| hannie | teolemon, if you are in, I just want to say I sent you an email a minute ago with my experiment results | 14:35 |
|---|---|---|
| teolemon | looking at it | 14:35 |
| hannie | ok | 14:35 |
| teolemon | you have to mark them as translated | 14:37 |
| teolemon | not fuzzy | 14:37 |
| teolemon | and import them as translated in the bogus project | 14:37 |
| hannie | you mean I have to approve them in Lokalize first? | 14:38 |
| hannie | In that case it is not different from how I used to do it | 14:38 |
| teolemon | and what you can upload first are machine translations | 14:38 |
| hannie | what you call machine translation is already in the merged file | 14:38 |
| teolemon | not quantal | 14:38 |
| teolemon | raring+ machine translations | 14:38 |
| hannie | in my case precise-e2 (100% translated) | 14:39 |
| teolemon | and then the po file with the fuzzy | 14:39 |
| teolemon | marked as translated | 14:39 |
| hannie | so precise-e2 + raring > merged | 14:39 |
| teolemon | can you send me the po file | 14:39 |
| hannie | ok it is about 928 kB | 14:40 |
| teolemon | machine translation for me is google translate | 14:40 |
| hannie | Is that not too big? | 14:40 |
| teolemon | and did you mark them as translated | 14:40 |
| hannie | teolemon, if I approve them in Lokalize there is no point in sending the file to manualautomated | 14:41 |
| teolemon | before uploading | 14:41 |
| hannie | Because then I have already done the work in Lokalize | 14:41 |
| teolemon | so that they show up as suggestions ? | 14:41 |
| teolemon | i should be ok | 14:41 |
| hannie | If I open the merged file in Lokalize there are more than 500 suggestions called Not ready | 14:42 |
| teolemon | give it a try | 14:42 |
| hannie | teolemon, I already have uploaded the merged file with suggestions to LP but the suggestions that show in Lokalize do not show in manualautomated. | 14:43 |
| teolemon | you mass approve them in a text editor or using keyboard shortcut | 14:43 |
| hannie | If I approve them in Lokalize there is no point in uploading the file to ubuntuautomated | 14:44 |
| teolemon | and then several translators can review them | 14:44 |
| hannie | I can just as well upload it to /ubuntu-manual/raring | 14:44 |
| teolemon | because you didn't mass validate offline first | 14:44 |
| hannie | I have to validate one by one. Is there another way? | 14:44 |
| teolemon | you feed the bogus project with strings you mass validated | 14:45 |
| teolemon | you don't care about the content | 14:45 |
| hannie | And where do I mass validate the fuzzies? | 14:45 |
| teolemon | select all | 14:45 |
| teolemon | validate | 14:45 |
| teolemon | save | 14:45 |
| teolemon | and upload | 14:46 |
| hannie | Hold on. Select all where? In a text editor? A text editor does not have a function Validate | 14:46 |
| teolemon | and then | 14:46 |
| teolemon | 10 people can look at them online | 14:46 |
| teolemon | and see if modifications are necessary | 14:47 |
| hannie | teolemon, I think I have to see if Lokalize has this function "validate the whole selection" | 14:47 |
| teolemon | yes | 14:48 |
| hannie | Now I think I see what you mean. Before uploading I have to approve the whole selection. I have to check if I can do this in Lokalize. I will let you know. | 14:48 |
| teolemon | using a text editor | 14:48 |
| hannie | you mean a CAT | 14:48 |
| teolemon | or using a keyboard shortcut | 14:48 |
| teolemon | as you wish | 14:49 |
| teolemon | #, fuzzy | 14:49 |
| teolemon | msgid "label 1" //Nouveau msgid | 14:50 |
| teolemon | you find and replace the fuzzy part | 14:50 |
| teolemon | to make it look like a regular string | 14:51 |
| hannie | just a minute, I'll open it in gedit | 14:51 |
| teolemon | i did it in one or two minutes using shortcut and typing like a crazy person on my keyboard | 14:53 |
| teolemon | but a find and replace in kedit | 14:53 |
| teolemon | or any text editor will be even faster | 14:54 |
| hannie | right, I see #, fuzzy in gedit | 14:54 |
| hannie | #, fuzzy must be changed to what? | 14:54 |
| teolemon | https://gist.github.com/anonymous/c4e32486f76e895ee2d4 | 14:54 |
| teolemon | here's an exemple | 14:54 |
| hannie | Yes, I spotted the #, fuzzy lines. Do I have to delete them or replace them by someting else? | 14:56 |
| teolemon | find all | 14:56 |
| teolemon | #, fuzzy | 14:56 |
| teolemon | replace by | 14:56 |
| teolemon | (and put nothing) | 14:56 |
| hannie | ok, so just delete the whole line #, fuzzy | 14:56 |
| teolemon | and you're done | 14:56 |
| teolemon | is that ok ? | 14:56 |
| hannie | I think I get it now. But you still have to find and delete them one by one. There is no bulk deleting in gedit | 14:57 |
| teolemon | wipe them | 14:57 |
| hannie | Or is there a command I do not know? | 14:58 |
| teolemon | to make them look like translated strings | 14:58 |
| teolemon | yes | 14:58 |
| teolemon | you can check the result in lokalize | 15:00 |
| teolemon | if you want | 15:00 |
| teolemon | eeeh | 15:00 |
| teolemon | replace all ? | 15:00 |
| CarstenG | Hannie, try CTRL+H | 15:00 |
| teolemon | gedit doesn't have a find and replace function | 15:00 |
| hannie | right. It takes a while to delete 500 #, fuzzy lines | 15:00 |
| CarstenG | gedit has a replace function. | 15:01 |
| teolemon | is that what you're implying ? | 15:01 |
| teolemon | ;-) | 15:01 |
| hannie | CarstenG, that's the right shortcut (didn't know it, thanks) | 15:01 |
| CarstenG | Open the window with CTRL+H | 15:01 |
| teolemon | http://projects.gnome.org/gedit/ | 15:01 |
| teolemon | hum hum | 15:02 |
| CarstenG | and there you fill the fields and click then »Replace all« | 15:02 |
| hannie | Ayay, my bad. Didn't know it existed in gedit | 15:02 |
| teolemon | otherwise, if you're into KDE, you can use a kde text editor | 15:02 |
| hannie | It is done in a jiffy | 15:02 |
| teolemon | with more functionnality :-P | 15:02 |
| hannie | Now I will upload it again to ubuntuautomated | 15:03 |
| teolemon | ten seconds ? | 15:03 |
| CarstenG | teo: We are making the manual for GNOME, so we use gedit :-p | 15:03 |
| * teolemon won't joke about German efficiency :-) | 15:03 | |
| hannie | or less ;) | 15:03 |
| hannie | I use gedit too | 15:04 |
| CarstenG | :-) | 15:04 |
| hannie | I only saw it as a toy compared to big brother Word | 15:04 |
| teolemon | ^^ | 15:04 |
| hannie | thanks for your patience with me | 15:05 |
| teolemon | http://manilovefilms.com/wp-content/uploads/2012/09/noisy-cricket.jpg | 15:07 |
| hannie | One thing I have to say to my defence: there is no Find and Replace in the menu bar of gedit | 15:07 |
| teolemon | small but powerful :-) | 15:07 |
| CarstenG | Teo: Can you explain this topic »bogus project«, »ubuntuautomated« on the mailing list? It sound interesting, but I did not yet understand it in detail. Also for other translator it would be useful, or? | 15:07 |
| teolemon | which ml ? | 15:08 |
| teolemon | ubuntu-translators ? | 15:08 |
| CarstenG | no our ml | 15:08 |
| CarstenG | ubuntu-manual | 15:08 |
| hannie | or both | 15:08 |
| teolemon | it's basically a rehash of what we did for the ddtp | 15:10 |
| CarstenG | Hannie: There is a entry in the menu bar \menu{Search \then Replace \ldots} :-) | 15:10 |
| teolemon | lite.framapad.org/p/ddtpUbuntu | 15:10 |
| teolemon | http://lite.framapad.org/p/ddtpUbuntu | 15:10 |
| teolemon | we detailed our methodology here | 15:10 |
| hannie | CarstenG, I found it under Search, I expected it to be under Edit. O, O, O | 15:11 |
| teolemon | you know things are bad when you start talking LaTeX to fellow team members | 15:11 |
| teolemon | hannie, have you uploaded it ? | 15:12 |
| hannie | I am going to do it in a second | 15:12 |
| CarstenG | teo: Thanks for this link. I'll have a look | 15:12 |
| hannie | teolemon, done. Thank you for your upload. In import queue | 15:14 |
| teolemon | i've added a paragraph | 15:14 |
| teolemon | on how to unfuzzy strings | 15:15 |
| teolemon | as we mainly used google translate for ddtp suggestions | 15:16 |
| hannie | teolemon, are you the one who approves the import? | 15:17 |
| teolemon | we didn't rely on fuzzy | 15:17 |
| teolemon | just approve | 15:17 |
| teolemon | approved it | 15:17 |
| teolemon | it should appear in a couple of minutes | 15:17 |
| hannie | ah, great! Thanks | 15:17 |
| hannie | I am so curious if the result is ok now | 15:18 |
| teolemon | but you'll have 100 or so strings left to translate fully by youself | 15:18 |
| hannie | better than 500 ;) | 15:19 |
| teolemon | possibly more | 15:19 |
| hannie | I may even decide to translate raring. so far we only do LTS versions of the manual | 15:19 |
| teolemon | i think so | 15:19 |
| teolemon | or maybe you can approve it yourself | 15:19 |
| hannie | 100 strings of the manual is different from gui strings (much shorter) | 15:20 |
| teolemon | as ubuntu-translators member | 15:21 |
| hannie | Not many of my fellow translators like the long strings | 15:21 |
| teolemon | I can guarantee that when you do that at DDTP scale | 15:21 |
| CarstenG | Teo: It looks like that the project https://launchpad.net/ubuntumanualautomated is for the current raring translation. Is it possible to use it also for the quantal version? | 15:21 |
| teolemon | you get a great chill | 15:21 |
| teolemon | (teaser ^^) | 15:21 |
| hannie | lol | 15:21 |
| teolemon | and we should be good ! | 15:21 |
| teolemon | have a look | 15:21 |
| hannie | We may start working on DDTP again if I can do the fuzzy trick there too | 15:22 |
| teolemon | you have 207 strings left | 15:22 |
| CarstenG | Or should we create a https://launchpad.net/ubuntumanualautomated-quantal | 15:22 |
| teolemon | do you want me to add automated suggestions ? | 15:22 |
| teolemon | we'd need to upload the quantal template | 15:22 |
| teolemon | but yes | 15:22 |
| CarstenG | ? | 15:22 |
| teolemon | we've finished quantal so i didn't do it | 15:22 |
| teolemon | (French love to brag) | 15:22 |
| teolemon | on the ddtp | 15:23 |
| hannie | We have a shortage of translators at Ubuntu Dutch Translators | 15:23 |
| teolemon | fuzzy is not enough | 15:23 |
| hannie | teolemon, I know I have to study the instructions you wrote. Haven't found time to do it yet | 15:24 |
| teolemon | you need automation | 15:24 |
| teolemon | not useful | 15:24 |
| hannie | Yes, automation is the key word | 15:24 |
| teolemon | and polluting launchpad | 15:24 |
| hannie | We are very grateful what you did for ddtp. I have started to work with Nightmonkey again | 15:24 |
| hannie | And a good idea to find people with interest in a specific field | 15:25 |
| teolemon | use Ubuntu Robot Dutch Translators | 15:25 |
| hannie | :) | 15:25 |
| teolemon | they don't complain | 15:25 |
| teolemon | do the work in 5 minutes | 15:26 |
| hannie | There is always a human behind the robot ;) | 15:26 |
| teolemon | don't eat | 15:26 |
| teolemon | don't breathe | 15:26 |
| teolemon | my wish is that we all automated | 15:26 |
| hannie | yeah, definitely | 15:26 |
| hannie | You have made a good start. Keep up the good work! | 15:27 |
| teolemon | the ddtp have been here for too long | 15:27 |
| CarstenG | Teo: Nice, if you could upload the template for quantal. But I first have to generate the merged po file based on the precise translation. | 15:27 |
| teolemon | i want to get done with it | 15:27 |
| hannie | I am going to have a look at ubuntuautomated | 15:27 |
| teolemon | once and for all | 15:27 |
| teolemon | it should be crowdsourced | 15:27 |
| hannie | Kissy, kissy, only 207 untranslated........ | 15:28 |
| teolemon | and then the translators act as reviewers | 15:28 |
| hannie | teolemon, there is a small risk that I approved strings that shouldn't have been approved | 15:29 |
| teolemon | robots >reviewed by> humans non translators >reviewed by> ubuntu professional reviewers | 15:29 |
| teolemon | it's a pyramid | 15:30 |
| teolemon | the thing is I can't push other teams | 15:31 |
| hannie | teolemon, I will try to revive the ddtp translation in the Dutch team | 15:31 |
| teolemon | i've requested a pot for quantal | 15:31 |
| CarstenG | Hannie: I have a problem with the msgmerge... | 15:32 |
| hannie | CarstenG, what's the problem? | 15:32 |
| teolemon | and google translate could give you suggestions for the rest | 15:32 |
| CarstenG | I do a | 15:32 |
| CarstenG | $ msgmerge precise.po quantal.po > quantal_merged.po | 15:32 |
| teolemon | we're down to 50 | 15:32 |
| hannie | CarstenG, are you in the directory with the po files? | 15:32 |
| teolemon | where ? | 15:32 |
| teolemon | in the automated project ? | 15:33 |
| CarstenG | and now I get in the merged file a fuzzy string wich was already translated in the quantal.po. | 15:33 |
| hannie | CarstenG, only one? Perhaps there was a slight change | 15:33 |
| CarstenG | So is there a way to ignore already translated strings in the quantal file? | 15:33 |
| CarstenG | No, there are more... | 15:34 |
| teolemon | https://translations.launchpad.net/ubuntu-manual/quantal/+pots/ubuntu-manual/nl/+translate?show=untranslated&direction=backwards&memo=430&start=420 | 15:34 |
| teolemon | have a look | 15:34 |
| teolemon | the suggestions are showing up in quantal as well | 15:34 |
| hannie | CarstenG, why would you ignore already translated strings? they should be in the merged file | 15:34 |
| teolemon | CarstenG | 15:34 |
| CarstenG | Ok, let me give an example for better understanding. | 15:35 |
| teolemon | i can put online automatic suggestions for multiverse if you're interested to test | 15:36 |
| hannie | by all means. I am willing to test things for you | 15:37 |
| hannie | or for us the Dutch team | 15:37 |
| teolemon | if you're satisfied with the quality we can then put it for main and universe | 15:37 |
| hannie | I do not like google translate very much, although it is getting better lately | 15:38 |
| hannie | I only use it for common words. Do you know trans.eu? | 15:38 |
| teolemon | i've uploaded the template for quantal | 15:38 |
| hannie | sorry opentran.eu | 15:38 |
| CarstenG | Lets use the glossary entry »Canonical« as example. | 15:38 |
| CarstenG | In precise I have: | 15:38 |
| teolemon | if you want we can do a google hangout so that I can share my screen and drive you through all this | 15:38 |
| CarstenG | \\newglossaryentry{Canonical}{name={Canonical}, description={Canonical, the financial backer of Ubuntu, provides support for the core Ubuntu system. It | 15:38 |
| CarstenG | has over 310 paid staff members worldwide who ensure that the foundation of the operating system is stable, as well as checking all the work submitted | 15:38 |
| CarstenG | by volunteer contributors. To learn more about Canonical, go to \\url{http://www.canonical.com}.}} | 15:38 |
| CarstenG | \\newglossaryentry{Canonical}{name={Canonical}, description={Canonical, der Geldgeber Ubuntus, leistet Unterstützung für das Ubuntu-Kernsystem. | 15:38 |
| CarstenG | Canonical verfügt über mehr als 310 bezahlte Mitarbeiter weltweit, die sicherstellen, dass die Basis des Betriebssystems stabil ist, und die die | 15:38 |
| CarstenG | Arbeit von Freiwilligen überprüfen. Mehr über Canonical erfahren Sie unter \\url{http://www.canonical.com}.}} | 15:38 |
| CarstenG | In qnatal this string is already translated: | 15:40 |
| CarstenG | \\newglossaryentry{Canonical}{name={Canonical}, description={Canonical, the financial backer of Ubuntu, provides support for the core Ubuntu system. It | 15:40 |
| CarstenG | has over 500 staff members worldwide who ensure that the foundation of the operating system is stable, as well as checking all the work submitted by | 15:40 |
| CarstenG | volunteer contributors. To learn more about Canonical, go to \\url{http://www.canonical.com}.}} | 15:40 |
| CarstenG | \\newglossaryentry{Canonical}{name={Canonical}, description={Canonical ist der Finanzsponsor von Ubuntu und betreut das Herzstück des Ubuntu-Systems. | 15:40 |
| CarstenG | Canonical beschäftigt weltweit über 500 Betriebsangehörige, die sicherstellen, dass die Basis des Betriebssystems stabil läuft, und die auch | 15:40 |
| CarstenG | sämtliche Beiträge gegenprüfen, die von freiwillig Mitwirkenden eingereicht werden. Um mehr über Canonical zu erfahren, folgen Sie \\url{http://www.canonical.com}.}} | 15:40 |
| hannie | CarstenG, so you are talking about the change from 310 to 500 staff members? | 15:41 |
| CarstenG | In qnatal this string is already translated: | 15:42 |
| CarstenG | \\newglossaryentry{Canonical}{name={Canonical}, description={Canonical, the financial backer of Ubuntu, provides support for the core Ubuntu system. It | 15:42 |
| CarstenG | has over 500 staff members worldwide who ensure that the foundation of the operating system is stable, as well as checking all the work submitted by | 15:42 |
| CarstenG | volunteer contributors. To learn more about Canonical, go to \\url{http://www.canonical.com}.}} | 15:42 |
| CarstenG | \\newglossaryentry{Canonical}{name={Canonical}, description={Canonical, der Geldgeber Ubuntus, leistet Unterstützung für das Ubuntu-Kernsystem. | 15:42 |
| CarstenG | Canonical verfügt über mehr als 310 bezahlte Mitarbeiter weltweit, die sicherstellen, dass die Basis des Betriebssystems stabil ist, und die die | 15:42 |
| CarstenG | Arbeit von Freiwilligen überprüfen. Mehr über Canonical erfahren Sie unter \\url{http://www.canonical.com}.}} | 15:42 |
| hannie | Yes, and the translation will also be in raring when you merge. What's wrong with that? | 15:43 |
| CarstenG | Hannie: Its no the change from 300 to 500 staff members... The problem is, that I get in the merged file a fuzzy string for an entry which is already translated in quantal. | 15:44 |
| hannie | CarstenG, it will not be a fuzzy anymore when you remove all the #, fuzzy lines from the merged file | 15:45 |
| hannie | As I have only just learned from teolemon | 15:46 |
| hannie | You DO want to keep the translation of a long message, even if there has been a slight change | 15:46 |
| CarstenG | In short words: | 15:49 |
| CarstenG | My "problem": | 15:49 |
| CarstenG | precise translated -- quantal translated --> in merged file string from precise, marked as fuzzy. | 15:49 |
| CarstenG | My expected behaviour: | 15:49 |
| CarstenG | * precise translated -- quantal translated --> in merged file string from quantal, not marked as fuzzy. | 15:49 |
| CarstenG | * precise translated -- quantal NOT translated --> in merged file string from precise, marked as fuzzy | 15:49 |
| hannie | teolemon, I suggest I start reading / experiment with your google translate instructions. After that I'll come back to you and see how we are going to do it for the Dutch language | 15:49 |
| teolemon | ok | 15:49 |
| teolemon | or if you're feeling like it | 15:50 |
| teolemon | i can screencast by doing multiverse | 15:50 |
| teolemon | while you're doing main | 15:51 |
| teolemon | and ask me any questions during the screencast | 15:51 |
| hannie | teolemon, forgive me, but what do you mean by screencast? | 15:51 |
| teolemon | and CarstenG can join too if he's interested for de | 15:52 |
| teolemon | we can have up to 8 persons attenting | 15:52 |
| hannie | do you mean the hangout? | 15:52 |
| teolemon | google hangout | 15:52 |
| hannie | ok | 15:52 |
| teolemon | where i share my screen | 15:52 |
| teolemon | instead of my face | 15:52 |
| teolemon | and meanwhile, we talk | 15:53 |
| hannie | yes, good idea. But I have to prepare before I join | 15:53 |
| teolemon | do you have a google account ? | 15:53 |
| hannie | teolemon, yes, but so far, for privacy reasons, I refused to participate in hangouts ;) | 15:54 |
| teolemon | ok | 15:54 |
| teolemon | but you don't need to prepare much | 15:54 |
| hannie | you may see if you can plan a date and time, I will do some reading in the meantime | 15:55 |
| teolemon | as we'll do the same thing at the same time | 15:55 |
| teolemon | it's a fuc***g efficient way to work | 15:55 |
| teolemon | ok | 15:55 |
| hannie | I suppose so. I'll participate | 15:55 |
| teolemon | if you so wish | 15:55 |
| teolemon | i had a couple of spare hours before our release lunch | 15:56 |
| hannie | CarstenG, if precise translated -- quantal translated --> merged there must have been a slight difference between the two messages | 15:57 |
| teolemon | so i intended to put them to good use :-) | 15:57 |
| hannie | otherwise you do NOT get a fuzzy | 15:57 |
| hannie | teolemon, se if you can find a convenient time and date for yourself and call for participants using doodle | 15:58 |
| hannie | *see | 15:58 |
| hannie | teolemon, I'm sure Redmar will also be interested | 15:59 |
| CarstenG | Hannie: yes, you are right. In this case is a difference between the precise and quantal string. But I'm wondering why I get the precise string as suggestion, if the quantal string is already translated. Or I have not yet understand the idea behind msgmerge... | 16:03 |
| hannie | The strings of the first file are merged with those of the second, resulting in fuzzies in the third file if there is a difference between the two | 16:07 |
| hannie | CarstenG, that is why it is a bit dangerous to change too many things in a new manual version | 16:08 |
| hannie | Example \newglossaryentry was removed in raring, resulting in more than 20 fuzzies (all glossary entries) | 16:09 |
| hannie | The same goes for splitting long paragraphs. It should be to the benefit of translators, but changing them means a lot more fuzzies | 16:10 |
| hannie | ok, guys, time for a pause. See you | 16:13 |
| CarstenG | ok, see you Hannie. | 16:17 |
| CarstenG | Teo: I see now two templates for the German language on https://translations.launchpad.net/ubuntumanualautomated | 16:17 |
| CarstenG | Thanks. | 16:17 |
| CarstenG | So I can now upload the merged po files there? quantal.po and raring.po? | 16:18 |
| teolemon | i goofed a bit i thin | 16:41 |
| teolemon | in both | 16:41 |
| teolemon | it doesn't matter | 16:41 |
| teolemon | launchpad will accept anything | 16:42 |
| teolemon | the suggestions will show anyway | 16:42 |
| CarstenG | what went wrong? | 16:46 |
| teolemon | i think i overwrote the raring template with the quantal template | 16:47 |
| teolemon | but the translations are still showing up for French and Dutch | 16:47 |
| teolemon | i think lp doesn't care much | 16:47 |
| teolemon | it swallows everything | 16:48 |
| teolemon | and then uses them as suggestions | 16:48 |
| CarstenG | Well, I see both, raring and quantal in German language. | 16:48 |
| teolemon | cool | 16:48 |
| teolemon | then yes | 16:48 |
| CarstenG | should the file I will upload, have a specific name? | 16:49 |
| teolemon | upload them in their respective templates | 16:49 |
| teolemon | no | 16:49 |
| CarstenG | ok | 16:49 |
| teolemon | give it the name you want | 16:50 |
| CarstenG | ok, ubuntu-manual-quantal-de.po is just uploaded. | 16:51 |
| teolemon | it's live normally | 16:58 |
| teolemon | i approved it | 16:58 |
| CarstenG | grrrr. | 17:03 |
| CarstenG | I forgot to remove the fuzzy string | 17:03 |
| CarstenG | Ok, then I have to upload it again. :-) | 17:03 |
| teolemon | ok | 17:10 |
| teolemon | it's good | 17:14 |
| CarstenG | ok, fine. | 17:33 |
| CarstenG | How long does it take, that the strings appear in the ubuntu-manual as suggestions? | 17:34 |
| teolemon | a couple of minutes to a couple of hours | 17:35 |
| teolemon | it depends | 17:35 |
| CarstenG | ok | 17:39 |
| teolemon_ | is it working ok now ? | 18:00 |
| CarstenG | no, not yet. | 18:16 |
| teolemon | can you try uploading it into raring as well ? | 19:19 |
| CarstenG | I would, but I did not yet received the po file from LP. | 19:28 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
