Time  Nick         Message
00:19 iqualfragile ah, k
00:19 iqualfragile in that case there is some android application violating at least that license
03:30 Sokomine     problem: repeatedly "minetest/src/emerge.cpp:570: virtual void* EmergeThread::Thread(): Assertion '0' failed." with pretty recent git + a new world (no guarantee that the mods fit and work together as intended)
03:31 Sokomine     will try with current git
03:34 Sokomine     same happens with latest git. that the mods do have some problem is likely. the world i try to generate is completely new
06:21 hmmmm        man, proller's weather crap is horrendous from an organizational standpoint
06:23 hmmmm        anyway my guess is that the weather bug is due to activateBlock not being called on time before it gets sent out to the client
06:24 VanessaE     um...wut?  http://pastebin.ubuntu.com/6430401/
06:24 VanessaE     didn't someone else just report this earlier today?
06:25 VanessaE     hm, maybe not the same error.  anyway, that ^^^ just happened out of nowhere
06:28 hmmmm        I want to get rid of the weather crap
06:28 hmmmm        it simply does not fit
06:29 hmmmm        it uses too many high-level constructs and uses them in with low-level block initialization without regard to any program structure
06:30 hmmmm        basically when he first wrote it he just did whatever mangling with pointers and references he needed to in order to get the objects necessary to carry out the weather calls
06:39 nore         https://gist.github.com/Novatux/7510145 <-- what do you think of that?
06:39 nore         I'm probably going to rewrite the while force loading code
06:40 nore         and it is a draft of the new API
06:40 nore         far less hacky
06:42 hmmmm        there's nothing that can go wrong with reference counting
06:42 nore         what do you mean?
06:44 nore         hmmmm, perhaps that then: https://gist.github.com/Novatux/7510145/revisions
06:46 hmmmm        that's probably better
06:46 hmmmm        you really cannot trust mods
06:47 hmmmm        they can and will do everything in their power to act stupid with minetest and cause problems
06:47 nore         (ah, and I forgot: the loaded blocks are saved on map shutdown)
06:47 nore         hmmmm: that's the objective of the handler
06:48 nore         you can limit any mod to a maximum number of force loadings, and the mod cannot force load more than that number of blocks
06:48 hmmmm        oh guys, do you know what the 'p' key does in minetest by default?
06:49 VanessaE     no idea
06:49 hmmmm        that's the 'throw exception' button
06:49 nore         what?
06:49 VanessaE     that does nothing here.
06:50 nore         nor for me
06:50 VanessaE     wait
06:50 hmmmm        ahh nevermind
06:50 hmmmm        that's not an exception, it looked like a backtrace to me
06:50 VanessaE     http://pastebin.ubuntu.com/6430462/
06:50 hmmmm        I thought it was like something that had a problem nobody noticed
06:50 VanessaE     it does that for me
06:50 hmmmm        yeah that's normal
06:51 hmmmm        doesn't it look like something being a bug and trying to crash at first?
06:51 VanessaE     yes in fat
06:51 VanessaE     fact*
06:52 VanessaE     fat lotta good a debug stack is if it isn't attached to some sort of actual crash, no?
06:52 nore         the traceback always stops at (Leftover data: #5  void MeshUpdateQueue::addBlock(v3s16, MeshMakeData*, bool, bool))
06:52 VanessaE     noreqditto
06:52 hmmmm        well I think the intention was to figure out where some thread is stuck in if it's taking too long to do something, or caught in an infinite loop
06:52 VanessaE     nore, ditto*
06:56 nore         hmmmm, are you ok for an API like that? should I code it?
06:56 hmmmm        i'm pretty indifferent toward it
06:56 hmmmm        why don't you ask people who would actually be using this
06:56 VanessaE     I assume he has things like technic, mesecons, and pipeworks in mind.
06:57 nore         yep...
06:58 nore         or a more general force load mod, that could allow you to place force loaders
10:25 PilzAdam     nore, I guess this should be done in minetest_game too: https://github.com/PilzAdam/pilztest/commit/a6a23b4f0ded32fa1708e88ea9007df9a025a701
10:26 nore         indeed... (do I make it, or do you want to do it yourself?)
10:27 nore         no, it is already done...
10:27 nore         https://github.com/minetest/minetest_game/blob/master/mods/default/nodes.lua#L160
10:29 PilzAdam     oh, good
10:30 PilzAdam     then another thing I noticed: https://github.com/minetest/minetest_game/commit/9fdbc1f4072dcf2296af9bcce6ec92bfa75972ca this adds some print() calls; print() is an alias for minetest.debug(); I think it should be changed to minetest.log("action", ...)
10:31 nore         didn't know that... but yeah, it should probably be changed (was tree growing action?)
10:31 PilzAdam     yea, I checked that
10:31 nore         oh, and I wrote a draft for the force loading API too: https://gist.github.com/Novatux/7510145
10:32 PilzAdam     woah, these abms use spaces as indentation
10:33 nore         what? when I coded them, I wrote tabs... (and I'm sure they were tabs)
10:33 PilzAdam     well, BlockMen changed that I guess
10:33 nore         ah, yes, I had forgotten that
10:34 PilzAdam     also, we should use mintest.get_item_group()
10:34 PilzAdam     (for is_soil)
10:34 nore         what does it use now?
10:35 PilzAdam     it checks it in minetest.registered_nodes[]
10:35 nore         and what is the problem with that?
10:35 PilzAdam     we have to check for nil and 0
10:36 nore         ah, didn't know that
10:36 nore         minetest.get_item_groupd returns 0 if not in group, is that right?
10:36 PilzAdam     yep
10:40 nore         should we add a minetest.swap_node function too? because now, it is very annoying that the formspec freezes when a furnace turns on/off, etc
10:43 Sokomine     any ideas what may cause that "minetest/src/emerge.cpp:570: virtual void* EmergeThread::Thread(): Assertion '0' failed." i'm experiencing? i havn't been able to nail it down to one particular mod yet
10:44 PilzAdam     nore, also the trunk should be able to replace leaves of other trees
10:44 VanessaE     Sokomine: related?  [11-17 01:24] <VanessaE> um...wut?  http://pastebin.ubuntu.com/6430401/
10:44 VanessaE     PilzAdam: the same should be true of the L-systems generator, btw/.
10:45 VanessaE     I griped about that months ago
10:45 nore         PilzAdam, perhaps even a grown_trough group or something like that
10:46 nore         nodes that can be replaced by growing trees...
10:48 PilzAdam     Im searching for other print()s in minetest_game currently; should it be logged when a player takes an item from the creative inventory?
10:48 Sokomine     vanessae: might be related. except that in my case the block was placed by a mod at mapgen time (new world, first visit to it) and not by a player
10:49 nore         PilzAdam, I think it is currently...
10:49 VanessaE     bbl
10:49 PilzAdam     nore, but with a print(), so I change that to actionstream
10:49 nore         yes, should be better
10:51 Sokomine     pilzadam: with creative inventory, it's probably less important. taking something out of/placing into formspecs of mapnodes that have an inventory ought to get logged. might avoid problems like with technic where thieves can't be identified
10:53 nore         Sokomine, that's a problem with technic... ;)
10:54 nore         a question about a possible minetest.swap_node: should it update light? should it be reported for rollback?
10:57 PilzAdam     nore, https://github.com/PilzAdam/minetest_game/commits/master
10:57 PilzAdam     are these 2 commits ok?
10:57 Sokomine     nore: it's not only technic. all these mods that support nodes with an inventory + formspec have to log what's taken and placed manually right now. logging that in general might be a good idea
10:59 nore         PilzAdam, yes
11:01 PilzAdam     whenever we create Lua code styleguidelines it should state that print() calls should be removed in a release
11:01 PilzAdam     now lets check builtin
11:06 PilzAdam     hmm... is modmgr.lua:935 "important" enough to be logged to errorstream?
11:08 nore         what is it?
11:09 PilzAdam     oh, right, its just mods being set in world.mt but not found in the modlist
11:09 PilzAdam     I log it to infostream then
11:09 pitriss      Important message should be warning if it's not fatal IMO
11:12 PilzAdam     nore, https://github.com/PilzAdam/minetest/commit/367b5382a394e853128092afd23fc7e360d369d1
11:12 nore         ok
11:19 Sokomine     regarding the crash i had: seems that's a more general message. the problem appears to lie in the way glooptest places treasure chests. the more exact error message is "ERROR[EmergeThread0]: ERROR: An unhandled exception occurred: basic_string::_S_construct null not valid"
11:22 Sokomine     while that particular mod is not important for now, i'd like to understand the error message and what causes it
12:47 nore         Any thoughts on this? https://github.com/Novatux/minetest/commit/649947b44aecd8a36f24458891f9fde84c02ae87
12:48 nore         (minetest.swap_node function)
13:00 sfan5        may I revert https://github.com/minetest/minetest/commit/3985c01ad71159b888677d0fe2019d7a05debeed ?
13:00 sfan5        it breaks things for me
13:02 nore         what does it break?
13:03 sfan5        https://forum.minetest.net/viewtopic.php?pid=118534#p118534
13:03 sfan5        linking
13:03 nore         you should perhaps wait for sapier to join then... he knows that more that I do
13:04 nore         ah, and do you have any thoughts on https://github.com/Novatux/minetest/commit/649947b44aecd8a36f24458891f9fde84c02ae87 ?
13:04 sfan5        I don't see why I should wait for sapier, it worked perfectly before..
13:05 sfan5        you should document the packet format in clientserver.h
13:05 nore         yep, but for him, that fixed things...
13:06 nore         sfan5, TOCLIENT_ADDNODE isn't documented either
13:06 sfan5        then that should be documented too
13:08 nore         oh, and do I need to change protocol version?
13:09 sfan5        for that, I don't think you need to
13:09 sfan5        s/,/?/
13:09 sfan5        the protocol is raised every release anyway
13:09 nore         what will happen when a client receives the packet then?
13:09 sfan5        nothing
13:10 nore         ah, so the swap_node call will not work for the client
13:10 sfan5        exactly
13:11 nore         so, should I change protocol version, and if client does not know it, send add node and set metadata?
13:15 nore         sfan5, and what about that? https://gist.github.com/Novatux/7510145
13:16 sfan5        hm
13:17 sfan5        I don't like the minetest.get_force_load_handler(), adds unnecessary java-like abstraction
13:17 nore         yeah, but how do I know which mod is asking for force load?
13:19 nore         without that, I don't see how it could be done...
13:20 sfan5        ..
13:20 sfan5        why do you need to know which mod force loads?
13:21 nore         sfan5, to add a limit to each mod, and so that mods handle their force loaded blocks separately
13:21 nore         i.e.: you can limit one mod to 5 blocks force loaded, and another one to 100
13:21 sfan5        mhm, then there is no other way
13:22 nore         and if one mod does not use wisely force loading, you can set a maximum of 0 blocks...
13:23 nore         the idea would be that the base API would only have 2 functions: minetest.force_load_raw(blockpos) and minetest.unload_raw(blockpos)
13:24 nore         and the remaining things would be done in builtin (including saving force loaded blocks on server shutdown, etc)
13:25 sfan5        that would be the best way to do it
13:26 nore         and do you think that one would have any chance to get merged, or that it will wait 1 year as the other ones?
13:27 sfan5        if PA doesn't disagree it will probably get merged quickly
13:28 nore         ok, so I will code it... (I reckon it will take 1 or 2 weeks)
13:30 nore         and about the swap node thing, no protocol version change, and a swapped node does not get swapped on old clients, is that ok?
13:49 celeron55    sfan5: don't revert; it was previously wrong, at least according to sapier
13:50 celeron55    ...but i think the build instructions weren't updated for it
13:50 celeron55    nor the build scripts or anything
13:50 celeron55    the thing is that mingw contains zlib already by itself which is linked in "mingw way", and with mingw that should be used instead of the downloadable one which is linked in "msvc way"
13:52 celeron55    this is an intolerable state though; either documentation needs to be updated or that has to be reverted
13:53 celeron55    why do people always leave things halfway done? it's terrible
13:55 sfan5        why do we need to fix something that isn't broken? that's terrible
14:01 celeron55    ask sapier
14:03 celeron55    it was something about possible issues coming from some libraries using a different zlib or so
14:03 celeron55    (those issues have never occurred)
14:29 hmmmm        oh shit don't tell me that sapier's mingw fix broke things
14:41 PilzAdam     hmmmm, sapiers mingw fix broke things
14:41 PilzAdam     my build fails too
14:41 hmmmm        we revert it
14:42 hmmmm        sapier has a track record of breaking stuff, what the hell happened? :(
14:42 hmmmm        is he getting bored with minetest and not really paying much attention to it or something?
14:43 PilzAdam     well, maybe celeron55 is right and sfan5 and me just need to update your scripts
14:43 hmmmm        try that first...
14:43 PilzAdam     but then he should at least tell us how (our scripts are based on buildbot.sh, so that needs to be updated too)
14:43 sfan5        I have no clue what I need to update
14:43 hmmmm        shit okay
14:43 hmmmm        want to just revert until we get more information from sapeir
14:44 sfan5        ZLIB_LIBRARY is set to the .lib
14:44 sfan5        that is apparently wrong
14:52 hmmmm        who was the guy who pasted that "assertion (0) failed" thing
14:52 hmmmm        oh nevermind
14:52 hmmmm        VanessaE, I found your bug
14:53 hmmmm        on database-sqlite3.cpp:225, sqlite3_column_blob() failed and returned NULL
14:55 hmmmm        "The return value from sqlite3_column_blob() for a zero-length BLOB is a NULL pointer."
14:55 hmmmm        so evidently the root cause of this was that a mapblock failed to save evidently and placed 0 bytes into the data column
14:56 hmmmm        although the database abstraction code should've been checking the return value before passing it along to a std::string ctor
15:09 hmmmm        PilzAdam did you see that patch from last night
15:10 PilzAdam     which patch?
15:10 hmmmm        https://github.com/kwolekr/minetest/commit/4c2e6fa5131d6fd0d99d024c09fbfeec191cc9df
15:10 hmmmm        i can't test that
15:11 PilzAdam     proller and VanessaE reported the issue
15:11 hmmmm        oh, change is_connected_player to is_player_connected in the lua before trying that
15:11 PilzAdam     I guess one of them has to run it for some time to see if still happens
15:22 nore         any thoughts on https://github.com/Novatux/minetest/commit/649947b44aecd8a36f24458891f9fde84c02ae87 ?
15:22 hmmmm        yup
15:22 hmmmm        my thought is that it's not being merged until after 0.4.8
15:26 hmmmm        VanessaE, cherry-pick this commit and try it out:  https://github.com/kwolekr/minetest/commit/c0ab4e495b2f76f08265a654798e21dfd95a6775
15:27 Calinou      why not release 0.5 instead of 0.4.8, due to the amount of changes we done?
15:27 Calinou      most projects would increment their "minor" version for such stuff
15:29 nore         IIRC, 0.5.0 would be a big protocol change...
15:30 nore         perhaps all protocol changing pulls at once, or something like that
15:43 Calinou      it would make the project seem to be more "advancing"
15:43 Calinou      protocol change doesn't have to be a specific version...
15:43 nore         I mean, incompatibility with previous versions
15:44 nore         http://dev.minetest.net/TODO#Big_protocol_changes
15:46 PilzAdam     Calinou, we have already plans for 0.5.0
15:58 hmmmm        https://github.com/minetest/minetest/commit/aa172bdda47555096524ac2331f0a285666f037b
16:07 kahrl        hmmmm: the same should be done in database-leveldb.cpp
16:07 kahrl        (I have no clue why that code is duplicated)
16:08 kahrl        guess it isn't needed in database-dummy.cpp
16:08 hmmmm        ahhh
16:08 hmmmm        yeah that's true
16:14 hmmmm        there's no real documentation for leveldb .
16:27 hmmmm        although honestly the leveldb one wasn't necessary, I presume you'd get a SerializationError with a blank block.  the reason why I had to fix that is because it was causing a really screwed up exception from implicitly converting NULL to a std::string (so now at least the player has the option of ignoring it)
16:28 hmmmm        I suspect VanessaE is the one who comes up with all these errors because her world is so damn old
16:28 nore         why would that cause problems?
16:29 hmmmm        there is literally no way to trace the cause of the blank mapblock there.  we don't know if it was a bug that was fixed or any information at all because there's a huge gap of time between those events
17:34 sapier       sfan5 the msvc change requires to change cmake zlib configuration for your build (if you do a mingw/cygwin build)
17:35 PilzAdam     sapier, you should fix that in buildbot.sh
17:36 sapier       buildbot was broken before
17:36 sapier       no idea how to make that thing work on any other system then the one who created it
17:36 OldCoder     sapier PilzAdam sfan5: thexyz suggests that I file an issue report for hangs during "item definitions". This is observed repeatedly by different people. I may wait until kahrl httpfetch is merged to rule out network glitches. I wished to ask if you have observed such hangs yourself.
17:36 PilzAdam     sapier, it works for me, and sfan5
17:36 PilzAdam     and c55
17:37 sapier       oldcoder httpfetch won't help as this most likely is a bug in shader loading
17:37 sapier       yes and doesn't work for me PA ;-)
17:37 sapier       that script is far from generic
17:37 PilzAdam     well, your patch cant go in as long as it breaks buildbot.sh
17:38 PilzAdam     we cant make your system work and break everyone else's
17:38 OldCoder     sapier, Hmm... Is there a known shader loading bug? Should I file an issue?
17:38 sapier       then revert it
17:38 PilzAdam     hmmmm already did that
17:39 sapier       ok I don't have any interest in fixing buidbot.sh as imho it's broken by design we can merge my cleanup once buildbot is fixed by someone else
17:40 sapier       oldcoder if I remember correct shader preparation takes to long to complete resulting in timeouts that's what you see ... I'm not yet up to date but if my fix fore vanessae's crash was merged you should see an error on timeout
17:40 sfan5        sapier: broken by design? where exactly?
17:40 hmmmm        so we're shipping bullshit that doesn't compile on mingw
17:41 hmmmm        that's good
17:41 OldCoder     sapier, Is the error on timeout intentional? Should I update in a few days?
17:41 sapier       buildbot downloads binarys instead of downloading sources and compiling on its own this way the libs wout REALLY match and not be some combination that happens to not crash
17:42 OldCoder     When did buildbot break? I was able to build Win32/64 MT clients OK two weeks ago
17:42 sapier       the error message is intentional, whole design is based on asumption a timeout never occurs ... that's been root cause for vanessae's crash
17:42 OldCoder     sapier, all right; so I should update after things settle down. Will the timeout be increased or do I misunderstand?
17:43 sapier       it's broken for latest mingw on win7 64
17:43 OldCoder     Ah
17:43 OldCoder     But cross-compile for Windows works
17:43 OldCoder     OK
17:43 sapier       yes I assume it works for the compiler combination it was built at
17:43 hmmmm        sapier, I'm gonna merge the RequestQueue fixes, any remarks before I do it?
17:44 sapier       there are two things in there
17:45 sapier       first commit is for broken multicaller handling (which isn't used at all atm) second is fix for vanessae crash
17:46 sapier       but both should be fine
17:46 hmmmm        the doxygen comment
17:46 hmmmm        i guess i might as well merge that too
17:46 hmmmm        pffffffffffffffffffffffffffffffffffffff
17:47 sapier       the information of that comment is required ... style ... as always a matter of personal preference
17:48 hmmmm        are you aware of any other critical bugs?
17:48 hmmmm        there's a recent EmergeThread assertion that sokomine got but I didn't get any more information about that
17:49 sapier       if they occured the last two days no I was offline
17:49 hmmmm        it's actually probably a mod
17:49 sapier       emegrge thread singleplayer or server?
17:49 hmmmm        I just fixed something that tried to implicitly convert NULL to a std::string, triggered when a MapBlock is blank in the database, soo
17:49 hmmmm        I don't have any more information about that one, it'd be nice if we had something like a backtrace
17:50 sapier       ok that's not a good idea of course :)
17:50 sapier       but a quite common error
17:50 hmmmm        there's also this https://github.com/kwolekr/minetest/commit/c0ab4e495b2f76f08265a654798e21dfd95a6775  but I am waiting for people who had the problem to test it because I can't
17:50 sapier       ok so atm there's only the modmgr stall left on way to 0.4.8 ?
17:51 hmmmm        I don't know about the modmgr stall, I mean that's a really hefty feature
17:51 hmmmm        isn't there any way to work around it?  can you fix over the stall if the modmgr is disabled in the config? (for people who don't use it at least)
17:51 sapier       it's a design issue with current implementation lacking async file doenload
17:52 sapier       no the only fix is merging the async things ... either mine or httpfetch
17:52 hmmmm        httpfetch isn't ready
17:53 sapier       then I suggest adding the async things they just add features and don't change existing critical things
17:53 sapier       risk of adding critical bugs is way less ... and httpfetch is mergable to async
17:54 PilzAdam     sapier, a bug for you: while it downloads stuff after a click on "Online mod repo" change the tab; the result is a screwed up menu
17:54 hmmmm        so your async thing is also a mechanism for inter-thread communication between lua instances
17:54 hmmmm        I really don't know about all this, it's huge
17:55 hmmmm        i'd rather wait on adding lua async
17:55 sapier       hmm I thought I already fixed that on PA I'll have look at it
17:55 hmmmm        there's really no way you can just temporarily disable it for people who don't use modmgr...?
17:56 sapier       no the stall in favourites is not related to modmgr but has same root cause as modmgr stall
17:56 hmmmm        well it's not like we're trapped with 0.4.8 for a half year
17:57 hmmmm        so I think I'd like to follow up with a 0.4.9 with a lot of the features that were being held back here
17:57 hmmmm        a month tops
17:58 sapier       considering it's not that far to christmas I expect a lot of new features within the next months so I guess we'll have a lot of things to merge
18:00 sapier       PilzAdam are you sure you did test the latest version of async?
18:00 PilzAdam     Im using latest master
18:00 sapier       are the async things merged?
18:01 sapier       ok I think I need to have a closer look what happend the last two days :-)
18:02 hmmmm        the async things are NOT merged yet and they won't be until after 0.4.8 is released
18:03 sapier       hmmmm I'm not sure if releasing with favourites and modmgr stall is really a good idea
18:03 hmmmm        people have been living with it up until now, and I'm more worried about some fatal bug being there
18:03 sapier       especially the modmgr things will get worse once there are more mods in there
18:04 ShadowNinja  hmmmm: Since you have a player object wouldn't simply player:is_connected() be better?
18:04 sapier       but if we plan to release another version this year it should be ok ... just waiting till march 2014 is way to long
18:04 hmmmm        the naming?
18:05 hmmmm        there's is_player already, so why not is_player_connected?  besides, what about get_player_name
18:05 * OldCoder   requests priority for bug fixes such as the possible shader hang; these seem more important than new features
18:06 ShadowNinja  Well get_name() could return something like tnt:tnt, while tnt can't be connected...
18:06 sapier       oldcoder the errormessage is now in master if your hang is what I guessed it to be it should be obvious now
18:07 OldCoder     sapier, I shall rebuild and test for you then
18:08 sapier       if you can proove my guess is correct it'd ease finding it
18:08 sapier       but it should be a client issue so no need to rebuild server
18:08 OldCoder     I shall rebuild in the next two hours; just need to finish editing some posts and explaining to a startuper how to do MP3 to Ogg conversion
18:09 OldCoder     Oh! Client issue... I will need to find people to try the rebuilt clients
18:09 * OldCoder   shrugs and will report results
18:09 sapier       if my guess is correct yes :-) if I'm wrong ... back to start ;-)
18:09 adama_       sorry if im butting, i was wondering if any one is having troubles with plants not growing in the latest git ??
18:10 sapier       hmm did you ask in minetest? people there usually spend more time playing
18:10 adama_       ya no luck
18:11 adama_       every thing was fine, then i compiled latest git and now no plants are growing
18:11 sapier       interesting when did it work last?
18:11 adama_       day be for
18:11 sapier       yesterday?
18:12 ShadowNinja  adama_: What kind of plants? Trees?
18:12 adama_       no day b 4 that
18:12 sapier       what exact version did you build?
18:13 adama_       i have the plants mod with flowers junglegrass and poison ivey
18:13 adama_       but no trees will grow from sapling or new plants spawn like flowers
18:13 PilzAdam     have you updated minetest_game too?
18:13 ShadowNinja  adama_: Perhaps you should speak with VanessaE in #minetest. Update minetest_game to get trees.
18:14 adama_       ya i delete old minetest folder then download and compile same as allways
18:15 adama_       ok thanks ShadowNinja
18:16 adama_       this is how i compile MineTest http://www.distrogeeks.com/install-minetest-ubuntu/
18:16 ShadowNinja  Fine if I close #1010?
18:16 PilzAdam     ShadowNinja, I guess you can
18:18 PilzAdam     adama_, thats outdated
18:19 PilzAdam     see README.txt for up-to date build instructions
18:19 adama_       hmmm is there a diff git then i can use
18:19 adama_       that would explain my issue maybe
18:21 adama_       ok PilzAdam looking now
18:29 adama_       ok im off to do more reading thank you every one for the help
18:29 adama_       bye
18:42 VanessaE     (still reading the backlog);  hmmmm, c0ab4e49 applied and the related changes to mobs removed.
18:43 VanessaE     whether we can trace the cause of a 'null' block, shouldn't the engine just refuse to bother with it, and perhaps even delete it?
18:44 VanessaE     though I see you also have a patch to handle those null blocks.  do I apply this also?
18:48 Sokomine     hmmmm: i had that assertion bug, and vanessa said she had encountered it before as well. there's more in the debug.txt, which points to a particular mod, but i don't get a clue from the error message
18:51 VanessaE     ok, I'm back.  hi.
19:01 Sokomine     here's the complete error message: http://pastebin.com/N4eLP62B   i've added some debug messages to the suspicious glooptest mod. when the mod is run on its own, it seem to work. in the admittedly mod-rich "game" (singleplayer) i intend to use it it is just a question of seconds until the bug occours, and it seems each placement of such a chest fails
19:08 adama_       just wanted to let you know PilzAdam you was 100% correct and that fixed my issue thank you verry much
19:09 hmmmm        VanessaE, I was briefly considering writing you a script to go through your entire map and remove bad blocks like that
19:10 VanessaE     hmmmm: it would seem to me that if the engine finds a block it thinks should be perpetually ignored, it would be prudent for the engine to just delete that block altogether...provided sqlite can do so safely.
19:10 VanessaE     (s/sqlite/the database backend/)
19:10 hmmmm        it's more of a mapgen problem I'd say
19:10 hmmmm        if you have one bad block it'll try to generate that entire chunk and overwrite everything else you've got
19:11 hmmmm        with v6 in particular you'd get all your buildings in the surrounding area covered with a nice layer of grass
19:11 VanessaE     yes, seen that before :P
19:12 VanessaE     ick..
19:25 sapier       c_internal.cpp Line 60 makes std::string from char* pointer without prior check
19:27 VanessaE     what's the status with https://github.com/sapier/minetest/commit/ffcf672125248926ec5481ce2c8578d87aa1011e
19:27 VanessaE     is this going to be merged soon?
19:28 PilzAdam     VanessaE, already done: https://github.com/minetest/minetest/commit/b2d9205796eef23fd5d9a436d438fa2ca31ec21a
19:28 VanessaE     oh!
19:28 VanessaE     ok
19:28 VanessaE     as usual, I'm a day late and a dollar short :P
19:38 sapier1      ShadowNinja can you fix c_internal.cpp L60?
19:39 ShadowNinja  sapier1: GTG right now, I will check it later.
19:39 sapier1      ok
19:40 hmmmm        "later"
19:40 hmmmm        I wanna release this like, soon
19:41 hmmmm        VanessaE, how is the get connected players fix working out so far?
19:41 VanessaE     hmmmm: get http_fetch in.  it (if done right) fixes the broken fetching of remote media
19:41 hmmmm        httpfetch is not ready yet first off, second of all, no new features
19:41 VanessaE     well, so far so good.  No crashes yet but the servers aren't super-active right now
19:41 VanessaE     httpfetch isn't a new feature.
19:41 VanessaE     it's also a bugfix.  remember, remote media is busted?
19:42 hmmmm        erm
19:42 PilzAdam     do we get #1000 in before 0.4.8?
19:42 hmmmm        kahrl himself says it's 'not ready', and besides, sapier's async api can solve that
19:42 hmmmm        yes we do pilzadam
19:42 VanessaE     PilzAdam: I've had #1000 in place since c55 first asked
19:42 VanessaE     it seems okay, but I haven't fully wrung it out yet.
19:43 VanessaE     hmmmm: well, if sapier's code can fix it, that's fine.  kahrl's code is shit-hot fast though :)
19:46 * OldCoder   has tested httpfetch. Does this help?
19:46 OldCoder     s/httpfetch/http_fetch/
19:50 sapier1      my async api does not collide with kahrl's changes it's basicaly the interface to lua while kahrl's changes are interface to webserver
19:57 hmmmm        sorry guys I don't mean to be pushy with the release but I do
19:58 hmmmm        0.4.8 was supposed to come like a month and a half after 0.4.7 but look at it
19:58 Sokomine     oh? a fix for the object duplication? that sounds very fine
19:58 PilzAdam     I dont feel very comfortable to release that fast
19:58 PilzAdam     we wanted to pack some other games besides minetest_game
19:58 PilzAdam     what do we do about that?
19:59 sapier1      imho that's a 0.4.9 issue (with targeted 0.4.9 christmas release ;-)
19:59 sfan5        0.4.9 should be released christmas?
19:59 Sokomine     hm. how many are there? there are diffrent clones of mc, dwarves, realtest, the space game mauvebic is working on, technic to a degree...
20:00 sapier       sfan5 :-) that's first time to say this
20:00 Sokomine     adding some mods from which players could choose from might be more helpful
20:01 PilzAdam     Sokomine, the problem with mods is that they all base on minetest_game, and thats not good
20:01 hmmmm        holy shit
20:01 hmmmm        christmas is way sooner than I thought it would be
20:02 sfan5        should we maybe include some mods in 0.4.[89] aswell?
20:02 PilzAdam     sfan5, modstore should do that
20:03 hmmmm        by the way about the "regular liquids not flowing" issue
20:03 hmmmm        that was basically caused by proller reducing the liquid queue for regular liquids by a factor of 10
20:03 PilzAdam     cant we just increase liquid_loop_max again?
20:03 hmmmm        we could, do you want to do that?
20:04 hmmmm        MAYBE it works for his finite liquid crap, but not so much for regular liquids.  he freaking broke liquid flowing for any large-ish liquid flow caused by caves
20:04 PilzAdam     defaultsettings.cpp and minetest.conf.example, anywhere else to change it?
20:04 hmmmm        no
20:06 PilzAdam     https://github.com/minetest/minetest/commit/a55c073ce9bd78a6ebbb867ab4db2ec611eb66cf
20:06 hmmmm        yuup
20:06 PilzAdam     hmmmm, oh, btw, I dont want to stop your rant, but it wasnt proller
20:06 hmmmm        I thought it was?
20:06 PilzAdam     it was me
20:07 hmmmm        I could've sworn he reduced it because it wasn't needed
20:07 hmmmm        when he added the randomly-remove-liquid-from-queue stuff
20:07 sapier       do we want to add a gamestore in 4.9?
20:08 PilzAdam     sapier, a _store_, not a tool to create games
20:08 sapier       we already have a tool so I was talking about that one
20:08 PilzAdam     just wanted to make that clear
20:08 PilzAdam     that tool should be removed from the mainmenu, btw
20:09 sapier       that tool is usefull if you download mods and want to use them
20:09 sapier       at least if you don't want to configure each world again
20:10 hmmmm        by the way, how's minetest_game looking for the release
20:10 PilzAdam     an option to copy world configs should be added then
20:10 hmmmm        any nagging issues with that?
20:10 PilzAdam     but games arent something that every user should create
20:10 sapier       that's one opinion PA
20:10 hmmmm        I realize a lot of the time a pull request in the core also needs a minetest_game update that we forget about
20:11 sapier       imho a game is just a collection of mods and everyone should be able to e.g. remove carts monorail simplemobs or whatever from his favourite game
20:11 VanessaE     regarding water flowing, hmmmm, we slowed it down a while back because we found it was causing major CPU load for any kind of largish waterfall, fountain, etc.
20:12 PilzAdam     sapier, what you talk about is a modpack
20:12 hmmmm        water flowing should be optimized instead of gimped like that
20:12 VanessaE     though at the time, we dod at least still have it flowing okay, if slowly.
20:12 VanessaE     did*
20:12 hmmmm        from what I understand that's the major CPU eater in the serverthread
20:12 PilzAdam     sapier, a game is more than, it has a goal and fixed set of features; the "mods" in it are more "modules" than mods
20:12 sapier       imho terms modpack/game are that similar they could be merged without loss of many information
20:13 PilzAdam     VanessaE, the problem is that it doesnt work in singleplayer with 1000
20:13 PilzAdam     VanessaE, and 10000 isnt a problem in singleplayer
20:13 PilzAdam     sapier, then you didnt get the idea of games
20:14 sapier       lol so you are in possession of the one and only truth PA? *smile*
20:14 PilzAdam     sapier, the game creator spends a lot of time thinking about the game, its features, what to add and what not to add
20:15 sapier       a game is (technicaly) nothing different then a modpack ... idealistic goals are not usefull for technical decisions
20:15 hmmmm        is that REALLY the case most of the time?  lol
20:16 sapier       it's not worth to have the term "game" just tell ppl some modpacks are "better" then others
20:18 PilzAdam     you should compare games to other games (_real_ games, outside of Minetest), and dont think about Minetest related stuff only
20:18 sapier       then adding multiple "games" to official release would be bogus
20:19 PilzAdam     well, we could either release the engine without games, or add some games to it; users have learn that Minetst is an engine, not a game
20:19 sapier       and third option is releasing multiple packages
20:20 PilzAdam     thats even better
20:20 PilzAdam     so people go to their package manager and say "I want pilztest", and the Minetest engine is just downloaded like a normal dependency
20:21 sapier       should be our goal yes but that requires some changes
20:22 sapier       I'd prefere to see how windows users use modstore prior deciding how to continue
20:22 PilzAdam     the problems with mods only is that they all base on minetest_game; they dont differ as much as games do
20:22 sapier       do they really do?
20:23 PilzAdam     mods cant do this: https://forum.minetest.net/viewtopic.php?id=6636
20:23 PilzAdam     well, they technically can, but they would need to disable minetest_game completly
20:24 PilzAdam     and then they are in-fact games
20:24 sapier       I still don't understand why you seem to be obsessed games are anything special
20:25 sapier       minetest_game is just a collection of mods not more not less
20:25 PilzAdam     I think it would be best for Minetest in the long-term if people get the engine idea
20:25 PilzAdam     sapier, minetest_game is a shitty game, its boring and featureless
20:26 hmmmm        and you made it that way!
20:26 sapier       I understand your frustration as you tried to make it better for some time but you'll have to live with fact that someone only designing a game will make better games than you as you split your attention through game/mod/engine
20:26 Exio4        why would minetest.get_connected_players even have non-connected players?
20:27 hmmmm        Exio4, because of the gap in time when on_leaveplayer gets executed and the environment step gets executed
20:28 Exio4        ah
20:28 Sokomine     pilzadam: i know how important that engine/game distinction is here on the channel, but it does not feel that important when actually playing the game. for that, it is good to have a common base - and that's minetest_game for me - with some crafts and blocks that are the same (more or less...) on all servers
20:28 hmmmm        i explained it completely in the channel yesterday i think
20:28 PilzAdam     the problems with minetest_games are different
20:29 Sokomine     no need to learn how to make a chest or a pick anew most of the time. you can concentrate on making it special the way you want (or the server owner decided to go)
20:29 PilzAdam     Sokomine, the problem is that when people usually say that Minetest isnt as good as MC they refer to minetest_game instead of the engine Minetest
20:30 hmmmm        that's because the only people aware of the different games aren't casual enough to compare it to minecraft :p
20:31 Sokomine     perhaps "selection of mods run on a server" is more important for most actual players most of the time. if a player looks for a rich survival-orientated game with mobs and what not (basically a copy of minecraft), then minetest_game alone will be disappointing
20:31 sapier       PilzAdam modstore isn't in any official release let people decide again if they have choice
20:31 PilzAdam     hmmmm, is_player_conneted() is the wrong name IMO, is_connected() is better since you already have a player ref
20:31 PilzAdam     hmmmm, also lua-api.txt doc missing
20:31 hmmmm        yes because it's an internal only function
20:31 hmmmm        remember we went over all this?
20:32 PilzAdam     then why not hide it outside of builtin?
20:32 hmmmm        iunno
20:32 Sokomine     yes, modstore will be very intresting
20:33 PilzAdam     sapier, mods based on minetest_game dont show the full potential of the Minetest engine
20:33 sapier       true but you can't have everything at once PA
20:33 sapier       minetest will evolve as it did till now
20:33 PilzAdam     who said that we are in hurry with this?
20:34 Sokomine     perhaps that minimal development game is confusing as well. new players will start with that. then they'll read that there's minetest_game and they'll think: that's it. later on, they'll learn that there are mods which do what they specificly want
20:34 PilzAdam     I just explained what I think should be the genral direction of Minetest
20:34 sapier       I guess we won't be able to find the one and only direction :-)
20:35 PilzAdam     we have to, directionless development sucks
20:35 sapier       but imho that's not a weakness but most positiv feature
20:35 sapier       no it doesn't linux doesn't have a direction too
20:36 hmmmm        well I have a direction
20:36 hmmmm        it's quite in line with what celeron wants too
20:36 sapier       we have some basic principless "make engine as generic as possible" thats a quite good common base
20:37 hmmmm        I'll tell you the big things that I am really looking forward to:
20:37 hmmmm        - client side lua
20:37 hmmmm        - more advanced environment synchronization
20:37 hmmmm        - hardware lighting
20:37 hmmmm        - the mapgen realm crap
20:38 PilzAdam     sapier, exactly: _engine_; not: engine for a single game that is modified via mods
20:38 hmmmm        - good mobs
20:38 sapier       I have good mobs hmmmm ;-P
20:38 hmmmm        sapier:  I mean even better mobs.  mobs that would make you shit your pants.
20:38 hmmmm        that's only possible with client side lua
20:38 sapier       it is
20:39 sapier       mobs suffer from spawning and lack of async
20:39 sapier       last one is on my list and gradualy improving ... of course it's not gonna be done tomorrow
20:40 sapier       I try to add it step by step to have a working stable version anytime till target is reached
20:40 hmmmm        so it's getting close
20:40 hmmmm        does anybody have anything to say before 0.4.8 gets released?
20:41 sapier       1000 and async :-)
20:41 hmmmm        any concerns or valid reasons why we should delay it further
20:41 sapier       or at least 1000
20:41 sapier       oh 1000 is in
20:51 PilzAdam     hmmmm, we cant release without thexyz
20:51 sapier       do we need a windows build?
20:51 PilzAdam     of course we do
20:52 sapier       mingw directx?
20:52 sfan5        do we need a windows build with directx?
20:52 PilzAdam     MSVC produces better results
20:52 sfan5        it does?
20:52 sapier       what is better in msvc?
20:52 PilzAdam     http://dev.minetest.net/Releasing_Minetest says so
20:52 hmmmm        it wouldn't surprise me
20:52 hmmmm        PilzAdam, thexyz is gone for the night?
20:53 PilzAdam     hmmmm, well, its kinda late in russia already, isnt it?
20:53 hmmmm        i suppose so
20:53 sapier       wouldn't be surprised msvc is better than that hybrid mingw version buildbot creates
20:53 * sfan5      has a windows pc with visual studio 2012
20:53 hmmmm        well it doesn't have to be ALL at once..
20:53 sapier       msvc version is broken in terms of localization
20:53 PilzAdam     hmmmm, "Be sure that the windows build works before continue to do anything!"
20:53 sapier       my fix does just cure worst effects
20:54 sapier       ok I'm gonna build mingw as well as msvc now
20:54 PilzAdam     we cant actually release without a win build
20:54 hmmmm        :/
20:54 sfan5        if someone can gimme irrlicht with directx its good to go
20:54 PilzAdam     sapier, we still need thexyz, he hosts the builds
20:54 sapier       yes but we can prepare and do the checks now
20:54 hmmmm        well I have work tomorrow, soo
20:54 PilzAdam     I personally would feel more comfortable with a bit more testing, though
20:55 hmmmm        well sure, okay
20:55 hmmmm        change the topic to code freeze
20:55 PilzAdam     "code freeze"?
20:55 hmmmm        yup
20:56 thexyz       yes?
20:56 hmmmm        upstream stays the same as it is right now until it gets released
20:56 PilzAdam     so no bugfixes?
20:56 hmmmm        we had like a week and longer of bugfixes
20:56 hmmmm        if they're really bad then sure
20:56 hmmmm        hey thexyz wanna do a msvc build
20:56 thexyz       uh, not now, sorry
20:56 hmmmm        ;/
20:57 hmmmm        pilzadam:  at this point I am pretty sure all of the most pressing issues are solved
20:57 hmmmm        if they're really critical bugs, like bugs with what we just fixed, then sure yeah
20:58 hmmmm        I suppose we can call what we have right now in upstream a release candidate
20:58 PilzAdam     thats what makes me feel uncomfortable, we _just_ fixed major bugs
20:58 hmmmm        so we'll let the dust settle
20:59 hmmmm        alright, 0.4.8 delayed again :/
21:00 PilzAdam     I would have time on tuesday from 16:00 to 20:00 UTC
21:00 PilzAdam     or do we wait until next weekend?
21:02 sapier       windows build has a lot of redefine warnings
21:02 hmmmm        I work this week until 23 UTC
21:03 sapier       guess this should be addressed for 0.4.9
21:03 hmmmm        so really I'm only free from 23:00 to 6:30 UTC
21:07 PilzAdam     we need to update the changelog in the dev wiki and pull in latest translations
21:07 sapier       current mingw32 native libs doesn't build but we knew that before
21:08 sapier       and msvc doesn't build either
21:11 hmmmm        that's great
21:11 hmmmm        so what's the problem now
21:11 sapier       no idea I don't see any issue in there
21:15 sapier       ok now it's really strange
21:16 sapier       msbuild minetest.sln /p:Configuration=Release /t:Clean,Build runs fine if run twice
21:16 sapier       but it should cleanup prior any run
21:16 sapier       so whats difference?
21:21 Exio4        can't https://github.com/minetest/minetest/issues/1009 and https://github.com/minetest/minetest/issues/950 get closed atm?
21:22 VanessaE     about #950 ... isn't that also indicative of a mapgen bug?
21:22 PilzAdam     Exio4, done
21:22 VanessaE     I've seen this countless times
21:23 hmmmm        it'
21:23 hmmmm        it's not
21:23 VanessaE     ok
21:24 hmmmm        things coming out messy like that is a natural consequence of procedural generation if you want anything to be interesting at all
21:24 VanessaE     well sure, but huge saves full of suspended-in-the-air lava layers?  is that also normal? :)
21:25 VanessaE     caves*
21:25 hmmmm        same reason with spines, floating blocks, and lava in water - if you really hate it that much then why don't you fix it yourself
21:25 hmmmm        see how easy that is :p
21:25 hmmmm        vanessae, that's what happens when liquid doesn't flow
21:26 VanessaE     fair enough
21:26 VanessaE     it just struck me as odd that the lava wasn't already being placed in the empty spaces is all.
21:26 VanessaE     but whatever
21:26 VanessaE     if making liquid reflow solves it, so be it :
21:26 VanessaE     :P
21:26 hmmmm        the thing in that picture is two 'large caves' being generated into eachother
21:27 Exio4        also, what's up with weather stuff?
21:28 VanessaE     hmmmm: ah, well I'm thinking of a different screenshot actually
21:28 VanessaE     (which, of course, I've since deleted)
21:36 BlockMen     sapier, i just pulled lastest master and it builds fine for me (msvc 2010)
21:38 VanessaE     hmmmm:  I meant stuff more like this:  http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2011172013%20-%2004%3a36%3a51%20PM.png
21:38 VanessaE     surely those partial meshes can't be a liquid reflow bug.
21:39 sapier       interesting ... didn't build for me right after cmake but did build again on second try MSVC2012
21:39 hmmmm        probably some sort of cache thing
21:39 PilzAdam     VanessaE, flowing liquids dont have a face at the bottom
21:40 PilzAdam     so if they "hang" in the air then the result looks like your image
21:40 BlockMen     sapier, what error you get?
21:40 BlockMen     *in first try
21:41 sapier       unresolved symbol init_gettext
21:42 BlockMen     thats caused by missing gettext.cpp in minetest project folder. it seems not be included there when creating project file with cmake
21:43 sapier       why?
21:44 BlockMen     idk
21:44 sapier       and why is it there on second run
21:45 VanessaE     PilzAdam: it's hard to tell from the screenshot, but some of those suspended, partial meshes are lava sources.
21:46 VanessaE     the darker ones (don't mind that, it's some sort of lighting or shaders glitch)
21:46 PilzAdam     VanessaE, change the visuals of flowing lava to "unknown node", and see if it fits my explenation
21:47 VanessaE     PilzAdam: lava *sources*.
21:47 VanessaE     might be easier if you join the survival server and teleport to those coords and have a look around
21:52 VanessaE     I can't exactly show the movement of the flowing vs. sources in a screenshot :P
22:00 PilzAdam     VanessaE, I have to see all the flowing nodes to see if my guess is wrong or right, so it would help if you change the visuals of it
22:01 VanessaE     it would take less time and disturb fewer players if you just sign in, teleport, and look around :P
22:01 VanessaE     I don't even care if anyone fixes it.  I just want someone who "matters" to SEE it.
22:05 RealBadAngel hi, ive pulled reworked shaders: https://github.com/minetest/minetest/pull/1013
22:30 Tesseract    sapier: Do you think that script_error/ModApiBase::scriptError should check that the value at the top of the stack is a string?
22:31 sapier       yes as a null pointer is returned if it isn't
22:33 Tesseract    Well it's intended to only be called if you have just run a failed pcall and have a traceback or other error on the top of the stack.
22:33 sapier       obviously it isn't
22:34 sapier       according to sokomines researches if you define a entity with nil collisionbox you get a instant crash
22:36 Tesseract    This is definitely caused by this?
22:37 Tesseract    The crash caused by this piece of code that is.
22:38 Tesseract    I beleive lua_pcall always places a string on the stack, even in the case of LUA_ERRERR or LUA_ERRMEM.
22:40 PilzAdam     RealBadAngel, why do you remove all shaders?
22:41 RealBadAngel because now there are no multiple shaders for a material
22:41 RealBadAngel only solids, liquids and alpha
22:41 PilzAdam     but why do you remove them then?
22:42 RealBadAngel because i renamed test* to solids_shader etc
22:42 PilzAdam     no, you didnt rename, you just removed them
22:42 PilzAdam     https://github.com/RealBadAngel/minetest/tree/48a9a95cbaf5b3f6f02cbe0d90dcbbe6979e2ec6/client
22:42 VanessaE     um, RBA, there are no glsl files in this patch?
22:42 RealBadAngel there are
22:43 RealBadAngel yes i did removed old files
22:43 RealBadAngel theyre no longer needed
22:43 PilzAdam     but you didnt add new ones
22:43 RealBadAngel ?
22:43 PilzAdam     you completly removed client/shaders/, see the link I posted
22:43 VanessaE     vanessa@rainbird:~/Minetest-related/minetest_client$ ls client/
22:43 VanessaE     serverlist
22:44 RealBadAngel shit :) i dont know how that could happen
22:44 RealBadAngel gimme a second
22:44 PilzAdam     btw, its nice that you use branches now
22:48 VanessaE     "liquids shader"?
22:48 VanessaE     please tell me this is not your experimental rippling water thing
22:49 RealBadAngel it is not
22:49 VanessaE     ok
22:49 RealBadAngel check the pull now, it contains missing files now
22:49 VanessaE     trying.
22:50 RealBadAngel i have absolutely no idea why smartgit havent pulled them before
22:51 RealBadAngel propably thx to playing with branches
22:51 RealBadAngel anyway its fixed now
22:52 VanessaE     um, why is flowing water opaque?  and why does lava source have a darker top side than flowing lava?
22:53 VanessaE     come to think of it, water source is opaque too.
22:54 RealBadAngel can you make a screenshot with water?
22:55 VanessaE     http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2011172013%20-%2005%3a55%3a13%20PM.png
22:55 RealBadAngel for me it looks fine: http://i.imgur.com/klVx8eO.png
22:56 VanessaE     for the channel log, RBA's screenshot is from my Survival server.
22:57 RealBadAngel can you check the water with Bumped texture pack?
22:57 VanessaE     no.
22:57 VanessaE     make it work properly without bumps first.
22:57 RealBadAngel lemme check the HDX then
22:58 VanessaE     the dark lava sources and opaque water might be a long-standing problem, but with shaders on it should "just work"
22:58 RealBadAngel lemme check why that happens
22:59 VanessaE     *checks without shaders*
22:59 VanessaE     looks fine without shaders, though lava's kinda dim by comparison
23:00 RealBadAngel i think i know why
23:00 RealBadAngel your water is simply opaque
23:00 VanessaE     (and now lava sources are slightly brighter than flowing)
23:00 VanessaE     yeh, so?
23:00 VanessaE     it works fine without shaders
23:00 RealBadAngel it doesnt provide alpha channel at all
23:00 VanessaE     so?
23:00 RealBadAngel so its wrong
23:00 VanessaE     provide it yourself in the code
23:00 RealBadAngel me? in irrlicht?
23:01 RealBadAngel sorry, just save it with alpha channel
23:01 VanessaE     um
23:01 VanessaE     it's a jpeg
23:01 VanessaE     jpegs don't support alpha.
23:01 VanessaE     and again *it works fine without shaders*
23:01 RealBadAngel save it in format that supports it then
23:02 RealBadAngel even default water is a png
23:03 VanessaE     then why do these textures work fine without shaders?
23:03 RealBadAngel because alpha here is faked
23:03 VanessaE     then fake it in the shader too?
23:03 RealBadAngel but shaders need actual real value
23:04 Tesseract    I don't like how your shaders break use_texture_alpha and similar things.
23:04 RealBadAngel and theres no way to say whether designer added alpha or not
23:04 VanessaE     RealBadAngel: exactly my point.  SO FAKE IT.
23:04 RealBadAngel thats the texture2D sampler thing
23:04 RealBadAngel not my idea
23:05 NakedFury    dont fake it
23:05 RealBadAngel repeat, theres no way to say if sampler holds the data or not
23:06 RealBadAngel so i cannot check if user provided alpha or not. only logical way is to assume it has to be provided
23:06 VanessaE     um..  isn't water supposed always have a constant alpha anyway?
23:06 RealBadAngel alpha channel is supposed to define alpha
23:07 RealBadAngel not some friggin tricks
23:07 VanessaE     not with water
23:07 PilzAdam     got an email from launchpad about a failed build: https://launchpad.net/~minetestdevs/+archive/daily-builds/+recipebuild/585830
23:07 RealBadAngel with everything, including water
23:08 VanessaE     RealBadAngel: that represents a fundamental change that may break more than just some textures then
23:08 RealBadAngel huh?
23:08 RealBadAngel how could that happen?
23:08 VanessaE     because the server also has a concept of how opaque the water is supposed to be
23:09 VanessaE     the WATER_ALPHA parameter in a node def.
23:09 RealBadAngel hrmmm
23:09 VanessaE     or rather the variable sometimes used therein
23:09 RealBadAngel ok, i will add that trick back
23:10 VanessaE     but wait
23:10 VanessaE     let me see how this looks with PNG's + alpha.
23:10 RealBadAngel for sake of compability, but its just wrong
23:11 RealBadAngel ofc it will work perfectly
23:11 RealBadAngel it does with default textures and both  my  texture packs
23:16 VanessaE     ok, that fixes the water transparency.
23:16 VanessaE     but lava is still screwed up.
23:16 VanessaE     see the previous screenshit
23:16 VanessaE     er screenshot*
23:16 VanessaE     http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2011172013%20-%2005%3a55%3a13%20PM.png
23:16 VanessaE     the darker lava is sources.
23:17 VanessaE     without shaders, at least when I looked earlier, they were almost the same brightness (in fact sources were slightly brighter then)
23:17 VanessaE     THIS is ugly as shit
23:21 VanessaE     RealBadAngel: I can accept the need for an alpha channel in the water to make it look right, but that lava's gotta be fixed, at least in the shader (I expect out-of-shader changes would be needed as well)
23:21 RealBadAngel propably the very same issue
23:22 RealBadAngel alpha when not provided is just a random value in uniform sampler
23:22 RealBadAngel im checkin it now
23:22 VanessaE     the lava images have alpha.
23:22 RealBadAngel jpgs?
23:22 VanessaE     at least now they do.
23:22 VanessaE     no, I changed all lava and water to PNG + alpha.
23:22 VanessaE     let me push to HDX
23:23 RealBadAngel ok
23:26 VanessaE     ok, go ahead and pull.
23:26 VanessaE     bbl, dinner
23:47 VanessaE     back