc64el] (cosmic-proposed) [1:6.1.0-0ubuntu1]14:53
-queuebot:#ubuntu-release- New: accepted node-csv-spectrum [sync] (cosmic-proposed) [1.0.0-2]14:53
-queuebot:#ubuntu-release- New: accepted python-cffi [amd64] (cosmic-proposed) [1.11.5-2]14:53
-queuebot:#ubuntu-release- New: accepted libminc [amd64] (cosmic-proposed) [2.4.03-1]14:53
-queuebot:#ubuntu-release- New: accepted libreoffice [armhf] (cosmic-proposed) [1:6.1.0-0ubuntu1]14:53
-queuebot:#ubuntu-release- New: accepted panko [amd64] (cosmic-proposed) [5.0.0-0ubuntu3]14:53
-queuebot:#ubuntu-release- New: accepted libminc [ppc64el] (cosmic-proposed) [2.4.03-1]14:53
-queuebot:#ubuntu-release- New: accepted llvm-toolchain-7 [sync] (cosmic-proposed) [1:7~+rc2-1~exp1]14:53
-queuebot:#ubuntu-release- New binary: node-csv-spectrum [amd64] (cosmic-proposed/universe) [1.0.0-2] (no packageset)14:55
-queuebot:#ubuntu-release- New: accepted node-csv-spectrum [amd64] (cosmic-proposed) [1.0.0-2]14:55
-queuebot:#ubuntu-release- New binary: linux-signed-gcp [amd64] (bionic-proposed/main) [4.15.0-1019.20] (no packageset)15:06
ahasenackanybody looking at the twisted migration?15:06
ahasenackI could help perhaps15:07
ahasenackcurrent version in cosmic (17.x) fails with py3.7, the new one has that fixed15:07
ahasenackor at least that particular failure15:07
ahasenackugh, http://autopkgtest.ubuntu.com/packages/t/twisted/cosmic/amd64 has one lonely green only, from 201715:08
-queuebot:#ubuntu-release- New: accepted linux-signed-gcp [amd64] (bionic-proposed) [4.15.0-1019.20]15:08
-queuebot:#ubuntu-release- New binary: linux-signed-azure [amd64] (bionic-proposed/main) [4.15.0-1023.24] (kernel)16:02
-queuebot:#ubuntu-release- New binary: linux-signed-hwe [amd64] (xenial-proposed/main) [4.15.0-34.37~16.04.1] (kernel)16:02
-queuebot:#ubuntu-release- New binary: linux-signed-hwe [ppc64el] (xenial-proposed/main) [4.15.0-34.37~16.04.1] (kernel)16:02
-queuebot:#ubuntu-release- New: accepted linux-signed-hwe [amd64] (xenial-proposed) [4.15.0-34.37~16.04.1]16:11
-queuebot:#ubuntu-release- New: accepted linux-signed-hwe [ppc64el] (xenial-proposed) [4.15.0-34.37~16.04.1]16:11
-queuebot:#ubuntu-release- Unapproved: ubuntu-release-upgrader (bionic-proposed/main) [1:18.04.24 => 1:18.04.25] (core)16:12
-queuebot:#ubuntu-release- Unapproved: accepted ubuntu-release-upgrader [source] (bionic-proposed) [1:18.04.25]16:23
-queuebot:#ubuntu-release- Unapproved: accepted ubuntu-image [source] (bionic-proposed) [1.4+18.04ubuntu2]16:25
infinityxnox, bdmurray: ^--- Can we verify that (a) s390x now works and (b) snap migration on amd64 still works?  If both are actually tested, I'm happy with that SRU being fasttracked post-verification.16:25
bdmurrayinfinity: Sure I can test the snap migration still working bit.16:27
-queuebot:#ubuntu-release- Unapproved: accepted ubuntu-image [source] (xenial-proposed) [1.4+16.04ubuntu2]16:29
infinityslangasek: Did you look at all into the python-cffi/i386 thing?16:52
infinityslangasek: Just to make sure glibc isn't off the deepend, I did this: https://paste.ubuntu.com/p/VRXRzT9gKK/16:52
slangasekinfinity: no; did mwhudson mutter about it? I don't recall now16:52
infinityslangasek: So now very curious where it's pulling NaN from.16:52
infinityslangasek: Oh.  So it does that cos(1.23) in two dlopen tests, and it's only one (dlopen_filename) that's returning NaN, so I think what it's actually doing is failing to dlopen by path.16:58
infinityWhich also seems weird, but slightly easier to debug.16:58
xnoxinfinity, do-release-upgrade xenial->bionic fails; do-release-upgrade --proposed xenial->bionic goes past that point and is downloading all the things for me. So I can reproduce original bug; and proposed stuff fixes everything.16:59
xnoxon s390x16:59
infinityxnox: Shiny.  Tell the bug, please.17:00
infinityxnox: Though maybe without the --proposed thing, don't want IBM people telling their customers that's how to upgrade. :P17:00
xnoxinfinity, the bug is commented17:02
xnoxslangasek, ubuntu-cloud-keyring binary from ubuntu-keyring needs to move from universe to main; it is seeded in bionic and cosmic. cosmic is easy.17:31
xnoxslangasek, but i don't know if there is components-missmatch report for bionic, or how to copy/publish it from bionic-release/universe to bionic-security|updates/main ?17:31
infinityslangasek: There is no c-m for stable series.  And the copy trick needs to be done by an AA.17:33
infinityxnox: ^17:33
infinityNo can brain.17:33
xnoxinfinity, yeah, that's what a vague recalled, but this happens to me only once every 4 years, so details are fuzzy.17:34
xnoxanyway volleyball17:34
infinityxnox: I'm a bit confused.17:35
infinityxnox: Oh.  Less confused now.17:36
-queuebot:#ubuntu-release- New binary: llvm-toolchain-7 [ppc64el] (cosmic-proposed/universe) [1:7~+rc2-1~exp2] (no packageset)17:37
-queuebot:#ubuntu-release- New: accepted llvm-toolchain-7 [ppc64el] (cosmic-proposed) [1:7~+rc2-1~exp2]17:41
slangasekinfinity: btw telepathy-logger-qt/ppc64el is definitely ignorable, if we get to that point.  hitting the threshold where abi-compliance-checker uses > 50% of free memory in the testbed is not a regression in any of the involved packages17:55
infinityslangasek: Fun.17:56
slangasek(I'm trying to work around the failure in a-c-c, but perl)17:57
infinityOne has to write some spastically bad Perl to eat all your RAM.17:57
infinityhttps://accidentallyquadratic.tumblr.com/ ?17:57
slangasekit's using Data::Dumper for the objects that represent the complete ABI of a very dense C++ lib17:58
slangasekso yes, that's less than ideal17:58
slangasekbut I'm not going to refactor that code to teach it how to stream to a file17:58
bdmurrayinfinity: xnox and I have verified bug 178766818:03
ubot5bug 1787668 in Ubuntu on IBM z Systems "ubuntu-release-upgrader crashed with KeyError: "The cache has no package named 'ubuntu-desktop'"" [High,In progress] https://launchpad.net/bugs/178766818:03
-queuebot:#ubuntu-release- New binary: linux-signed-azure [amd64] (xenial-proposed/main) [4.15.0-1023.24~16.04.1] (kernel)18:06
infinitybdmurray: Danke.  Releasing.18:31
-queuebot:#ubuntu-release- New binary: llvm-toolchain-7 [amd64] (cosmic-proposed/universe) [1:7~+rc2-1~exp2] (no packageset)19:36
-queuebot:#ubuntu-release- Unapproved: fwupdate (cosmic-proposed/main) [12-3ubuntu1 => 12-3ubuntu1] (core)20:40
-queuebot:#ubuntu-release- Unapproved: fwupdate (cosmic-proposed/main) [12-3ubuntu1 => 12-3ubuntu1] (core)20:40
-queuebot:#ubuntu-release- Unapproved: fwupdate (cosmic-proposed/main) [12-3ubuntu1 => 12-3ubuntu1] (core)20:40
-queuebot:#ubuntu-release- Unapproved: fwupdate (cosmic-proposed/main) [12-3ubuntu1 => 12-3ubuntu1] (core)20:41
-queuebot:#ubuntu-release- Unapproved: fpc (bionic-proposed/universe) [3.0.4+dfsg-18ubuntu1 => 3.0.4+dfsg-18ubuntu2] (no packageset)20:42
cyphermoxugh, fwupdate probably shouldn't have been uploaded at all21:08
cyphermoxslangasek: infinity: do you recall what the state is for fwupdate vs. fwupd?21:09
slangasekcyphermox: I've said that going forward we will not sign uefi binaries for the unseeded one; however I see that both are in main, I could've sworn I had already processed that demotion21:13
slangasekI did remove fwupdate-signed21:14
slangasekfwupdate needs fixed to not spit out uefi signing requests21:14
mwhudsoninfinity, slangasek: i haven't looked at cffi21:15
-queuebot:#ubuntu-release- Unapproved: rejected fwupdate [amd64] (cosmic-proposed) [12-3ubuntu1]21:15
-queuebot:#ubuntu-release- Unapproved: rejected fwupdate [armhf] (cosmic-proposed) [12-3ubuntu1]21:15
-queuebot:#ubuntu-release- Unapproved: rejected fwupdate [arm64] (cosmic-proposed) [12-3ubuntu1]21:15
slangasekcyphermox: so, I've rejected the above21:15
-queuebot:#ubuntu-release- Unapproved: rejected fwupdate [i386] (cosmic-proposed) [12-3ubuntu1]21:15
mwhudsoni can, in a couple hours, if you want21:15
mwhudsoni still hate everything about scipy and am going to ignore that today so i actually achieve something21:16
cyphermoxslangasek: ok21:17
cyphermoxslangasek: can you block shim-signed 1.37 in proposed? LP is timing out here...21:17
cyphermoxmaybe I'm being extra paranoid, but I wouldn't mind just breaking my system if it turns out I made some error.21:18
cyphermoxnvm, got LP to let me do what I want after all21:21
xnoxslangasek, infinity - so can one of you please promote ubuntu-cloud-kerying into bionic-updates/main somehow?21:23
xnoxslangasek, infinity - so can one of you please promote ubuntu-cloud-keyring into bionic-updates/main somehow?21:23
slangasekxnox: what references it in bionic?21:23
xnoxsame thing as in cosmic, supported-misc-servers21:24
xnoxbut no reports...21:24
slangasekand promoted21:28
slangasekcyphermox: I see shim-signed having migrated to cosmic, was that what you expected?21:41
cyphermoxbut it's fine22:01
cyphermoxeverything can migrate.22:02
xnoxslangasek, linux-image-generic|lowlatency|oem do not depend on intel-microcode in bionic. still.22:03
xnoxthey appear to do so in cosmic.22:04
xnoxslangasek, is this expected?!22:04
slangasekxnox: which version? I see the dep from linux-image-generic in bionic22:05
xnoxreverse-depends -r bionic intel-microcode must be telling me lies then!22:05
xnoxlet me finish upgrading this instance and double check with apt22:06
slangasekxnox: does reverse-depends know about -updates?22:06
xnoxslangasek, yeah, it's there, thanks!22:07

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