Time  Nick          Message
14:40 messa4        hello
14:41 messa4        HELP. how to get minetest in ENGLISH? im using windows xp [french version]. but i want to play minetest in english
14:41 messa4        its serious bug
14:42 messa4        imagine: u are using chinese computer in cyber-caffe in china, u want to play but minetest is AUTOMATICLY changing launugage to chinese instead english
14:43 VanessaE      wrong channel, messa4
14:44 VanessaE      these questions belong in #minetest
14:44 messa4        I CANT
14:44 messa4        PLEASE HELP ME
14:44 messa4        I have only 40 minutes lefty of cybercafe
14:44 messa4        and i
14:44 messa4        want to play in english
14:44 ironzorg      messa4: "language" variable in minetest.conf
14:44 messa4        but menu if FORCING french menus becasue of french windows installed on machine
14:44 messa4        why its not in some menu?
14:45 messa4        also why minetest is doing that without question?
14:45 messa4        THERE IS NOT "language" in minetest .conf!!!!!!!!!!!!!!!!!!
14:46 messa4        any professinal here? HOW TO CHANGE LAUNGUAGE of game?
14:46 messa4        why this is not in the menu?
14:46 ironzorg      language=en
14:46 VanessaE      messa4: this is a community, volunteer effort, please show some respect.
14:46 messa4        please add it to menu ok?
14:46 messa4        i was using minetest for year
14:46 messa4        on pc with english windows
14:47 messa4        so i was shocked when i enetered the game [usb pen drive] and suddenly my game is FRENCH
14:47 messa4        this is serious bug
14:48 ironzorg      who is bug
14:48 ironzorg      bug better be serious about this.
14:48 zat           How can it be a bug that a game uses the language of the OS
14:48 messa4        autoamtic launguage change based on windows version.
14:48 messa4        please add change in menu
14:48 VanessaE      this is not a bug, messa4.  the normal behavior of most software that is i18n compatible is to adapt to the language of the system being run on.
14:48 messa4        otherwise - people would be not able tyo play
14:48 messa4        but its opposite
14:49 zat           Windows is in french, game decided to start in french because of that
14:49 messa4        re-read my "travel to china to cybercafe" line
14:49 zat           how can that behavior be incorrect?
14:49 messa4        and what if i will be in china?
14:49 VanessaE      change the language variable, or add it if it isn't in there.
14:49 messa4        THIS SHOULD BE IN MENU IN GAME
14:49 messa4        not in some myhsterious config
14:49 messa4        that i would never find without asking on irc
14:49 messa4        :(
14:49 messa4        think of normla users
14:49 messa4        normal
14:50 zat           normal users use Windows in their native language AFAIK
14:50 VanessaE      minetest.conf is not "mysterious".  Almost eeryone who uses minetest knows of this file.
14:50 messa4        i didnt know about it
14:50 zat           messa4: if it is so urgent for you, you can add the feature to the code.
14:50 messa4        and i played minetest for almost year
14:50 VanessaE      and am I wrong, or doesn't every minetest install also come with a copy of the .example also?
14:50 ironzorg      yep, that's where I found the variable
14:51 messa4        zat: from me its ok. as i can ask for help on irc. but im afriad that normal users would never know how to change it. imo there should option in menu. thats it :)
14:51 messa4        ps. minetest is great. and thx for making it
14:51 zat           sure :)
14:52 messa4        zat: like flag on first menu that u see. u click on falg and choose: default os launguage or english or other transaltions avalible
14:52 zat           yes I know how language selection works
14:52 messa4        good
14:53 messa4        do it then. and add living nimals and mosters :) thanks
14:53 messa4        animals*
14:53 VanessaE      messa4: if you feel the need to add something like this, please code it and submit a pull request
14:53 VanessaE      even an initial language selection screen as you propose would probably suffice.
14:53 messa4        VanessaE: in future maybe. now i dont have access to computer, electricity or internet. [cyber cafe now]
14:53 VanessaE      morning, Jordach
14:54 VanessaE      messa4: if no electricity, how have you been playing for the past year?
14:54 messa4        VanessaE: I moved to diffrent country month ago :) traveling
14:54 VanessaE      I se
14:54 VanessaE      see*
14:54 Jordach       wrong channel
14:54 VanessaE      Jordach: been there, done that.
14:55 VanessaE      now, please take this to #minetest
14:55 VanessaE      you're already signed into that channel also.
14:55 messa4        i cant talk there
14:55 VanessaE      why not?
14:55 messa4        Cannot send to channel
14:56 VanessaE      try now.
15:30 messa4        when u will add the launguage choose option to default build?
15:30 messa4        its crtitical bug
15:31 messa4        u will go to china and unable to play
15:31 messa4        at current state
15:31 zat           If I ever travel to china I really doubt I would be worried of playing Minetest in a computer.
15:31 messa4        what if u will move to china?
15:32 messa4        some people travel and not stay in hotels
15:32 messa4        for months
15:32 messa4        think about it
15:32 zat           I doubt I will ever be, and if I move, I think I am smart enough to change a config.
15:32 zat           $"#&%$&#%&#$
15:38 ironzorg      messa4 | its crtitical bug
15:38 ironzorg      :)
15:40 zat           extreme unstability
16:37 ShadowNinja   zat: I will try your SQLite rollback. Where should I get it from? And which branch is it in?
16:41 zat           ShadowNinja it is the last commit in the master branch at git://bros.ivn.cl/mbarrera/minetest.git
16:46 ShadowNinja   zat: rollback.txt is left in place for you to delete if all went well?
16:47 zat           yes if all went well it will be deleted
16:47 zat           so better back it up just in case
16:47 zat           I made a hard link of mine to not use more space
16:48 ShadowNinja   zat: It shouldn't delete it.
16:48 zat           (/rollback_check is not implemented in the mod, only the rollback function)
16:48 zat           ShadowNinja: it should not, but how does it know it should not migrate the file again?
16:49 zat           I could not find more useful functions in filesystem.h to make it do something else
16:49 zat           I had originally though of appending a .done to the file
16:49 ShadowNinja   zat: It should only migrete if rollback.sqlite doesn't exist, rollback_recording is enabled, and rollback.txt exists.
16:49 ShadowNinja   migrate*
16:49 zat           I let it migrate is the rollback.sqlite does not exist in case of incomplete migrations
16:50 zat           you can kill minetestserver during a migration, and it will continue on next start
16:50 ShadowNinja   zat: If the migration is incomplete you just delete the sqlite file.
16:51 zat           if the migration is incomplete it will more likely be for a reason that did not let the server correctly close
16:51 zat           but well, I have to take your opinion into consideration
16:51 zat           I would rather rename the original file anyway
16:51 Exio4         you need to think the rollback generated could be corrupted
16:52 Exio4         and you would want to go to the "old format" or keep it or whatever
16:52 ShadowNinja   Yes, a pertially-converted db may very well be corrupted.
16:52 zat           I am using transactions to minimize that possibility
16:53 zat           still I need more useful functions in filesystem.h
16:53 zat           how do I guarantee that the resulting database is not corrupted?
16:53 ShadowNinja   zat: Then write them. :-)
16:53 Exio4         you can't? that is why you shouldn't delete it
16:53 ShadowNinja   zat: I don't think you can.
16:53 zat           ShadowNinja: I could write them, but I cannot guarantee it would be cross-platform.
16:54 zat           Exio4: I cannot guarantee it would not be corrupted, but I cannot tell whether should I migrate the existing file again or not
16:54 zat           it would be easier if filesystem.h provided a function to rename the old file afterwards
16:54 Exio4         if the sqlite file is there, you shouldn't migrate
16:54 zat           I let it do that in case of interrupted migrations
16:55 ShadowNinja   zat: Add a file, set a value in the db, or something similar while migratin and remove it afterg. If the key or file exists then restart the migration.
16:55 zat           other thing I could do is recording the modification date of rollback.txt and record somewhere after the migration is done
16:55 zat           well lol like that yes
16:57 zat           whatever method I am implementing it after confirming the new rollback works fine for everybody
16:57 zat           how big is your txt ShadowNinja?
16:57 ShadowNinja   zat: Could the speed be calculated more acurately? It seems to round to 5000.
16:57 zat           actions per second?
16:58 ShadowNinja   $ wc -l worlds/b/rollback.txt
16:58 ShadowNinja   5426636 worlds/b/rollback.txt
16:58 zat           ok that is a bunch
16:58 ShadowNinja   Took a few minutes.
16:58 zat           the speed can be calculated more accurately yes, I think it just does 10000 actions per commit
16:58 ShadowNinja   So rollback_check doesn't work?
16:58 zat           yes I did not implement that one yet
16:59 zat           but it is easy to implement... I abstracted the query mechanism to a single function
17:00 zat           I can probably improve that part of the code and have the migrator adjust its actions per commit depending on its speed.
17:00 ShadowNinja   Alright, that is the best function to profile this since it doesn't involve changing the map.
17:00 zat           did not think of that huh
17:00 zat           but aren’t you working on a test world?
17:00 zat           (not on production/public)
17:01 ShadowNinja   A old copy of my public server.
17:01 zat           ah good
17:01 zat           then I guess you can destroy :D
17:08 ShadowNinja   rollback.sqlite: 464M rollback.txt.bkp: 833M
17:09 zat           it optimized too
17:19 ShadowNinja   Using the sqlite3 tool on the DB shows aproximately 1 second rollback checks.
17:20 zat           did you perform actual rollbacks?
17:20 ShadowNinja   Much better than >256s checks. (My server started pinging out from the IRC server it was connected to)
17:21 ShadowNinja   Nope, I want to keep this world for now, I may want to copy things from it.
17:21 ShadowNinja   I used it because it had a big rollback file.
17:22 zat           cannot you copy the world and change that one?
17:23 ShadowNinja   Yes, but the SQLite rollback os definitely better. The bottleneck should only be the ma now.
17:24 zat           ma?
17:24 ShadowNinja   map. grep shows speeds similar to SQLite.
17:30 ShadowNinja   zat: Make sure that when you change it you make rollback_check more informative, eg: BadPlayer changed air to lava 12345 seconds ago <newline> GoodPlayer changed lava to stone 1234 seconds ago <newline> GoodPlayer changed stone to air 1232 seconds ago.
17:30 celeron55     zat: i didn't bother reading the full discussion, but you're free to add a function to filesys.h (it can just use the C standard rename() like safeWriteToFile does)
17:32 zat           ShadowNinja: I guess that would go in the file that interfaces with lua but sure I can improve it
17:33 zat           celeron55: thx for permission and I noticed what you meant the other day... when trying to amend my own commit I soft-resetd by accident and ended up amending a commit that wasnt mine D:
17:34 celeron55     do it in a separate commit so things stay organized
17:34 zat           right, I never meant to use an existing commit to begin with roflol
17:35 celeron55     has Sokomine already tested sqlite rollback? sokomine is one of the people who've been eager to get some improvements on rollbacks
17:36 zat           Sokomine asked me how to get it but hasn’t tell me how it went.
17:36 zat           told*
17:39 celeron55     having it as a pull request on github would be good for making it seen by more people
17:40 ShadowNinja   ^ The proper way to do this is to fork the repo and make your changes in a sqlite_rollback branch.
17:40 celeron55     (which is important because the more eyeballs, the more issues are found before users get hold of it)
17:41 zat           oh goood
17:41 zat           I am going to do that
17:41 zat           after I finish cleaning the kitchen...
17:42 celeron55     i'm totally in for having as much improvement on the rollback functionality as possible; the current one is really just a proof-of-concept
17:44 celeron55     (i know it is because i made it 8))
17:44 zat           kik
17:48 celeron55     one thing i guess would be very desired would be automatic removal of old rollback logs
17:48 celeron55     so that server admins can decide that they tend to hear about griefings within a week and then save space by not holding a month of log
17:49 celeron55     i haven't really been running a server in a year or two, and never ran a server with rollback enabled so i don't know what i'm talking about
17:49 VanessaE      +10000
17:50 VanessaE      an active server will generare 2-3 GB worth of rollback.txt in a few weeks' time
17:50 ShadowNinja   DELETE FROM action WHERE timestamp<current_time-12345;
17:50 celeron55     yeah, extremely easy with SQL
17:50 celeron55     altough not hard with sequential a plaintext file either, but whatever
17:51 ShadowNinja   It's so easy to do with a external script and the sqlite3 utility that you could just provide a shell script to add to crontab.
17:52 celeron55     it should be just a minetest setting
17:52 zat           and vacuuming every once in a while
17:53 zat           the part I am liking the most now is that sqlite files tend to be half or less in size than the original txt files
17:53 celeron55     it seems to me that it could be optimized many times more
17:53 zat           I wonder, could they get even smaller by adding an index for the node names?
17:56 celeron55     dunno; it's a bit hard to guess what takes most space with that database layout
17:56 celeron55     how tight does sqlite squeeze NULL values?
17:57 celeron55     i wonder if you add a million rows to that table with everything being NULL, how much space is that
17:58 celeron55     hmm, altough, actually that is a useless experiment because most of them are always used
17:59 zat           I have done such tests
17:59 celeron55     i guess the node names don't really add a lot to the 18 other columns
17:59 zat           in postgres though
17:59 zat           hold on I am testing something now
18:01 ShadowNinja   zat: Perhaps there should be two tables, set_node and inentory or some such. That way you don't have rows that are half NULL.
18:01 ShadowNinja   inventory*
18:02 zat           That was the first thing I did, queries got extremely complex because of that.
18:03 zat           also, there was a really ugly limitation that forced me to use a single table
18:04 zat           anyway, it would not be hard to change the layout if ever needed
18:06 celeron55     KISS is quite important
18:07 celeron55     don't make anything more complicated unless the benefits are like 2x
18:10 Sokomine      celeron55, zat: i'll only be able to test it on a very mixed world - month-old rollback.txt with a world that has been built on in for several month. thus, migration can most likely be checked; but actual rollback will too often fail due to changes in the world
18:11 zat           Its something :P
18:12 zat           I did not make the migration procedure fault tolerant so I do not know what to expect on syntax errors
18:12 ShadowNinja   Sokomine: With SQLite rollback there should be no need to periodically delete the rollback log.
18:12 Sokomine      for the actual ingame checking of rollback data, i'd recommend my inspector-mod. i think it's very useful. currently it calls grep via a shell-script on the rollback.txt file
18:13 Sokomine      celeron55: how did you manage to make current rollback *that* slow? :)
18:13 zat           Speaking of periodically deleting stuff... a MongoDB backend for maps would be good since mongodb can use TTL per record.
18:14 zat           (and can also provide high-availability/fault-tolerance)
18:15 zat           OK........... I vacuumed my sqlite file, it changed from 550mb to 556mb
18:17 Sokomine      btw, is my pull request #951 now ok? i'd really like to see rotating windmill wings and other things like that
18:17 ShadowNinja   Hah. Make sure to set auto-vacum on the DB if you plan to add clearing of old entries.
18:23 celeron55     Sokomine: yeah, i guess i'll push it as everybody else is just idling
18:24 ShadowNinja   Sokomine: It looks good, but you declare s three times. You should also have the image scale outward for torchlike nodes. (Should just be x*scale*2, x instead of x*scale, x*scale)
18:24 ShadowNinja   (At least IMO)
18:24 ShadowNinja   celeron55: ^
18:24 celeron55     i mentioned the torch thing before
18:24 Sokomine      celeron55: that's fine :-)
18:24 celeron55     but it's kind of very arguable
18:25 VanessaE      the torch thing is easily correctable with the chosen texture also I'd think
18:26 ShadowNinja   Yes, a centered texture should fix that. But is there any good reason for the textore to be embeded in the node that the torchlike node is attached to?
18:26 Sokomine      scale outwards? do you mean "center moved up" like in plantlike or just scaling factor * 2?
18:26 celeron55     for example it's kind of meaningful to have it wrap around the node at the top and bottom of the node it's attached to at visual_scale=3
18:26 celeron55     i can imagine some decorations that should look like that
18:27 Sokomine      at visual_scale=1, nothing ought to change. and i noticed no changes while playing with the modified client
18:27 Sokomine      it's easiest to remember that such overlarge signlike-/torchlike-nodes will have their center at the location where the node was placed
18:28 VanessaE      being able to display a flat texture at 45 and 90 degree angles but scaled out past the edges of the node could be potentiall7y useful.
18:28 VanessaE      I'd say don't change the centering.
18:28 VanessaE      -7
18:28 celeron55     it might be a good idea to not implement this for torchlike until we have an actual example of something that would be implemented with scaled torchlike
18:29 celeron55     bah, whatever, i'll just push this; it can't really hurt anyway
18:30 Sokomine      right now, torchlike can't be scaled. there's nothing lost if it's implemented. only when people start using it and we change it later on it'd be bad. through usage requirements may change
18:30 Sokomine      perhaps a comment in the lua_api would be best then
18:32 celeron55     i'll do that change
18:33 Sokomine      ok
18:33 Sokomine      thank you
18:36 celeron55     why does github still show your commit altough i push -f'd it out of there immediately after pushing it?
18:36 celeron55     well whatever, don't pull it because i'm going to replace it
18:40 celeron55     now i pushed the changed one, but github doesn't update
18:40 celeron55     well it's their problem 8-)
18:41 celeron55     oh now i see what i did wrong
18:41 celeron55     now it's right
18:45 celeron55     (i had some screwed up remotes in my ancient local repository)
18:50 Sokomine      aah. ok. i had trouble with getting everything to work as well. vanessa and shadowninja helped me "communicate" with my git. next time i'll know that i have to start a new branch first
18:53 ShadowNinja   thexyz: Have you installed that lua template extension on the wiki yet?
18:54 thexyz        ShadowNinja: no, do we need it?
19:01 ShadowNinja   thexyz: It would make some templates much cleaner and would allow for, ge, a Hearts template that allowed 20.5 hearts.
19:01 ShadowNinja   eg*
19:02 ShadowNinja   Currently it only supports 5 because more would mean more than doubling the size of the template since it is only a switch statement.
19:02 thexyz        okay, I'll do that tomorrow
19:58 proller       https://github.com/minetest/minetest/pull/952
20:35 ShadowNinja   zat: I noticed that your SQLite converter has a line rewrite feature for the progres so that it doesn't print 1000 lines. Could you add this to the LevelDB conversion?
20:36 zat           I would have to find where it is I guess
20:43 ShadowNinja   zat: https://github.com/minetest/minetest/commit/58841ef12f6cba1bb622353c1fcaa0e3c6fb46c9#diff-7ec3c68a81efff79b6ca22ac1f1eabbaR1263
20:45 zat           anyway I don’t recall a bunch of lines being printed when I migrated to leveldb
20:45 * ShadowNinja does
20:45 ShadowNinja   zat: I also notice a few trailing tabs and spaces when using git diff --color.
20:46 zat           where?
20:53 ShadowNinja   zat: Multiple places, and my diff is probably pretty old. The code style for the for statement also seems a little off (See http://pastebin.ubuntu.com/6237794/ )
20:54 zat           ??
20:54 zat           you are just complaining that you dont like the style of a for?
20:55 ShadowNinja   "using namespace std;" is discouraged, but if you are going to bring std::string into the nurrent namespace you shoud reper to it as simply "string".
20:56 ShadowNinja   zat: Yes, see the code style guidelines on the dev wiki.
20:56 zat           discouraged by who?
20:57 ShadowNinja   zat: Me, hmmm, celeron?
20:58 zat           I see, the kind of devs that like enforcing their opinion on coding
20:58 zat           I say that for forcing people to code in a certain way regarless of the readability
20:59 kahrl         if you are ready to fix it when there is a name collision in the global namespace on $RANDOM_SYSTEM
20:59 kahrl         then go ahead
21:00 ShadowNinja   zat: In order to maintain readability you have to have consistent code style.
21:01 zat           since when does consistency means readability
21:01 zat           that is just an opinion
23:18 Exio4         sure, having a file where you have mixed tab and spaces and complete different code styles doesn't make the code ugly and less mantainable
23:18 Exio4         and your editor probably likes it too
23:19 proller       astyle can solve most style problems