Time  Nick         Message
06:05 sfan5        ShadowNinja: we should use z-slice for sqlite3 and individual for leveldb
06:07 ShadowNinja  sfan5: I switched to the individual because z-slice was giving me errors.  I'll try to convert it back later.
06:08 sfan5        ok
06:34 celeron55    http://sprunge.us/NRgQ
06:34 celeron55    so have you tested that this results in correct negative indices?
06:35 celeron55    modifying that is like poking a nuclear bomb
06:37 ShadowNinja  celeron55: Yes, it had issues before testing, but works now.
06:37 celeron55    on every platform and compiler?
06:38 ShadowNinja  And I've had my share of headaches from mixing negative signed numbers with bit twiddling.
06:38 celeron55    that index thing was done by someone who hasn't been around for any time before it nor any time after it, so you have nobody to blame
06:39 ShadowNinja  celeron55: Um, that's pretty straightforward code.  If it fails on a platform we probably don't support it.
06:39 ShadowNinja  I added plenty of () too.
06:40 celeron55    the good thing about using negative indices is that it naturally extends to different coordinate ranges
06:40 celeron55    like if minetest switches to v3s32 some day, it will work just the same
06:40 celeron55    if it used only positive indices, chances are it wouldn't be so easy
06:40 celeron55    (depending on what offsets were used)
06:41 celeron55    hmm actually no
06:41 celeron55    it explodes even like it is now because there are the multiplications nevertheless
06:41 celeron55    oh whatever
06:42 celeron55    the index thing should be moved to be part of the backend instead of a widely used thing; the only reason it exists is because all existing sqlite maps use it
06:43 celeron55    i hope no new backend is using them
06:43 celeron55    because it would be insane
06:43 ShadowNinja  Yes, seperate x, y, and z fields would really be more appropriate for SQLite3.
06:44 celeron55    they are more appropriate for anything
06:44 celeron55    i don't think there exists a usable database where your only option is to index solely by a single integer
06:45 ShadowNinja  I don't understand the s64 thing though.  I thought that it should be a u64, but my DB apparently has negative indices.
06:45 celeron55    maybe you should look at the code again
06:45 ShadowNinja  celeron55: LevelDB uses a single string.
06:45 celeron55    the string can be "x,y,z"
06:46 ShadowNinja  You could seperate...^
06:47 celeron55    ShadowNinja: do you realize you can de-wrap and multiply multiple signed values to a single value like you can do for unsigned ones?
06:47 sfan5        e.g. leveldb does not support seperate keys
06:47 sfan5        aka (x, y, z) -> value
06:47 celeron55    you just go to both sides of zero then
06:47 sfan5        doing that in redis would be possible but very .. let's say.. hacky
06:48 celeron55    you don't need separate keys; you just need a sane "hash" of the coordinates that isn't so obscure like the integer used by the sqlite backend
06:49 celeron55    a string where they are appended is such a thing
06:49 ShadowNinja  celeron55: What do you mean?  Positions are signed, but it seems like the hash shouldn't be.
06:50 celeron55    ShadowNinja: shouldn't? maybe in a sane world, but nothing prevents you from doing it, and it behaves completely logically
06:50 celeron55    ShadowNinja: what is this "shouldn't"? who defines it? it's not undefined behavior at least
06:50 ShadowNinja  You need a seperator, otherwise 1,15,5 == 11,5,5
06:50 celeron55    it's simply not separated by bits but by multiplications
06:51 ShadowNinja  celeron55: Why are there negative keys in my DB?  Do you know?
06:51 celeron55    it just raises the non-x-values to a higher range
06:51 celeron55    because of this
06:51 celeron55    how can you not understand?
06:51 celeron55    this is incredibly simple
06:53 celeron55    it's exactly how it works; it doesn't use bits; it multiplies the Z value by the ranges of the X and Y values, so that when the Y value multiplied by the range of the X value and the X value is added to it, the effect each coordinate has to the final value doesn't wrap onto others
06:53 celeron55    it's not separated to their own bit fields; they are separated by multiplication
06:53 celeron55    i.e. negative values have very clear meaning
06:55 celeron55    you just made it harder to understand by changing it to bit shifts
06:55 celeron55    because it now uses your way of thinking, which is wrong
06:55 celeron55    or at least you don't understand it by using that way
06:58 ShadowNinja  Well, I don't understand what you mean.  ;-)
06:58 celeron55    the reverse is rather unintuitive though
06:59 celeron55    maybe you shouldn't change code that you don't uunderstand
07:02 celeron55    can i ask you to focus on fixing bugs and implementing features instead of changing code around for no good reason?
07:19 sfan5        celeron55: did I tell you that I got this working yet? http://sprunge.us/COMR?cpp
07:23 celeron55    try to integrate it to the pull request then
07:24 sfan5        I'll do that when I have time
07:24 sfan5        the (more or less) compiler I wrote for that also outputs sane error messages now
07:24 celeron55    i think it would be fine to just include the actual serialization code inside the packet definition
07:24 celeron55    because some of them are going to get different formats for different versions and so
07:25 celeron55    and they are bound to have special types with special formats
07:25 sfan5        I'll see if I can add that without making it too complicated.
07:25 sfan5        error message example:
07:25 sfan5        packets.txt:39:28: Expected identifier got '{'
07:25 sfan5        |  list u16 SomeListContainer {
07:25 sfan5        |                             ^
07:26 celeron55    you can do it based on indentation
07:26 celeron55    then it's trivial to see where it ends
07:27 sfan5        I'm for using { ... };
07:27 sfan5        (because that is what I currently use and wouldn't require changes to the lexer)
07:27 celeron55    oh well, i guess something like C++ {} would look reasonable
07:28 celeron55    you just need to match the curly brackets
07:30 sfan5        it ends up looking like this in the code: http://sprunge.us/UBNc
07:30 celeron55    actually more like C++::serialize { ... } and C++::deserialize { ... } in the long run (currently deserialization isn't done by that)
07:31 celeron55    (but deserialization is lower priority anyway; it's more important that the data correct to begin with)
07:31 celeron55    +is
11:02 seolfor      what IDE are you guys using? im talking to linux users ...
11:16 celeron55    vim and bash
11:18 seolfor      celeron55, and now tell me it's a joke :p you also debuging under vim ?
11:18 seolfor      how about call stack do you have it also in vim ? ;D
11:18 celeron55    gdb is pretty good
11:19 celeron55    also valgrind and various smaller command-line utilitiees
11:20 seolfor      sfan5, it's freeking unfair i spend 4 days on attempts to compile minetest under windows and in linux i done it like in 4 shell commands ..
11:21 seolfor      celeron55, yep i'd been using valgrind from time to time but with gdb im not to familliar ..
11:22 seolfor      any way since i cappable to bild minetst already, what can i do ?
11:23 seolfor      i mean is ther some small job or jobs to be done for good warm up :D
11:25 celeron55    suitable small jobs are always very hard to find in the engine
11:27 seolfor      celeron55, all right it shouldn't be much problem.. i will just play a bit with a code ..
11:30 celeron55    you could continue this, altough the work will be replaced by an automatic generator if sfan5's attempt turns out to be working: https://github.com/minetest/minetest/pull/1228
11:31 celeron55    (integrating the generator to that would be one thing, but sfan5 hasn't released it yet)
11:32 celeron55    but yeah, there just isn't anything really; you're free to find something you'd like to do and ask here if it's okay
11:33 celeron55    alternatively you could try to fix some bug listed here, but most are too complicated: https://github.com/minetest/minetest/issues?direction=desc&sort=created&state=open
11:35 celeron55    if i may, i'd propose that you start off from developing some game or mod, and then come to work on the engine when it's missing something that is needed
11:35 celeron55    it would be the healthiest way to find things to do in it
11:36 seolfor      celeron55, thx for guidance and one more thing it will be faster if i simply ask are you have some specific codding standart?
11:37 seolfor      for minetest ofcourse
11:38 celeron55    you will find a lot of stuff here: http://dev.minetest.net/Main_Page
11:38 celeron55    http://dev.minetest.net/Code_style_guidelines
11:38 celeron55    some of that is rather arguable though; like almost everything in this project...
11:42 celeron55    seolfor: this gives some idea on who knows and does what: http://dev.minetest.net/Organisation
11:44 BlockMen     who is [OPEN]? :P
11:45 seolfor      BlockMen, i think its open position to take :p
11:47 PenguinDad   [OPEN] seems to be very lazy (s)he never commited something ;)
11:53 seolfor      I experiencing some strange mose behavior in newst commit - its have some very curved speed sensitivity
11:53 seolfor      it's only my problem or it's common thing ?
12:07 seolfor      and i have the same problem in official arch release 0.4.9-3, i curies was it already reported as issue ? ;]
12:22 celeron55    BlockMen: it's thexyz who left his position
12:23 seolfor      celeron55, it's seems i already found somthing small ;] mose sensivity in arch for me is much to high and even minimum is only "playable" not even close to confortable
12:23 * BlockMen   thinks "at least PenguinDad got it"
12:26 seolfor      celeron55, 1 more thing you will din't answerd my prvies question about codind stanadrt, and are you guys on purpose making so HUGE source files ?
12:27 seolfor      it's not even splited by class per file somthig like that is not acceptable in my job for stable release
12:28 seolfor      but if it's yours policy here.. i can't oposide it
12:28 celeron55    they are split by purpose; i think the java ways of putting 100 lines in a file is not useful
12:29 seolfor      ye but reading file close to 4000 lines it even worse
12:30 celeron55    well those are too large; yes; there are only a couple of them though
12:30 seolfor      somthing more then 1000 line's its very hard to read for me
12:30 celeron55    they're badly designed otherwise too as the 4000 lines consists of a single class
12:30 seolfor      not realy
12:30 seolfor      for ex. im right now looking at game.cpp
12:31 celeron55    that's the only one that doesn't
12:31 celeron55    ...maybe
12:32 seolfor      ;] i thinl i will be pain in the ass for you guys ;D
12:32 seolfor      think*
12:32 celeron55    maybe you'll just learn to browse large files and then you're arguing with some other newcomer about them 8)
12:35 seolfor      hahaha may be bit isn't whoud be simpler to just split the freking file for example to 4 files by 1000 lines each and put them into folder called game? Ah ok will not asy anything more about it you are the boss here ;P
12:35 celeron55    as a matter of fact i'm not the boss
12:36 seolfor      but what about the codding standart you known.. spaces/tabs/brackets/naming etc ..
12:36 celeron55    http://dev.minetest.net/Code_style_guidelines
12:36 celeron55    i linked this already
12:36 seolfor      you are not ? i thought you are
12:36 seolfor      o .. sory i didn't notice
12:36 celeron55    i'm not much more than one of the core developers
12:38 celeron55    anyway - it's been long desired that game.cpp would be split to meaningful parts so if you can do it well, that might be upstream-worthy
12:39 celeron55    also the game state should be stored in a simple class or struct instead of local variables
12:40 celeron55    it has been discussed before too: http://irc.minetest.ru/minetest-dev/2012-08-12#i_2418283 http://irc.minetest.ru/minetest-dev/2013-10-21#i_3385585
12:40 seolfor      all right i will take a look as soon i will do somthing with this mouse configuration
12:41 seolfor      and i was i thinking why are there so many pull request unclosed on github with one are definetly usefull
12:41 seolfor      for ex. game confing in game etc ..
12:42 celeron55    some of them are waiting for testing, some of them are incomplete or buggy (as stated by the author), some of them are being argued about
12:42 celeron55    take a look at how many pull requests have been closed 8)
12:42 celeron55    there's just a huge flow of them so they buffer up a lot
12:43 seolfor      ah indeed ;]
12:43 celeron55    if you want, you could look through them and point out those that seem directly mergeable
12:44 celeron55    or that should be closed
12:45 celeron55    then maybe we find something that needs such tweaking that you can do
12:47 seolfor      and about that .. how should i report / made pull request
12:47 seolfor      should i made fork for my sell then report pull request ?
12:48 seolfor      and do i have to keep there linear history ? i realy don't like rebase functionalyty (simply i didn't used it to much)
12:49 celeron55    we use rebase for everything; but it can be done at the moment when the pull request is finally added upstream
12:50 celeron55    keeping all work-in-progress comments separate until that point is encouraged
12:50 celeron55    but yeah, it's best if you make yourself very familiar with rebasing
12:56 seolfor      sorry but "keeping all work-in-progress" what did you mean by that ? and aobut upstream, sory im not familiar with them
13:01 celeron55    s/comments/commits/
13:01 celeron55    work-in-progress means stuff that isn't yet finished (but will be finished after more is added to it)
13:01 celeron55    and upstream means the main minetest repo to which people base their personal forks
13:02 seolfor      hmy.. trivial explenation for trivial problem's but i get it now thx ;]
13:04 seolfor      i read http://dev.minetest.net/Code_style_guidelines but how about line ending chareacters? i hope you are using unix style ?
13:06 celeron55    of course
13:08 celeron55    i'm guessing someone should add things to the wiki based on this discussion as it's apparently missing things that it could have
13:15 seolfor      most of it is's already at wiki i will add notice about line ending charcter
13:17 seolfor      and about game.cpp i will be good exercise for me. I will made fork for my slef after i will do acualy somthing then we will be able to discuss it..
13:18 seolfor      i have to and for now thx for you time celeron55 ;]
15:44 seolfor      i have some stange mouse bechawior in arch linux for example if i will move mouse to left lets say about "2cm" (looks good) and then to right agin "2cm" (here is the problem) camera in game will move more to left side even more and only afther that to the right
15:45 seolfor      my question is: it's normal in newest commit (not linear mose movment) or it's only at arch linux bechewing strange ?
15:52 celeron55    i have no idea what could cause that
15:58 celeron55    also arch linux has not much special compared to other distros
15:58 celeron55    but it could be specific to your mouse driver or something
15:58 celeron55    or window manager
15:58 celeron55    or irrlicht version
16:00 seolfor      on windows is working qute well,
16:01 seolfor      i got irrlicht in 1.8.1-2 ( so i dont think it's a problem)
16:03 seolfor      it might be a drivers becouse im using vmware virtual machine i will check what i got installed, but I don't think window manager can have somthing with it
16:05 sfan5        trying to run 3d stuff in a VM is usually a bad idea
16:08 celeron55    sounds like the VM isn't working right
16:08 celeron55    which isn't surprising at all and probably can't be fixed
16:09 sfan5        >afaik those shadows are caused by invisible caves that are generated in the sky...
16:09 sfan5        wat
16:09 sfan5        apparently air != air
16:10 seolfor      ah .. i don't care any more i wan'a code somthing i will instal tomoro native linux any proposition for distro ?
16:10 sfan5        I would go for Xubuntu or Crunchbang
16:11 seolfor      i whould preper somthing based on arch .. ;]
16:11 sfan5        archbang?
16:12 seolfor      yep i was thinking about it.. or again plain arch + xorg + somthing
16:13 seolfor      and btw sfan5 are you deverloping under win or linux ?
16:13 sfan5        are you insane?
16:13 sfan5        as if anyone would dev under windows
16:13 sfan5        horrible...
16:13 sfan5        I use Xubuntu 13.10
16:14 seolfor      and how about ide .. dont tell me you are also uning vim ?
16:15 sfan5        I find vim hard to use
16:15 sfan5        gedit/geany, command line, gdb & valgrind
16:18 seolfor      a bit better then celeron but still it's to small for me ;] i think i will stick with qtcreatro for time beeing ..
16:18 sfan5        netbeans works good too
16:19 seolfor      i will check it but not today :p i'm messing with game.cpp :p
16:20 seolfor      i's give me so much plesure to ctrl+c/v 500+ lines of code ;]
16:27 seolfor      and btw what encoding are you using ? UTF-8 i hope so ?
16:28 BlockMen     "[18:13] sfan5: as if anyone would dev under windows", hey!
16:28 sfan5        seolfor: UTF-8 IIRC
16:28 seolfor      BlockMen, i whould if i could
16:29 BlockMen     seolfor, its possible ;)
16:30 seolfor      BlockMen, not for me i tried to make build like 4 day or about
16:31 BlockMen     seolfor, i also needed a lot time to get it run
16:32 BlockMen     and since then im trying to not touch the running system^^
16:32 seolfor      ye but's freeking unfair i made it runable under linux like in 5 min
16:38 sfan5        linux is superior~
16:38 BlockMen     so the windows users are the real skilled ;P
16:40 seolfor      windows don't care about deverlopers other then them that's the main diffrance
16:41 BlockMen     oh, thats not true. windows care about developers, but for windows exclusive only
16:41 BlockMen     *s/for/about
16:45 twoelk       in the newest minetest build I can rotate slabs but not blocks. Is that a bug or a feature?
16:47 sfan5        normal blocks look same at every side
16:47 sfan5        +te
16:47 sfan5        *the
16:52 twoelk       You can make patterns with wood for example
16:54 sfan5        you mean rotating y?
16:55 seolfor      for ex or at least 2 sides
16:56 seolfor      i mean first texture for 4 sides and secound for last 2\
16:56 twoelk       yeah, like creating a floor with wood planks, https://forum.minetest.net/viewtopic.php?id=9084
16:58 seolfor      and btw global functions/varialbes are trully painfull >.> are ther realy neccesery I highly don't think so
19:35 spillz       !tell seolfor to try Code::Blocks with CMake generated project files. Works well. irrlicht code completion could be better though.
19:35 ShadowBot    spillz: O.K.
19:36 sapier       "celeron55  vim and bash" ?? really?
19:37 sfan5        yes
19:37 sfan5        celeron55 always used just vim and bash
19:38 sapier       everytime I hear this I don't know if the one using it is genius or crazy :)
19:39 sapier       seolfor I know I'm gonna be the bad guy on saying this but what do you think about fixing some bugs to get known to code prior starting to refactor core components?
19:42 ShadowNinja  sfan5: I tried the range thing again, but I got the same issue.  The mapper writes outside the image causing a segfault, and if i increase the image size I get this: http://imgur.com/T784lwy.png  patch http://ix.io/bJ4/diff
19:43 ShadowNinja  Notice that the bottom left section has a extra lower layer, and there are block patches.
19:43 ShadowNinja  black*
19:43 ShadowNinja  The LevelDB backend works fine.
19:43 sfan5        static inline int64_t stoi64
19:43 sfan5        this sounds wrong
19:44 sfan5        the string passed by reference can change
19:44 sfan5        while the arg itself does not change (because reference)
19:44 sfan5        a static decl is therefore not right
19:44 sfan5        oh also
19:45 sfan5        I am merging redis database backend thing in 30 minutes
19:45 sfan5        celeron55 said it was okay as long as I maintaned it
19:45 seolfor      <sapier> everytime I hear this I don't know if the one using it is genius or crazy :) - i think it's propably bouth
19:46 ShadowNinja  sfan5: It's a static global function.  Meaning that it only exists in it's own module (source file).
19:46 sfan5        hm, ok
19:46 ShadowNinja  static has a few meanings in C++.
19:46 sfan5        Apr 12 11:46:49 <celeron55>	sfan5: i guess it can be added, but it's the first one that will go if some internal interfaces are to be changed and you aren't around to modify it
19:46 seolfor      sapier, about fixing bug's i dont have enything against
19:47 sfan5        ^ 'evidence'
19:47 sapier       and it's quite small difference between statics meaning ... but those small differences sometimes decide
19:47 ShadowNinja  sfan5: Does it use that hashing thing like SQLite3 and LevelDB?
19:48 spillz       What is with all this fiddling with backend databases and code formatting? these aren't the real problems with the game, but make up much of recent activity.
19:48 sfan5        ShadowNinja: dunno
19:48 sfan5        spillz: redis is even better than LevelDB
19:49 ShadowNinja  sfan5: It does, it uses getBlockAsInteger.
19:50 sfan5        what should it use instead?
19:50 sfan5        now is the time where I can change it
19:51 ShadowNinja  sfan5: It should seperate each number with a space or comma.  Or it should use seperate collums if redis supports multi-column keys.
19:51 sfan5        redis does not
19:51 ShadowNinja  columns*
19:51 sfan5        so it should use commas
19:51 proller      redis cant be faster than level
19:52 sapier       because of what reason proller?
19:52 ShadowNinja  proller: And why not?
19:52 sfan5        proller: lolwut, redis is always in-RAM | LevelDB is not
19:52 proller      also about block keys - https://github.com/freeminer/freeminer/commit/e1f7349bdae070856b0545fb7f2bb4a72b4c5bf2
19:53 sapier       proller can you please tell what makes you believe it can't be faster?
19:53 sfan5        are you telling me dummy is faster than redis?
19:53 proller      levedb - one process file access, redis - access with locking
19:54 proller      no, look at keys
19:54 proller      fm will use this for leveldb too
19:54 sfan5        lol
19:54 ShadowNinja  proller: Wut?  Why would you use strings rather than integers?
19:54 sfan5        I am about to use this too
19:55 proller      keys in leveldb is strings
19:55 ShadowNinja  proller: Yes, but dummy doesn't have the restriction.
19:55 proller      why need to jerk with s64 instead of direct use 10,20,-30
19:55 sapier       are you sure string operations are faster then shift operations?
19:56 proller      https://github.com/freeminer/freeminer/commit/b60c3bd0b95ca05f7326ce5b11a761e05bd41ae3
19:56 sfan5        ShadowNinja: should I use seperated number now or not?
19:56 ShadowNinja  You can store a number efficiently in a string using writeUNN.
19:56 ShadowNinja  sfan5: Yes, according to out discussion yesterday.
19:56 sfan5        apparently no
19:56 sfan5        so yes
19:56 proller      block store not slowest place in mt
19:57 sfan5        <sapier> are you sure string operations are faster then shift operations?
19:57 sfan5        what about that?
19:57 ShadowNinja  proller: So you want to make it slower?
19:57 proller      you can use whole blocks as keys without slowing
19:57 sapier       I don't think that much about writing it but parsing?
19:57 proller      ShadowBot,
19:57 ShadowNinja  sfan5: It's already written as a string.
19:57 proller      ShadowNinja, how much slower?
19:58 proller      1ns vs 2ns ?
19:58 sfan5        ShadowNinja: yeah, but then it is parsed into a number and shift ops vs. parsed into numbers
19:58 sapier       10e20 * 1ns is still quite a lot of time
19:59 ShadowNinja  proller: No, comparing an integer and comparing a string are very different time-wise.
19:59 sapier       there's a simple solution for this problem write a working benchmark  and provide values
19:59 * proller    facepalm
19:59 sapier       and description of benchmark to validate it's really doing what it's supposed to do
19:59 proller      paramat lua mapgen generate one block per 4 second
20:00 ShadowNinja  sfan5: True.  Storing it as one int is convoluted in the first place though.  And a int string is even worse.
20:00 ShadowNinja  proller: And?
20:00 sfan5        ShadowNinja: how should I do it then?
20:00 sfan5        I think I'll just leave it as is
20:00 sapier       so we're slowing down core because of it's not relevant in a already broken scenario? ... proller thats a non argument
20:00 ShadowNinja  sfan5: I don't know.  ;-)
20:01 sfan5        ok
20:01 sfan5        I'll leave it as is
20:01 ShadowNinja  sfan5: Any ideas about my mapper issue?
20:01 sfan5        no, not yet
20:01 sfan5        merging redis in 15 mins
20:02 proller      dont forget to rename minetest to sapiertest, or maybe sapierpilztest, or sapierpilzshadowninjatest, it will beat minecraft and all other games
20:02 sfan5        you forgot to rename freeminer to prollerminer
20:16 spillz       The database and code formatting so not explain the frame rate drops, draw distance issues, ugly gui (e.g. inventory screen), various issues with entities, shortage of nice entity and player animations, unfinished android port, and general lack of a cohesive game. It looks like you are no longer making minetest the game, but playing github the game.
20:16 jin_xi       nice rant
20:16 ShadowNinja  Good riddance (*roller).  sfan5: Well, look at it soon and see if you can find anything out of place.
20:17 sfan5        lol
20:17 ShadowNinja  Hah.
21:53 celeron55    well i agree with spillz and proller in their main points
21:53 proller      welcome to club ;)
21:56 hmmmm        what
21:56 hmmmm        on earth
21:56 hmmmm        that has got to be the most useless patch i've ever seen
21:56 hmmmm        just because something isn't the current bottleneck means that you have free reign to make it slower
21:56 hmmmm        if someone dropped a dollar on the street, wouldn't you pick it up?
21:57 hmmmm        sure, it's insignificant in the grand scheme of things, but look, it's an easy way to optimize - no - not optimize - just not screw up for no good reason
21:57 hmmmm        so it's like dropping a dollar bill on the street on purpose
21:57 hmmmm        it takes you more effort to do something stupid
23:00 RealBadAngel hi