| micahg | arand: BTW, abiword 2.8.4 just hit maverick | 00:00 |
|---|---|---|
| crimsun | thank goodness :-) | 00:01 |
| ajmitch | does 2.8.4 fix a few problems? | 00:01 |
| arand | Nice. | 00:02 |
| micahg | s/hit/uploaded/ | 00:02 |
| crimsun | ajmitch: yes | 00:03 |
| arand | It fixes the crash-on-help one, although that should hopefully be fixed with my svn cherrypick. | 00:03 |
| arand | ...In lucid as well | 00:04 |
| crimsun | not least of which is the xmlCleanupParser() madness | 00:04 |
| crimsun | which historically makes pulse look bad, but it's in fact people not bloody reading the libxml2 API | 00:04 |
| ajmitch | uh | 00:04 |
| ajmitch | how does that have anything to do with pulse? :) | 00:04 |
| arand | Uh, what? Was pulse involved in this? | 00:05 |
| crimsun | calling xmlCleanupParser() multiple times tramples all over the pulse client's TLS vars | 00:05 |
| crimsun | as a result, the offending pulse client (the program abusing xmlCleanupParser()) segfaults, with the stack trace pointing to pulse | 00:05 |
| ajmitch | oh that must have been a bundle of fun | 00:06 |
| crimsun | I combed through the archive toward the end of the Lucid cycle to fix this | 00:06 |
| arand | One good thing with a new abiword is that 2.8.2 seems to FTBFS on my MM system as well :/ | 00:13 |
| crimsun | would a member of ~ubuntu-release please unsub the team from bug 568619 so I won't spam everyone? | 01:01 |
| ubottu | Launchpad bug 568619 in gorm.app "FFe: Sync gorm.app 1.2.10-1 (universe) from Debian sid (main)" [Wishlist,New] https://launchpad.net/bugs/568619 | 01:01 |
| === nobawk|away is now known as nobawk | ||
| === nobawk is now known as nobawk|away | ||
| === nobawk|away is now known as nobawk | ||
| AnAnt | Hello, how can I fix this bashism: echo -e "text text \n text text \t text ..." ? | 03:33 |
| AnAnt | if I remove -e, it would work fine with dash, but not with bash | 03:33 |
| crimsun | AnAnt: check the xpg_echo shell option and act appropriately | 03:58 |
| crimsun | AnAnt: in bash, that is | 03:58 |
| AnAnt | crimsun: thanks, I used printf instead | 03:59 |
| imbrandon | evening all | 05:07 |
| ajmitch | hi imbrandon | 05:13 |
| imbrandon | heya ajmitch | 05:14 |
| imbrandon | so its kinda cool, i picked up this tiny 11inch dell from like 1999, its funny its like its got all the features of the new "mini"s just slower | 05:17 |
| imbrandon | got lucid loaded on it, but it dosent like gnome/unity much , toooooo slugish , like firefox takes 2 minutes to load | 05:18 |
| imbrandon | lol | 05:18 |
| imbrandon | so i got lucid but i apt-get installed fluxbox + a few other lighter apps and its great little mobile machine to use to ssh into other boxes etc | 05:18 |
| imbrandon | for free ;) | 05:19 |
| imbrandon | its 700mhz, so its not "terrible" | 05:20 |
| ajmitch | not bad | 05:22 |
| imbrandon | i think most of the slowness is the only 128mb ram, it maxes out at 256 says dells website if i can run accross a 256mb pc-100 sodimm at somepoint | 05:24 |
| imbrandon | sooo whats new in motu ? | 05:25 |
| imbrandon | ajmitch: http://support.dell.com/support/edocs/systems/latl400/en/ug/specs.htm | 05:27 |
| imbrandon | heh | 05:27 |
| === nobawk is now known as nobawk|away | ||
| === nobawk|away is now known as nobawk | ||
| _Andrew | #debian | 06:35 |
| _Andrew | woops | 06:35 |
| _Andrew | Anyone know why when I do "dh_install --sourcedir=debian/tmp --fail-missing" and inside my install script it has the following.. "usr/bin" my package fails to find something installed in debian/tmp/usr/bin/xyz ? | 07:00 |
| persia | _Andrew: I may be mistaken, but I believe --fail-missing only fails if there are things in the temporary directory that are unallocated to any package, rather than if there are things missing from the temporary directory. | 07:11 |
| _Andrew | Here's the error message.. "debian/tmp/usr/bin/cpack exists in debian/tmp/ but is not installed to anywhere" | 07:13 |
| dholbach | good morning | 07:14 |
| _Andrew | I'm using compatibility level 6 and have tried "debian/tmp/usr/bin", "debian/tmp/usr/bin/*", "usr/bin", "usr/bin/*" and probably some other combinations. | 07:14 |
| _Andrew | It never seems to find it in the .install file | 07:15 |
| persia | One of the changes in compatibility level 7 was some new logic to make the .install files make more sense :) | 07:16 |
| persia | I tend to list stubborn files explicitly and individually. | 07:17 |
| _Andrew | Can't because this package is for hardy | 07:18 |
| _Andrew | Otherwise yes, I'd bump up the compat in a heartbeat :D | 07:19 |
| persia | How did it work before you started fixing it? | 07:19 |
| _Andrew | I specifically lists all the programs.. I think I already tried that. but I'll give it another go | 07:20 |
| _Andrew | It** | 07:20 |
| persia | That seems about right, from my memory. Best practice when working with older packages is to touch the absolute minimum. | 07:20 |
| === nobawk is now known as nobawk|away | ||
| === nobawk|away is now known as nobawk | ||
| === nobawk is now known as nobawk|away | ||
| === ApOgEE__ is now known as ApOgEE | ||
| === wgrant_ is now known as wgrant | ||
| G | It's been a while since I've done Debian style packaging, but I'm looking to get back into it, I've got a package that has downloaded as: foo-0.2.1-source.tar.gz and extracts as foo-0.2.1-source, the example at https://wiki.ubuntu.com/PackagingGuide/Howtos/PackagingFromScratchHelloDebhelper shows 'tar -xf hello-x.y.tar.gz hello-x.y' so my question is, does the -source matter, or is that something I have to rename out? | 10:27 |
| persia | The name of the directory in the upstream tarball matters not at all. | 10:28 |
| Rhonda | It neither matters for dpkg-source, it though issues a warning which can get ignored. :) | 10:29 |
| Rhonda | When packaging inside a VCS I usually have the directory only named VCS for consistency issues. | 10:30 |
| umang | Hi, I maintain a package on Debian that entered sid after the Lucid DIF. If I want it to be imported into maverick then do I have to file a bug or will it automatically happen after MM's DIF? | 10:31 |
| G | persia: Rhonda: great, thats for that (and thanks for the hint abotu the warning) | 10:31 |
| persia | umang: It ought enter the NEW queue automatically. | 10:32 |
| Rhonda | umang: It hopefully will automatically happen _before_ MM's DIF. :) | 10:33 |
| Rhonda | … because after DIF there isn't any automatism for that anymore. | 10:33 |
| umang | persia, Rhonda: so I don't need do anything for it to be included in maverick? | 10:34 |
| persia | umang: It wouldn't hurt to track the progress through https://launchpad.net/ubuntu/maverick/+queue but it ought require no action on your part. | 10:35 |
| umang | persia, I'm not able to find it under any of the statuses. So I presume it will enter one of those queues later? | 10:37 |
| persia | Right. The archive admins usually wait until the current packages sync has made significant progress before syncing the new packages. | 10:37 |
| umang | persia, Thanks! :) | 10:38 |
| persia | I'd recommend checking again in a week or so. if you still haven't seen it in any of the queues by a few weeks before DIF, then it may merit investigation. | 10:38 |
| persia | With luck, you'll only ever notice it in DONE. | 10:38 |
| umang | persia, Sure, I'll do that. :) | 10:40 |
| * Rhonda still waits for packages.ubuntu.com to display maverick … | 10:45 | |
| Rhonda | … and for wesnoth-1.8 syned into maverick. ;) | 10:45 |
| mok0 | Any schroot users? I am getting loads of messages of the following type, and it is starting to annoy me. I already tried to edit /etc/schroot/schroot.conf, but to no avail. | 12:15 |
| mok0 | W: line 19 [karmic-0f03ab72-3520-4df4-a561-e3d55c218892]: Deprecated key 'run-exec-scripts' used | 12:15 |
| mok0 | I: This option will be removed in the future; please update your configuration | 12:15 |
| mok0 | W | 12:15 |
| === nobawk|away is now known as nobawk | ||
| azeem | mok0: yeah, it's pretty annoying | 12:20 |
| geser | mok0: do you have a "run-exec-scripts" line in your schroot.conf? | 12:20 |
| mok0 | I removed it | 12:20 |
| geser | and still this warning? | 12:21 |
| azeem | mok0: karmic-0f03ab72-3520-4df4-a561-e3d55c218892 means it's from a session | 12:21 |
| persia | The only way I found to make that go away was to regenerate the schroots. | 12:21 |
| azeem | mok0: so umount those bind mounts and close all sessions | 12:21 |
| mok0 | Lemme see | 12:21 |
| azeem | (I don't understand why schroot needs sessions anyway) | 12:21 |
| persia | `schroot -e -c ${SESSIONID}` automates that. | 12:21 |
| persia | azeem: Because sometimes you want to run several in parallel, or you might want to have persistent schroots for some reason, etc. | 12:22 |
| azeem | hrm | 12:22 |
| mok0 | There are no mounted work volumes | 12:22 |
| mok0 | persia: what is SESSIONID? | 12:22 |
| persia | mok0: I'm not entirely sure. When I use schroot with lvm, the necessary magic value appears in the lvs output. | 12:23 |
| azeem | mok0: probably karmic-0f03ab72-3520-4df4-a561-e3d55c218892 or without the karmic- | 12:23 |
| * persia is still learning about schroot-without-lvm | 12:23 | |
| persia | I think it's with the karmic | 12:23 |
| azeem | persia: those should be a specially requested operation though, not the default | 12:23 |
| azeem | I assume sbuild is to blame here | 12:23 |
| persia | azeem: Kinda. So, when sbuild crashes, the schroot doesn't close. this can be useful in many ways, but also annoying sometimes. | 12:24 |
| mok0 | There are similar lines for my jaunty builder | 12:24 |
| persia | By default, schroot opens a session, and closes it when complete. This is essential if one has configured schroots to be used on a multiuser box, or if one wants to do parallel builds with sbuild, etc. | 12:25 |
| mok0 | You think I should enter the schroots, and make the change to /etc/schroot... etc there? | 12:25 |
| persia | Keeping the session open after exit requires a special command-line argument. | 12:25 |
| persia | mok0: That may work: regeneration is likely cleaner. | 12:25 |
| mok0 | persia: yeah, sometimes the roots stay around even after the build finishes, but that's not the case for me now | 12:25 |
| mok0 | persia: I'm not sure I understand "regenerate"... run mk-sbuild again? | 12:27 |
| persia | What I did was remove the stanzas from schroot.conf, remove the storage (either LV or /var/lib/schroot/... ) and then rebuild them from scratch. | 12:27 |
| persia | This is not necessarily the optimal solution. | 12:28 |
| mok0 | persia: I have so many schroots | 12:28 |
| mok0 | 13... | 12:28 |
| * persia wrote one-line wrapper script around mk-sbuild that day | 12:28 | |
| mok0 | Hmm :-) | 12:29 |
| persia | for i in dapper hardy jaunty karmic lucid maverick; do for j in amd64 armel i386 powerpc; do mk-sbuild --arch=$j $i; done; done takes care of Ubuntu. Debian is similar. | 12:30 |
| * persia may have mistyped syntax | 12:30 | |
| persia | Note that this will take a while, and it is advised to be network-near a mirror of at least the buildd stuff. | 12:31 |
| persia | (and use --debootstrap-mirror=...) | 12:31 |
| mok0 | persia: fortunately, I have an apt-proxy | 12:31 |
| mok0 | on the local network | 12:32 |
| persia | That works too, at least for some subset :) | 12:32 |
| mok0 | persia: can't remember... do you need to edit schroot.conf first? | 12:32 |
| persia | I just deleted everything that I was replacing from there. | 12:33 |
| persia | Also remember to remove the backing storage configuration, as mk-sbuild is a bit fragile about that. | 12:33 |
| mok0 | *sigh* | 12:33 |
| * mok0 procrastinates: I think I'll do it... sometime | 12:34 | |
| mok0 | In fact, I might just figure out what part of sbuild issues those messages and zap them :-) | 12:35 |
| persia | I'm sure it's something inside the schroot. if you figure out what, probably best to write an exec script that cleans up elderly schroots on intialisation. | 12:35 |
| mok0 | Hm. Nothing inside the schroot | 12:39 |
| === nobawk is now known as nobawk|away | ||
| === ttx_ is now known as ttx | ||
| === txwikinger2 is now known as txwikinger | ||
| G | does lucid have a buggy autoconf? it's generated a configure script with includedir=$PREFIX/include, instead of includedir=$PREFIX/usr/include, is there a standard fix for that? | 13:13 |
| Rhonda | G: What's wrong with that? includedir=$PREFIX/include seems to be correct to me? | 13:15 |
| Rhonda | … because usually PREFIX itself is /usr | 13:15 |
| G | is there any way I can verify what debuild is executing (such as -v) so I can see what it's passing to autoconf.sh? | 13:16 |
| azeem | G: try DH_VERBOSE=1 | 13:17 |
| azeem | (wild guess) | 13:17 |
| G | azeem: ohhh thanks | 13:18 |
| G | (it's been a while) | 13:18 |
| === nobawk|away is now known as nobawk | ||
| G | hmmm, this is confusing I can't seem to get any debhelper logging :S | 13:30 |
| piju | http://9w2pju.blogspot.com/2010/05/ubuntu-hams-t-shirt-artwork-from-9w2pju.html | 13:32 |
| piju | what do you think? | 13:32 |
| === nobawk is now known as nobawk|away | ||
| mok0 | piju, Interesting. How is HAM radio related to Ubuntu? Just curious | 13:51 |
| mok0 | Internet over HAM? | 13:51 |
| === nobawk|away is now known as nobawk | ||
| piju | mok0; join #ubuntu-hams | 13:53 |
| mok0 | piju: I'm not a HAM operator. Just curious | 13:54 |
| piju | ubuntu got lots of ham radio applications | 13:54 |
| piju | try apt-cache search ham radio | 13:54 |
| piju | shackbox is a distro for ham radio, made from ubuntu distro | 13:55 |
| piju | you can use ubuntu for ham radio | 13:55 |
| porthose | http://packages.ubuntu.com/lucid/hamradio/ | 13:55 |
| piju | morse code, APRS, echolink (VOIP), etc | 13:55 |
| piju | design your antenna, and many more | 13:55 |
| mok0 | Woa | 13:55 |
| mok0 | Heh | 13:55 |
| piju | where do you come from ? | 13:56 |
| mok0 | Denmark. in this country, I think it is pretty near impossible to get a license unless you want to spend every free hour | 13:56 |
| piju | you can ask your local ham radio club, how to get a ham license | 13:56 |
| piju | we can talk over ham radio | 13:57 |
| mok0 | piju: You need to learn morse code | 13:57 |
| piju | im from Malaysia | 13:57 |
| piju | mok0; in the US, no more morse code test | 13:57 |
| mok0 | piju, what's wrong with IRC :-) | 13:57 |
| mok0 | piju: yeah I read that | 13:57 |
| piju | there is lot of open source developer who are ham radio too | 13:58 |
| piju | #hamradio | 13:58 |
| mok0 | piju: my interest in ham would be more like tunneling internet over it | 13:58 |
| piju | yes, we have echolink for that | 13:58 |
| piju | VOIP | 13:58 |
| piju | i can talk using my pc to a man who are driving in England | 13:58 |
| piju | and vice versa | 13:58 |
| mok0 | piju, somehow I'm not impressed :-) | 13:59 |
| piju | you can talk to astronout if you a ham radio | 13:59 |
| piju | mok0; yes, i know that. you are generation of computer, not radio | 13:59 |
| mok0 | piju: errh | 13:59 |
| piju | even most advanced military still use radio | 13:59 |
| piju | international space station too | 14:00 |
| mok0 | piju: encrypted probably | 14:00 |
| piju | encrypted ? | 14:01 |
| piju | yes. for sure. military will use encryption | 14:01 |
| mok0 | piju: I would assume that the military encrypt their radio communication. I may be wrong | 14:01 |
| mok0 | ah | 14:02 |
| piju | but, not for ham radio | 14:02 |
| mok0 | piju: I see | 14:02 |
| piju | ISS also not encrypting their broadcasting | 14:02 |
| piju | you can listen to them | 14:02 |
| mok0 | piju: that would be research thought | 14:02 |
| mok0 | though | 14:02 |
| piju | mok0; i know a ham radio from Denmark | 14:02 |
| mok0 | piju: where? | 14:02 |
| mok0 | where in dk I mean | 14:03 |
| piju | i dont remember | 14:03 |
| piju | but he is a ham | 14:03 |
| mok0 | I think ham guys are pretty rare here | 14:03 |
| mok0 | b/c of the steep requirements for getting a license | 14:04 |
| === mathiaz_ is now known as mathiaz | ||
| tarzeau | how can i get someone to have a look at a package in revu? | 14:10 |
| Rhonda | By dropping in a hint with more content, like URL or package name or similar. | 14:10 |
| tarzeau | http://revu.ubuntuwire.com/revu1-incoming/fracplanet-1005071518/fracplanet_0.4.0-1ubuntu3.dsc it's a tool to create 3d fractal planets | 14:11 |
| tarzeau | http://revu.ubuntuwire.com/revu1-incoming/sea-defender-1005071521/sea-defender_0.9-1.dsc is a small recreational game where you defend oil tankers being hit by torpedos | 14:12 |
| tarzeau | Rhonda: thanks for the hint | 14:12 |
| G | so, last question for now (I'm about to head to bed), should I run this 'autogen.sh' and have it as a patch to the build, or should I add it to the debian/rules file? | 14:13 |
| tarzeau | G: which package/software is it? | 14:15 |
| G | tarzeau: it's a library (dep for an end package I'd like to do) qpixmap | 14:16 |
| G | tarzeau: the tarball comes with an autogen.sh file (they released straight from git) | 14:16 |
| tarzeau | G: i see | 14:17 |
| G | Doing it as a patch would be cleaner right? | 14:17 |
| tarzeau | G: the cleanest would be upstream releasing without you having to run it | 14:17 |
| G | true, but for now (while I contact upstream) | 14:18 |
| tarzeau | G: i don't know, all my upstreams cooperate on these things with me | 14:18 |
| G | so there isn't a given standard | 14:18 |
| tarzeau | G: i think that's the best i know about it: http://sam.zoy.org/lectures/20050910-debian/ | 14:22 |
| tarzeau | G: what does your autogen.sh do? | 14:23 |
| eagles0513875 | hey guys i have a question relating to a piece of software that has made some others in the current lucid repos obsolete | 14:24 |
| tarzeau | yes? | 14:24 |
| eagles0513875 | can anyone point or direct me as to the proper procedure in getting something into the next release | 14:24 |
| tarzeau | eagles0513875: what is it, url? | 14:24 |
| eagles0513875 | tarzeau: mysql workbench | 14:24 |
| eagles0513875 | it has replace mysql admin and query browser which are still in the repos | 14:24 |
| eagles0513875 | those have reached their eol | 14:25 |
| eagles0513875 | http://dev.mysql.com/downloads/workbench/5.2.html#downloads tarzeau | 14:25 |
| eagles0513875 | im more then willing to repackage it if need be as i need to learn about packaging and want to learn about it | 14:26 |
| tarzeau | eagles0513875: i have no idea. i wonder what the debian people do with the mysql packages... | 14:26 |
| eagles0513875 | tarzeau: that package was created upstream at mysql | 14:26 |
| tarzeau | i really don't know | 14:27 |
| G | tarzeau: basically everything except clean the kitchen sink (generates ./configure, Makefile's etc | 14:27 |
| eagles0513875 | G: any idea then how i should go about getting a new package into the new release and possibly backported to lucid | 14:28 |
| tarzeau | eagles0513875: i don't think backporting it to lucid makes sense here? | 14:28 |
| eagles0513875 | ill be back later to continue working on this | 14:29 |
| Rhonda | \o/ - seems like maverick sync is open, bug #109434 sent me a mail :) | 14:44 |
| ubottu | Launchpad bug 109434 in wesnoth-1.8 "Installing a server for a game automatically auto-inits and runs every boot." [Undecided,Fix released] https://launchpad.net/bugs/109434 | 14:44 |
| tarzeau | Rhonda: they already synced on 9.5 if not earlier | 14:49 |
| Rhonda | tarzeau: Well, there was an ubuntu diff in lucid, so I guess that might be the reason for the "delay". | 14:54 |
| tarzeau | Rhonda: aha. any idea what happens with sea-defender and fracplanet now? | 14:54 |
| Rhonda | No clue, I don't have the time for that myself, sorry. | 14:55 |
| === nobawk is now known as nobawk|away | ||
| === nobawk|away is now known as nobawk | ||
| fagan | Does anyone package any vala apps for the repo? | 16:34 |
| fagan | Im looking for guidance about how to do the vala ppa stuff for quickly | 16:35 |
| === nobawk is now known as nobawk|away | ||
| === nhandler_ is now known as nhandler | ||
| MTecknology | !info php5 | 17:08 |
| ubottu | php5 (source: php5): server-side, HTML-embedded scripting language (metapackage). In component main, is optional. Version 5.3.2-1ubuntu4.1 (lucid), package size 1 kB, installed size 20 kB | 17:08 |
| === yofel_ is now known as yofel | ||
| === nobawk|away is now known as nobawk | ||
| Riddell | slangasek: you uploaded bug 519541 but didn't accept it or note your SRU approval? | 17:23 |
| ubottu | Launchpad bug 519541 in abiword "Abiword 2.8.1 freezes with document lost when help is clicked or F1 is pressed" [Critical,In progress] https://launchpad.net/bugs/519541 | 17:23 |
| slangasek | Riddell: I uploaded it in a sponsor capacity, would prefer a second vote of confidence from another SRU team member for the acceptance | 17:23 |
| benste | hi, this channel was fullfilled by those who do some packaging work right ? - I'm trying the latest gnome-power-manager from git master but didn't finish to compile it becuase autogen quotes it needs gnome-keyring0 but there is no such package in ubuntu - what can I do ? | 17:45 |
| benste | someone here ? | 17:47 |
| benste | solved - hidden in libgnome-keyring-dev | 17:51 |
| === kirkland` is now known as kirkland | ||
| === mok0_ is now known as mok0 | ||
| === erhesrhsrtb54vyh is now known as Elbrus | ||
| === nobawk is now known as nobawk|away | ||
| === nobawk|away is now known as nobawk | ||
| fabrice_sp | mannyv, will you take care of libclass-trait-perl merge/sync? | 19:59 |
| micahg | ari-tczew: i'll try to get to libjdic-java early next month | 20:00 |
| mannyv | fabrice_sp, yeah I can look into that | 20:00 |
| fabrice_sp | mannyv, you're the one that last touched it :-) | 20:01 |
| ari-tczew | micahg: I'm glad and amazed of yours memory and solidity :)) | 20:02 |
| micahg | ari-tczew: I noticed all the merges you were doing so I figured it would be on you mind ;) | 20:03 |
| ari-tczew | micahg: yea, I'll try to merge libjdic-java if you'll have done fix ftbfs on debian :> | 20:04 |
| micahg | ari-tczew: I was going to merge in ubuntu first and then push the fixes up to debian | 20:05 |
| ari-tczew | micahg: feel free, your choice :) | 20:06 |
| micahg | ari-tczew: k, thanks | 20:06 |
| === mok0_ is now known as mok0 | ||
| MTecknology | how can I see if I have any packages that were installed that werent' from the repos that are currently running | 21:02 |
| carstenh | ask google about ask-listrepository and grep for dpkg/status in its output | 21:03 |
| MTecknology | carstenh: thanks | 21:04 |
| MTecknology | carstenh: hm?? I'm not finding anything useful | 21:23 |
| carstenh | MTecknology: sorry, a typo ... | 21:27 |
| carstenh | MTecknology: bugs.debian.org/504460 | 21:28 |
| carstenh | MTecknology: download it, run chmod a+x apt-listrepository and then ./apt-listrepository | grep dpkg/status | 21:29 |
| MTecknology | carstenh: thanks | 21:43 |
| ari-tczew | fta: are you planning merge chromium-browser with debian unstable? | 22:05 |
| fta | ari-tczew, no. please read the bug(s) as for why | 22:05 |
| fta | ari-tczew, and remember debian used our package, not the opposite | 22:06 |
| ari-tczew | fta: ok thanks for information | 22:08 |
| === nobawk is now known as nobawk|away | ||
| mirsal | hello | 23:14 |
| === TheMuso` is now known as TheMuso | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!