Minetest logo

IRC log for #minetest-dev, 2014-06-18

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:00 asl i talk about some memory leak in minetest a few days/weeks ago,  after running minetest for awhile, it slowly hogging the ram and before you know it, it use up all the memory and freeze the whole os, exiting to the menu doesn't free the memory, only closing minetest then the memory would be free
00:01 VanessaE huh.
00:01 asl it slowly hog more and more memory as more block/chunk of the map get loaded and that's is the only lead i got
00:01 VanessaE [06-17 20:00] <VanessaE> Jordach: also, 6.133 GB with HDX 256 at Creative's spawn after pitching and panning my view around and letting every last bit of map around me load up
00:01 VanessaE [06-17 20:01] <VanessaE> no preload, all settings enabled except for parallax, have not opened the inventory yet.
00:12 Eater4 joined #minetest-dev
00:42 VanessaE Ok after enabling some swap, Minetest uses 14.342 GB virtual, and somewhere in the 12 GB resident range, with HDX 512, just standing here looking around at the spawn on my Creative server.
00:42 VanessaE THAT'S FUCKING INSANE.
00:42 VanessaE the G*d damn pack is only ~420 MB.
00:42 VanessaE and I am NOT using preload visuals
00:43 VanessaE so there's no way in hell it should be using this much, this soon.
00:43 VanessaE all I'm doing is standing at the spawn, looking around.
00:43 VanessaE FPS:  3-4.
00:45 VanessaE and that's with the view range turned all the way down to 5m.
00:45 VanessaE what is the problem here?
00:50 VanessaE hello?
00:51 VanessaE and yes, I'm ranting
00:56 Sokomine whatever the problem is, hope it can be fixed. less good hardware may profit from such fixes as well
01:04 VanessaE well if Jordach can do the same server with less than half the usage, and roughly equivalent CPU and graphics power, but under Windows, then surely it can
01:04 VanessaE but does anyone care?
01:04 VanessaE or is everyone obsessed with the server?
01:04 VanessaE (except RealBadAngel, whose focus is on shaders)
01:32 Megaf joined #minetest-dev
02:36 Eater4 joined #minetest-dev
02:38 VanessaE if I turn off aniso/mipmap, it barely even touches the RAM usage (not enough to even measure).
02:38 VanessaE (and gives me back a ton of framerate, probably because of GPU overload).
02:38 VanessaE if I turn off normalmaps, that gains back about 2.5GB, but still leaves me using over 9.5 GB of RAM.
02:39 VanessaE I simply can't use any less than that, and that's still almost twice the usage Jordach was getting with windows, and he was using normalmaps.
02:39 VanessaE someone has to explain this.
02:45 Exio4 i've been trying, i seem to end with the resource usage as a lot vs the win build (using latest's sfan5 build at the moment, cb3b42e) against latest git
02:45 Exio4 with same settings it ended being as much as 375% (3.75x :P)
02:48 VanessaE ^^^ that is, his experience mirrors mine, only worse.
02:57 VanessaE https://github.com/minetest/minetest/issues/1386
05:25 Miner_48er joined #minetest-dev
05:46 Hunterz joined #minetest-dev
05:53 grrk-bzzt joined #minetest-dev
06:12 LemonLake joined #minetest-dev
06:17 werwerwer_ joined #minetest-dev
06:23 werwerwer_ joined #minetest-dev
06:23 NakedFury joined #minetest-dev
07:09 OldCoder joined #minetest-dev
07:35 CraigyDavi_ joined #minetest-dev
08:13 CraigyDavi joined #minetest-dev
08:15 CraigyDavi__ joined #minetest-dev
08:37 kaeza joined #minetest-dev
08:38 CraigyDavi_ joined #minetest-dev
09:24 kaeza joined #minetest-dev
09:55 restcoser joined #minetest-dev
10:03 Jordach joined #minetest-dev
10:43 Eater4 joined #minetest-dev
10:46 proller joined #minetest-dev
11:01 ImQ009 joined #minetest-dev
11:26 CraigyDavi joined #minetest-dev
11:28 CraigyDavi__ joined #minetest-dev
11:29 Megaf joined #minetest-dev
11:39 CraigyDavi_ joined #minetest-dev
11:43 vifino joined #minetest-dev
11:49 alexxss joined #minetest-dev
12:03 khonkhortisan joined #minetest-dev
12:04 Megaf_ joined #minetest-dev
12:08 EvergreenTree joined #minetest-dev
12:52 hmmmm joined #minetest-dev
12:54 RealBadAngel hi hmmmm, whats up?
12:55 hmmmm skipping work today
12:56 RealBadAngel i do have 2 weeks off
12:56 hmmmm well good for you
12:56 hmmmm hopefully I'll have more time soon
12:57 RealBadAngel i will also have a bit more time later, i will work as a coder soon
12:57 RealBadAngel and propably will work at home
12:58 hmmmm yeah same here, work at home seems great
12:58 hmmmm today i have the final interview for this new job
12:58 hmmmm 3 hours long
12:59 hmmmm all with cell phone voice quality
13:01 RealBadAngel and you have this job?
13:01 hmmmm i will.  i'm switching over
13:02 RealBadAngel cool
13:03 RealBadAngel about my 2 weeks off, im gonna spend them on coding
13:03 RealBadAngel i already started with some small things
13:03 hmmmm lol, i know exactly what i need to get done
13:04 hmmmm which involves more trial and error
13:06 RealBadAngel after im done with several small features, im gonna focus on two things
13:07 RealBadAngel rendering buffers and mesh drawtype
13:08 werwerwer joined #minetest-dev
13:15 RealBadAngel -
13:15 RealBadAngel +-+--
13:15 RealBadAngel +-
13:16 RealBadAngel ooops
13:17 proller joined #minetest-dev
13:24 kaeza joined #minetest-dev
13:37 RealBadAngel hmmmm, any plans on pushin mgv7 forward?
13:38 RealBadAngel also, i was working lately on tooltips for some formspec elements and now i agree with ShadowNinja
13:39 RealBadAngel we definitely have to drop that shit, it leads to nowhere
13:39 RealBadAngel adding new features is pain in the ass, keeping compability is almost impossible
13:40 RealBadAngel we have to use just tables
13:40 RealBadAngel sooner we drop strings the better
13:42 diemartin joined #minetest-dev
13:50 PilzAdam joined #minetest-dev
13:52 Hunterz joined #minetest-dev
14:04 hmmmm yes, i would like to move mgv7 forward
14:12 Krock joined #minetest-dev
14:16 NakedFury joined #minetest-dev
14:17 ShadowNinja So I rewrote the server list in Python: http://sprunge.us/QEHP?python
14:18 ShadowNinja It isn't feature-complete just yet, but it has some advantages, such as lots more checks on the server data.
14:23 alexxss joined #minetest-dev
14:25 LemonLake joined #minetest-dev
14:36 RealBadAngel ShadowNinja, so, are we moving to tables instead of strings?
14:36 RealBadAngel im ready to help with that
14:37 diemartin joined #minetest-dev
14:37 sfan5 merging in 10 mins: https://github.com/minetest/minetest/pull/1375
14:46 Krock joined #minetest-dev
14:50 ShadowNinja RealBadAngel: Yes, do you think you can code that?  You'll have to use JSON or core.serialize for sending over the network and the client should still support the old format for a release or two.
14:51 ShadowNinja I know Lua, but not network or client.
14:52 PenguinDad joined #minetest-dev
14:59 Eater4 joined #minetest-dev
15:09 crazyR joined #minetest-dev
15:13 GhostDoge joined #minetest-dev
15:18 pitriss joined #minetest-dev
15:20 vifino_ joined #minetest-dev
15:20 RealBadAngel ShadowNinja, i think serializing/deserializing could be better, readability is not the point here
15:21 RealBadAngel about old format... im not sure
15:22 RealBadAngel i think we should rather release 0.4.10 soon as last version with formspecs and go for new stuff
15:23 RealBadAngel a) we havent released stable for a long time already, theres lotsa features onboard already
15:23 Piggybear87 joined #minetest-dev
15:23 RealBadAngel b) it will give us some time till next release to work on new system
15:24 kaeza joined #minetest-dev
15:24 RealBadAngel my proposal is, lets gather all the possible fixes we have around
15:24 RealBadAngel till the end of the week, lets say
15:25 RealBadAngel and then we can start pushing new code
15:25 RealBadAngel i do have at least 2 weeks off just for coding
15:26 RealBadAngel so i will be definitely hyper active
15:31 kaeza joined #minetest-dev
15:31 VanessaE I don't recommend dropping support for formspecs for a very LONG time, if ever
15:31 VanessaE though making them as deprecated and basically never talking about them, is a good idea
15:31 VanessaE marking*
15:34 sfan5 I think it's ok to make a new release.. http://sprunge.us/XfiT
15:34 RealBadAngel as i said, next weekend
15:35 RealBadAngel if we do have some fixes, push them and go
15:35 sfan5 how about next friday?
15:35 sfan5 I won't be having much time next saturday
15:35 Hunterz joined #minetest-dev
15:44 kaeza joined #minetest-dev
15:48 Krock yay! a new version!
15:54 RealBadAngel sfan5, you mean 27.06?
15:54 sfan5 yes
15:55 RealBadAngel its ok with me
15:55 RealBadAngel enough time to gather forsaken fixes and work on needed ones
15:56 sapier joined #minetest-dev
15:56 VanessaE how about using REAL irrlicht particles instead of minetest entities for particle spawners?  too late for that? :)
15:56 RealBadAngel i think not for this release
15:56 sfan5 VanessaE: irrlicht particles can't have collision IIRC
15:56 VanessaE didn't think so.
15:56 VanessaE sfan5: so don't use them when collisions are needed?
15:57 RealBadAngel lets focus on bugfixes right now
15:57 sapier entities for particles? hope you're not talking about increasing those :-)
15:57 RealBadAngel we do have a few issues open
15:57 sapier numbers
15:57 * VanessaE shrugs and gives up.
15:58 RealBadAngel VanessaE, dont give up, but this is like new feature, not a bugfix
15:58 kaeza joined #minetest-dev
15:58 VanessaE RealBadAngel: it would be a bugfix, imho, because particle spawners are unusable in their current state.
15:58 RealBadAngel same as formspecs
15:58 VanessaE no, I mean literally
15:59 sapier vanessae you forgot about png being compressed images while textures arent
15:59 VanessaE formspecs at least work
15:59 RealBadAngel to some point
15:59 VanessaE particle spawners cause a horrible performance hit when they're used
15:59 VanessaE and this should not happen
15:59 VanessaE sapier: I did not forget.
15:59 sapier meaning each of your 512hdx textures is at leas 1MB ram
15:59 RealBadAngel what do you think, how much it takes to compare two formspec strings?
16:00 VanessaE sapier: but that doesn't explain why Linux uses 2 to 3 times as much RAM as Windows under the same conditions, settings, and hardware capabilities.
16:00 sapier that's true
16:00 VanessaE sapier: ok, 3400 * 1 MB.  big deal.  that's only 3.4 GB.
16:02 VanessaE excuse me, 3279 * 1 MB.
16:02 sapier I'm against a fixed schedule for formspec because there's noone who wants to do that work right now, everyone demands it to be replaced but once someone is asked to do it that one usually becomes very very silent
16:02 RealBadAngel sapier, me and SN
16:02 VanessaE even if you include normalmaps, that's only 1855 more, or 5134.
16:03 RealBadAngel we want to work on that
16:03 sapier don't get me wrong I'm not against doing it in a better way but it has to provide all current features and compatibility code
16:03 RealBadAngel we will see
16:03 VanessaE sapier: if your 1MB-per-texture guess is correct, then that's 5134 MB for every last texture in existence for the HDX 512px pack and its normalmaps, and that would be a perfectly reasonable amount.
16:03 sapier and I'm against doing a release prior merging android port too it's 95% done
16:04 sapier 1mb per texture isn't a gues but a exact calculation
16:04 RealBadAngel we cant delay releases forever on the other hand
16:04 sapier 512px * 512px * 4 bytes
16:04 RealBadAngel theres too much code in already
16:04 VanessaE sapier: ok, exact calculation.  that makes it even worse.
16:05 sapier it's been waiting for 6 months now why do you wanna delay android support till christmas
16:05 RealBadAngel modders do need stable releases
16:05 VanessaE sapier: that means I'm losing no less than 7-8 GB of RAM to some random leak.
16:05 sapier especially as it's about 2 weeks prior merge
16:05 RealBadAngel we can wait 2 weeks
16:05 RealBadAngel its just about we have chosen
16:06 RealBadAngel 27.06 or lets say end of the month
16:06 RealBadAngel it is enough for you?
16:06 sapier modders do need a stable version for about 4-5 months now ... our build schedule is plain crap
16:06 sapier depends on how active ppl are at reviewing the remaining changes
16:07 RealBadAngel i do have 2 weeks off
16:07 RealBadAngel started yesterday
16:07 Jordach why should we remove formspecs?
16:07 RealBadAngel because theyre dead end
16:07 Jordach not good enough answer
16:07 RealBadAngel overengineered
16:07 sapier jordach in mid to long term they have to be replaced by a more generic mechanism but my personal opinion is this will take at least 2-3 releases
16:08 RealBadAngel more features we add its gettin worse
16:08 sapier come one rba just because they don't fit your current requirements ;-)
16:08 RealBadAngel no
16:08 RealBadAngel i can see we are standing on the edge
16:08 sapier any generic implementation will be way more complicated then formspec
16:08 RealBadAngel tables....
16:08 sapier most likely we need a gui designer for it
16:09 RealBadAngel what we do need is just to transfer lua tables to c++ side
16:09 sapier lol
16:09 RealBadAngel and put the elements into existing structures
16:09 RealBadAngel without strings parser
16:09 sapier sorry but believing c++ is the answer is almost always wrong
16:10 RealBadAngel no
16:10 RealBadAngel i said on c++ side everythin is ok
16:10 sapier and next time irrlicht changes all of our gui api is broken
16:10 RealBadAngel we need just flexibility tables can provide
16:11 RealBadAngel changing an element, adding new one
16:11 Jordach what i'm thinking is a master table, and then other tables inside said table, containing the data, such as slots[1] = "itemstring <number of item>"
16:11 RealBadAngel piece of cake while using tables
16:11 Garmine joined #minetest-dev
16:12 RealBadAngel think of a table like of a folder
16:12 sapier I don't believe that's gonna be siginificant more readable but way more hard to parse
16:12 RealBadAngel put into root size, backgrounds, etc
16:12 RealBadAngel and then in parent elements
16:12 Jordach sapier, considering i get headaches from current formspecs - yes, that'd probably help in the long run
16:13 sapier yes but in long run we shouldn't invent another proprietary format as we already did with formspec but use some more common one
16:13 RealBadAngel cmon, tables are what we do have
16:13 RealBadAngel we dont have to invent anything
16:14 RealBadAngel just iterate over the table and send it to c++ side
16:15 RealBadAngel adding new features will be nothing more like adding new field
16:15 sapier sorry I don't believe that's gonna provide a real benefit especially on guessing the amount of work required, but you're free to proove me wrong
16:16 RealBadAngel that will give us flexibility we dont have
16:17 RealBadAngel generating complex formspecs is lookin as a bad joke
16:17 RealBadAngel see UI's code
16:17 sapier well usually flexibility adds complexity too
16:17 RealBadAngel complexity we have ^2 alreadty
16:17 sapier formspecs benefit is it's quite easy to get started
16:17 RealBadAngel without ANY flexibility
16:18 RealBadAngel for simple forms, yes
16:18 sapier you don't need to do complex things while a table thingy might be complex for simple things too
16:18 RealBadAngel but it evolved into something worth monthy python's award of the year
16:19 RealBadAngel in current state formspecs are a bad joke just
16:19 RealBadAngel they were good for signs
16:19 RealBadAngel and for that we can leave them in the engine
16:20 RealBadAngel but UI screams for something usable and flexible
16:20 sapier noone stops you from implementing a better suggestion rba ... but don't you have enough work with shaders? ;-)
16:20 RealBadAngel code that builds single string out of hundreds of pieces? oh cmon
16:21 RealBadAngel i believe even UI can hit thousand rite now
16:21 RealBadAngel now try to debug such string
16:22 RealBadAngel its DEAD END
16:22 sapier noone is questionong that formspecs should be replaced but noone is there to do the work too
16:22 rubenwardy joined #minetest-dev
16:24 RealBadAngel sn said that about two weeks ago
16:24 RealBadAngel now after my attempt to add tooltips joined him
16:25 sapier ok is there anyone who wants to rewrite formspecs as well as main menu?
16:26 RealBadAngel are you blind or what? :P
16:26 RealBadAngel me and sn
16:27 RealBadAngel i can see its not that hard to do
16:27 sapier then start doing it ... but believe me 2 weeks is far from enough
16:27 RealBadAngel huh
16:27 RealBadAngel try me
16:28 sapier as I said I'd be glad if you proof me to be wrong
16:28 RealBadAngel before month ends you will have that
16:29 RealBadAngel and be ready to buy crate of beer for both of us :P
16:30 sapier don't forget about doc and don't break any mod
16:31 sapier and of course, it's up to you to persuade modders your's is really better ;-P
16:32 sapier still RealBadAngel I'd prefere if you did things really providing benefit like doing the shaders
16:32 sapier formspec is ugly but not broken
16:33 sapier and debugging a lua table isn't better then debugging a string
16:37 sapier comments? https://github.com/minetest/minetest/pull/1373 I'd like to merge it
16:37 Eater4 sapier: I do not understand how to use (your) faction mod
16:39 VanessaE sapier: any regression potential?
16:39 sapier none I know about
16:40 sapier it's just an addon
16:40 sapier so where nothing happened before something will happen ... of course if someone relied on nothing to happen (that'd be silly) there might be issues
16:40 VanessaE you mean like how the old signs formspecs were? ;)
16:41 * Krock likes the new dropdown box
16:42 sapier it's not new ... just the behaviour
16:43 Krock I see
16:44 zat joined #minetest-dev
16:44 sapier by now you had to read it's content on pressing some button ... you can do this in future too but now you can react immediatly once it's changed
16:45 kaeza joined #minetest-dev
16:45 Krock list boxes already have this feature, so it's good to have it there too
16:47 jin_xi joined #minetest-dev
16:48 sapier did anyone test it by now?
16:50 Calinou joined #minetest-dev
16:54 Krock just merge it - the revert function exists when something went wrong
16:54 kaeza ...
16:55 sapier well reverting things is last thing we wanna do
17:12 sapier ok after doing some additional testing I'm gonna merge 1373 now
17:14 sapier could someone please test this changes https://github.com/minetest/minetest/pull/1359 they provide up to 20% more fps
17:14 RealBadAngel can you squash them?
17:15 RealBadAngel git rebase - i HEAD~8
17:15 Megaf sapier: Im testing #1359 when I get home
17:16 sapier pushing #1373 now
17:17 Megaf sapier: VanessaE: in unified_inventory, It wont be possible to click on an item if the text from chat/server is at the same "height" of the item row.
17:17 VanessaE Megaf: not limited to Unified Inventory.
17:17 Calinou no disable_jump? lua_api.txt talks about it, but it has no effect: I can jump on a nodebox that uses it, even if nothing is below and it and besides it (to be sure there are no collision errors)
17:17 VanessaE any formspec is covered by the chat area.
17:18 VanessaE Megaf: it also affects pipeworks' mese sorting tubes' formspecs (the white button is inaccessible while chat is enabled)
17:18 Krock Agh I really must buy a new PC in the next days... I'd like to test those commits but the compiling time of 31min ppisses meoff
17:18 RealBadAngel build it incrementally
17:18 sapier or use make -j<number of cores+1>
17:18 Calinou RealBadAngel, isn't this the default?
17:19 Calinou when you modify only one file, only this file and a few more ones are recompiled
17:19 sapier depends on what is modified
17:19 Krock sapier, make -j 2 :P
17:19 RealBadAngel it is, but maybe he continuosly is building mc in new folder
17:19 RealBadAngel *mt
17:19 sapier we've got a insane header coupling within mt
17:20 RealBadAngel sapier, btw, could you fix tooltips rectangles resizing while in mainmenu?
17:21 sapier there's no simple way to do this
17:21 RealBadAngel pro tip, it works correctly while in game
17:21 sapier I already know why it happens
17:21 RealBadAngel without that, my tooltips commit is useless
17:22 sapier mainmenu is a fixed size formspec thus viewport is set to 800x600
17:22 sapier no matter how big window is
17:22 Megaf [14:18:09] <Krock> Agh I really must buy a new PC in the next days... I'd like to test those commits but the compiling time of 31min ppisses meoff
17:22 VanessaE sapier: but the bounding boxes go way beyond 800px wide.
17:22 Megaf It takes 3 minutes to compile on my ARM board Krock
17:22 VanessaE like, 3000+ pixels.
17:23 sapier calculation error?
17:23 Megaf Krock: try -DBUILD_REDIS=0 -DBUILD_LEVELDB=0 -DBUILD_SERVER=0 options
17:23 VanessaE sapier:  http://digitalaudioconcepts.com/vanessa/h​obbies/minetest/screenshots/Screenshot%20​-%2006152014%20-%2008%3a54%3a20%20PM.png
17:23 Krock Megaf, leveldb doesn't work, I don't have redis libs, server is in client too
17:24 Megaf yep, so you can disable support for them
17:24 VanessaE sapier: I can reproduce ^^^^^ that with as much as 3000 pixels of separation between the pointer and the NOT-focused Minetest window :)
17:24 sapier that's something completely different
17:24 sapier that's checkboxes having fixed lenght
17:24 Megaf [14:18:01] <VanessaE> Megaf: it also affects pipeworks' mese sorting tubes' formspecs (the white button is inaccessible while chat is enabled)
17:24 Megaf That's really bad
17:25 sapier fix for this should be easy
17:25 Megaf Can't chat be sent a layer before that?
17:25 sapier I'll push the fix direct
17:26 Megaf Please, and thanks.
17:27 VanessaE Megaf: he's talking about the main menu glitch, not the chat thing.
17:27 Megaf oh
17:27 Megaf too bad
17:27 Megaf =/
17:27 Megaf And I don't get that menu glitch here
17:27 kaeza <Calinou> when you modify only one file, only this file and a few more ones are recompiled
17:27 kaeza because C++
17:27 sapier you don't have tooltips so you don't see it
17:27 VanessaE Megaf: you don't have the patch in place that was used to gen... yeah
17:28 sapier megaf I once did a fix for this but someone had a better way to do it ... sadly I don't remember who
17:28 VanessaE sapier: "better" fixes, imho, are useless if they aren't committed :P
17:29 Megaf indeed
17:29 VanessaE (so whosever better way it was botched it by blocking you and not following through with their idea)
17:30 sapier well I thought it was and realized months later it wasnt
17:30 Megaf A fix that fixes and dont cause regreations/performance loss is a good fix
17:30 RealBadAngel i can live with tooltips just on checkboxes box
17:30 RealBadAngel but resizing have to be done
17:38 RealBadAngel sapier? can you fix that mainmenu resizing?
17:39 sapier no
17:40 sapier but I can fix those bounding boxes causing the bug in vanessaE's screenshot
17:40 RealBadAngel bounding boxes are not a problem
17:40 RealBadAngel position of the rectangles is the real problem
17:41 sapier if you want your tooltips to use full screen size use porting::getWindowSize() to get the real screen size not the menu size
17:41 RealBadAngel tooltips rectangles doesnt match elements after resizing
17:41 VanessaE RealBadAngel: um, that screenshot was taken with the window at its default size...
17:41 sapier sorry I lost track what are we talking about now
17:42 RealBadAngel VanessaE is talking about hardcoded rectangles for checkboxes
17:42 RealBadAngel which is wrong but not critical
17:43 RealBadAngel im talkin bout that defined rectangle which is used to check whenever cursor points the element is broken while in main menu
17:43 RealBadAngel it works ingame
17:43 RealBadAngel so i cannot define them another way
17:43 RealBadAngel theyre not scaled to respect viewport changes
17:44 sapier I don't think so
17:44 VanessaE why would they scale if the main menu doesn't change size anyway?
17:44 sapier I'll have look at your code but I think you did something wrong
17:44 RealBadAngel they do work properly ingame
17:45 RealBadAngel whenever you rescale viewport in mainmenu theyre shifted
17:45 VanessaE so you meant the *origin* moves
17:45 RealBadAngel yes
17:45 VanessaE not that the size of the target bit changes
17:45 VanessaE target box*
17:46 RealBadAngel same code is used for all the elements no matter ingame or mainmenu
17:46 RealBadAngel in game all is ok
17:47 sapier why don't you just ask irrlicht what element the cursor is at? that's what the engine is for?
17:47 RealBadAngel ok, keep asking me to trash that formspec and related code
17:47 RealBadAngel and i will do so
17:48 sapier give me a few minutes to finish the bounding boxes then I'm gonna show you how to do the tooltips without own rects
17:48 sapier irrlicht already does this for us so we don't need to reimplement it
17:48 RealBadAngel we already did long time ago
17:49 sapier +if (tooltip_x + tooltip_width > (s32)screenSize.X)                                                                        +tooltip_x = (s32)screenSize.X - tooltip_width - 15;
17:49 RealBadAngel im asking you to make it work in mainmenu
17:49 sapier that's causing the issue as screenSize isn't really screensize there
17:50 sapier but that doesn't actually matter using own rects while irrlicht already has them is broken anyway
17:50 sapier and yes I know that's in there for some time
17:50 RealBadAngel since always...
17:51 sapier obviously the one implementing it didn't know she/he could make irrlicht do the work
18:05 RealBadAngel if so, its really let the engine do that
18:05 RealBadAngel *better
18:14 lemonlake joined #minetest-dev
18:14 lemonlake joined #minetest-dev
18:15 sapier can someone explain to me why I always have to find bugs at locations where noone expects bugs ... text width calculation is broken
18:15 lemonlake because you look in areas where nobody looks
18:15 RealBadAngel hmmm, broken?
18:16 RealBadAngel it worked for me correctly, it is takin into account \n 's
18:16 RealBadAngel and returnin length of the longest line
18:16 sapier according to irrlicht doc's it's supposed to calculate lengths in pixels a string would be rendered to
18:16 sapier but that's wrong
18:16 sapier it's way below that length
18:16 rubenwardy I have tested #1359 The view difference doubles for me when keeping the same fps. Overwise it goes from 23fps -> 29/30 fps.
18:16 RealBadAngel for tooltips it is working more than perfect
18:17 RealBadAngel it is height that is broken
18:17 RealBadAngel i had to manually count the lines
18:17 sapier hmm
18:17 RealBadAngel look at my commit
18:17 sapier what do you use? font->getDimension()
18:17 sapier ?
18:17 RealBadAngel getWidth and getHeight
18:18 RealBadAngel lemme check
18:18 sapier hmm you use the created element
18:21 RealBadAngel http://pastebin.com/CypR7kGF
18:22 sapier yes ... sadly I can't do it that way because I don't have a gui element by that time
18:24 RealBadAngel thats why i wasnt able to calculate proper width for checkbox
18:26 vifino joined #minetest-dev
18:40 RealBadAngel sapier, indeed irrlicht have tooltips
18:40 RealBadAngel but its not enough for u
18:40 RealBadAngel us
18:41 RealBadAngel we do have our own gui elements, like itemstacks, itemimagebuttons, imagebuttons etc
18:41 sapier hmm didn't even notice you didn't use the irrlicht tooltips
18:41 sapier I was talking about detecting where the mouse cursor is only
18:42 RealBadAngel its not my code
18:42 RealBadAngel i havent invented that, but i can see now why it was done in the first place
18:43 sapier as I understand why labels have fixed size ... the one implementing it most likely didn't want to look for the getDimension bug
18:43 RealBadAngel theres way around that
18:44 sapier what do you suggest?
18:44 RealBadAngel make invisible label of same text and check its width
18:44 sapier that's a ugly hack
18:44 RealBadAngel but will work for sure
18:44 sapier well mine is ugly too ... but at least way more performant ;)
18:45 sapier I'm just gonna multiply witdth by 1.5 ... still not correct but way better then fixed 300
18:45 RealBadAngel that could be generic function that will create label examine it, delete and return dimensions
18:45 sapier still it'd create a gui element
18:45 RealBadAngel wont work with different fonts
18:46 RealBadAngel vanessa will be bitching about it for sure ;)
18:46 RealBadAngel shes using larger font
18:46 sapier I'll not add something causing that much additional memory operations as quick n dirty hack
18:46 sapier I'll look for the original bug later but for not it's at least better then the current state
18:46 RealBadAngel lables are not time critical
18:47 sapier but memory operations cause heap fragmentation
18:47 Krock left #minetest-dev
18:47 sapier I know we already do to much but adding additional ones wont make it better ... and may even cause ghost texts
18:47 RealBadAngel not a big deal while in a menu
18:47 RealBadAngel you can clear whole stack while leaving
18:48 sapier fragmentation doesn't vanish once you enter game and it's in game formspec too
18:48 sapier no you can't clear stack
18:48 sapier heap
18:48 sapier stack is always cleared but not heap
18:59 sapier hmm no I don't like hacks ... that 1.5 is too ugly I'm gonna spend the time for a real fix
19:02 OldCoder joined #minetest-dev
19:02 nore joined #minetest-dev
19:03 sapier argh ...
19:03 sapier it's not the calculation being broken ...
19:03 sapier a checkBOX isn't just the text only but a box in front of it ..
19:17 sapier https://gist.github.com/sa​pier/b4b3de15c639f16af352 fixes labels as well as vertlabels too
19:17 sapier any comments?
19:30 LemonLake joined #minetest-dev
19:30 nore joined #minetest-dev
19:37 Fresh_me_ joined #minetest-dev
20:02 BlockMen joined #minetest-dev
20:16 ImQ009 joined #minetest-dev
20:30 proller joined #minetest-dev
20:35 RealBadAngel sapier, its ok, but its not what i asked for...
20:35 sapier it's not supposed to be your fix
20:36 sapier that's the fix for vanessaE's issue only once it's done I'm gonna have a look at your issue ... at least if you could explain to me whats wrong because I haven't really understood it by now
20:37 sapier I'm gonna push this now and look for your issue right after it
20:40 RealBadAngel ok
20:41 smoke_fumus joined #minetest-dev
20:41 RealBadAngel for tooltip you define rect which is checked later
20:41 sapier ok pushed now please explain to me what's wrong
20:43 RealBadAngel after resizing window in mainmenu, rectangles for tooltips do not match elements anymore
20:43 RealBadAngel they do work without problems ingame
20:44 sapier well first of all please fix your tooltip patch breaking all mods while I find out whats wrong
20:44 RealBadAngel lol
20:44 RealBadAngel show me such mod in the first place
20:44 sapier mobf
20:45 RealBadAngel what mobf has to do with imagebuttons?
20:45 sapier the menu does heavyly use checkboxes for settings and I'm extremely annoyed because I already have 3 different settings menus to match different apis
20:45 sapier no but checkboxes
20:45 sapier which you have broken too
20:45 Piggybear87 joined #minetest-dev
20:46 RealBadAngel so throw away code for anything but buttons
20:46 RealBadAngel because any rectangle is wrong
20:46 sapier you don't have to throw it away why don't you just ADD your code like everyone else??
20:46 RealBadAngel ok i will, but its not like 5 seconds
20:47 sapier ok I'm gonna do it myself if you think its a big issue
20:47 RealBadAngel and i have to think how the hell im supposed to guess what user gave as parameter
20:47 RealBadAngel that is about imagebuttons especially
20:48 sapier of course you can add this for exactly where you need it but thats selfish
20:49 RealBadAngel its not elements code broken
20:49 RealBadAngel regenerateGUI is broken for mainemenu
20:49 RealBadAngel all the elements and rectangles work ingame
20:50 RealBadAngel imagebuttons defined same way work perfectly ingame, but are broken in mainmenu
20:51 RealBadAngel try to display item in mainmenu, it will be broken too
20:51 Piggybear87 joined #minetest-dev
20:54 RealBadAngel apply my patch and check tooltip for "Bumpmapping"
20:55 RealBadAngel ive added this one
20:55 Miner_48er joined #minetest-dev
20:56 RealBadAngel well, you can try this even without my patch
20:57 RealBadAngel add tooltip to game buttons
20:57 RealBadAngel its using imagebuttons
20:58 sapier no it's not broke you assume local screensize variable provides real screensize  thats wrong
20:58 sapier true it's name is missleading but I told this about 3-5 times now
20:58 sapier do I understand correct that you did even reorder image button parameters?
20:59 RealBadAngel ok, i will try that
21:00 RealBadAngel yes i did but i will make it to be compatible
21:00 RealBadAngel i mean i will put tooltip at the end
21:00 RealBadAngel which is ofc wrong, but let  it be
21:01 RealBadAngel i wanted to keep convention, name, label, tooltip
21:01 tomreyn joined #minetest-dev
21:01 tomreyn joined #minetest-dev
21:02 RealBadAngel but propably its easier to have things fucked up in the engine than edit some mod
21:03 sapier well rba that's what happens if it's not done from begining ... curse of compatibility
21:03 RealBadAngel fuck it
21:03 RealBadAngel we will be sorry about that sooner or later
21:04 RealBadAngel we are all the time anyway
21:04 sapier but that's the way it is ... it's hard for us true, but do you know how annonying it is for modders to always have to update their mods with each version?
21:05 RealBadAngel why my mod that uses formspec extremaly heavily is not affected?
21:05 RealBadAngel but some mod that uses a few boxes is?
21:06 RealBadAngel now we are makin formspecs even slower
21:06 RealBadAngel constantly checking elements length, if 5 if 6 or 6 or 8 or maybe 9
21:07 RealBadAngel and each combination of above, even in function code
21:07 RealBadAngel its not compability, its insanity
21:07 sapier so you're gonna volonteer to update all broken mods?
21:07 RealBadAngel well, at least im sure i will cut that shit with tables
21:07 sapier if not stop complaining
21:08 RealBadAngel you could probably count to a few
21:08 RealBadAngel with broken mods
21:08 sapier checkboxes aren't some obscure never used gui elements you're free to change on every release
21:09 sapier expecially if the only benefit is some more nice parameter order
21:09 RealBadAngel ok ok, i said i will change that today
21:10 sapier if we really wanna do things like that lets add some formspec version field and add a second parser ... not sure if you wanna do that much work for a single moved parameter
21:10 RealBadAngel after that patch i dont wanna work on formspecs anylonger, it just made me sure theyre dead end
21:11 sapier nothing is more permanent then temporary solutions ;-)
21:11 RealBadAngel adding more features will make them even more complicated and slower
21:11 sapier yes
21:12 RealBadAngel we do have tables
21:12 sapier and at some point adding new features will be more painfull then rewriting it in total ... that's the point where switch will be realistic
21:12 sapier of course that point is highly subjective
21:12 RealBadAngel i feel like im already there
21:13 RealBadAngel motivated enough to code it
21:13 sapier well look for some generic gui language used by someone else then as and suggest how to integrate it to minetest
21:15 proller html
21:16 sapier well if you're talking about html5 maybe
21:23 crazyR +1 ^
21:23 crazyR sorry had to add my 2 pence worth lol
21:28 RealBadAngel why the duck using anything else than a simple table?
21:30 RealBadAngel easy to modify on the lua side, easy to transfer to the engine
21:30 RealBadAngel on the engine side whole parser thing (almost whole) will become obsolete
21:31 RealBadAngel and we will just read the table and fill existing structures
21:32 RealBadAngel also problem of varying number of variables will just not apply there
21:33 RealBadAngel theres no point in bringin there another language
21:36 sapier https://gist.github.com/sa​pier/81453b7c276ec902d2b7
21:37 VargaD_ joined #minetest-dev
21:38 Megaf_ sapier; how do I add this to my fork? https://github.com/minetest/minetest/pull/1359
21:38 sapier no matter if it's a table or something else rba you DO define another formspec language on implementing it
21:38 sapier and that's another proprietary one
21:38 sapier ah ... you remember me I have to push the rebased version of this one
21:41 RealBadAngel sapier, ok, push it
21:41 sapier have you tested it?
21:41 sapier is your issue fixed?
21:43 RealBadAngel wait a sec, will test it a bit more
21:44 Megaf aff, I hate github, it doesnt make sense at all
21:44 Megaf I created by mistake pull requests to sapier's fork and minetest already
21:44 ShadowNinja sapier: Your gist has style issues but looks good otherwise.
21:45 sapier where do I have to fix?
21:45 ShadowNinja sapier: spaces around binary operators.  And maybe use string.empty() instead of string != "".
21:46 ShadowNinja sapier: You're missing an if in parseButton I think.
21:47 sapier where?
21:48 RealBadAngel and it doesnt apply to current sources
21:48 ShadowNinja sapier: if (!tooltip.empty()) missing before spec.tooltip = tooltip;
21:48 sapier strange, thought I did base on it
21:48 RealBadAngel http://pastebin.com/F3ct5sZr
21:49 sapier ShadowNinja I don't wanna change all usages in there and there are way more then the new ones
21:49 ShadowNinja sapier: O.K. Either way, but fix the if.
21:49 sapier ahh ... usuall diff error on copying from console .. sorry
21:49 sapier what if?
21:49 sapier I don't see it
21:50 sapier can you be a little bit more precise? ;-)
21:52 ShadowNinja sapier: In parseButton it doesn't check if the tooltip is empty before setting it, but it does in other placed.
21:52 ShadowNinja -d+s
21:52 sapier hmm is this necessary?
21:53 Megaf [14:14:15] <sapier> could someone please test this changes https://github.com/minetest/minetest/pull/1359 they provide up to 20% more fps
21:53 Megaf It wont compile http://paste.debian.net/105724/
21:54 sapier oops pushed wron version
21:54 sapier give me a few minutes to cleanup the tooptip things
21:54 Megaf ok
21:56 sapier what about removing those ifs around tooltip comletely
21:56 ShadowNinja sapier: Also, use simply tostring, or rely on Lua's automatic string conversion, when converting booleans to strings.  dump is designed for more complex values and it's output is undefined since it's only for humans to read (normally for debugging).
21:57 sapier I don't see any lua code in there Shadow
21:57 ShadowNinja sapier: If the empty string is interpreted as no tooltip just set it directly without a seperate tooltip variable.
21:58 ShadowNinja sapier: Lua code where?
21:58 paramat joined #minetest-dev
21:59 sapier in that diff there ain't lua code
22:00 ShadowNinja sapier: Also, tbl.X is equivalent to tbl["X"], but shorter and simpler.  Why do you use the second format in the mainmenu so much?
22:00 ShadowNinja sapier: Yes, I know.  Did I say there was?
22:00 ShadowNinja sapier: Oh, not in that patch.
22:00 sapier then I don't know how it's related to it ;-)
22:01 ShadowNinja sapier: I was refering to usage of dump() in mainmenu/tab_settings.lua.
22:01 ShadowNinja I was reviewing a related PR.
22:01 sapier most likely because I started to use it and kept that style
22:02 ShadowNinja RealBadAngel: Can you feasably fix https://github.com/minetest/minetest/issues/1388 ?
22:04 sapier https://gist.github.com/sa​pier/b4b3de15c639f16af352 ready to merge?
22:06 ShadowNinja sapier: Can you just do spec.tooltip = parts[4]; instead of using a seperate temporary variable?
22:06 sapier no
22:06 sapier parts 4 is optional
22:06 sapier so it might be a out of bounds access
22:07 ShadowNinja sapier: I mean in the appropriate if block.
22:08 ShadowNinja (((parts.size() >= 5) && (parts.size() <= 9)) && (parts.size() != 6)) < Unnecessary grouping.
22:08 sapier possible but more understandable at least to me
22:09 paramat hmmmm, please could you look at this pull request for a linear lighting table? https://github.com/minetest/minetest/issues/1362 seems to me there must be good reasons for an exponential table, unfortunately the pull is popular and may be merged soon unless someone who understands lighting better than me explains
22:11 ShadowNinja paramat: It's a issue, not a pull request.  Although it does suggest a very specific code change.
22:11 sapier https://gist.github.com/sa​pier/81453b7c276ec902d2b7 now ready to merge?
22:12 hmmmm the reason for an exponential decay is because minecraft does it
22:12 hmmmm now, I'll agree that lights are too dark
22:12 hmmmm but that's because the range of light is so limited
22:12 hmmmm because there are only 4 bits for lighting
22:12 ShadowNinja sapier: Line 91, you still create the variable but don't use it.
22:13 ShadowNinja sapier: And in other places.  Otherwise it's fine.
22:13 sapier could someone enable unused variable warnings? :-)
22:14 hmmmm I think it could be worthwhile trying to remove the day night ratio, add 4 bits to the light values, and make day/night happen with hardware lighting
22:14 hmmmm scene lighting
22:14 hmmmm having hardware lighting for light source nodes isn't even entirely necessary
22:15 hmmmm but, I think we discussed how we would do it if we wanted light source nodes to use scene lighting, by dynamically calculating lightmaps
22:15 hmmmm which could be done alongside the mesh gen
22:16 hmmmm adding 4 bits to light would be a decent step toward the greater goal of removing node lighting information
22:20 asl joined #minetest-dev
22:22 sapier I'm pushing the tooltip fix now
22:24 BlockMen left #minetest-dev
22:27 Fresh_me_ joined #minetest-dev
22:45 sapier https://github.com/minetest/minetest/pull/1359 pushed a rebased version
22:47 paramat thanks, seems to me linear table would produce a different attenuation of light around light sources of different levels, exponential makes a consistent drop from node to node. i sensed a disturbance in the force, seems linear would break so much other stuff. oops yes it is an issue :)
22:48 VanessaE joined #minetest-dev
23:06 paramat hmmmm, there are now 3 voxelmanip floatland mods, 2 by me, and HeroOfTheWinds is working on large undergrond caverns, so i feel indev mapgen (and math mapgen) can now be removed. BTW here's my new mapgen https://forum.minetest.net/v​iewtopic.php?f=9&amp;t=9296
23:07 proller +1 for removing indev and math
23:08 sapier is a lua based mapgen already fast enough?
23:14 paramat my simple floatland mods are 2 sec per chunk non luaJIT, probably a little slow for a server. however i think hmmmmm is planning a core 'aether' type mod
23:14 paramat (not mod, mapgen)
23:35 sapier left #minetest-dev
23:48 iqualfragile joined #minetest-dev

| Channels | #minetest-dev index | Today | | Google Search | Plaintext