Minetest logo

IRC log for #minetest-dev, 2015-12-30

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

All times shown according to UTC.

Time Nick Message
00:01 sapier and you're gonna add +1 if protocol is updated next time? ;)
00:02 cmdskp I already did ;)
00:02 cmdskp but that may be incremented further before I find a way round myself ^_^
00:02 sapier usually you should change things in a way old clients wont even notice it
00:02 sapier e.g. by adding additional fields
00:02 sapier old clients just ignore them
00:03 sapier new clients should handle both old style and new one prefering the new one if present
00:03 red-001 joined #minetest-dev
00:03 cmdskp yes, unfortunately, I'm not up to scratch on the network system to determine the best way to handle it, so I'm hoping for some help with this bit :)
00:04 sapier it's always better to check what you actually need then check some number that might or might not be correct
00:04 gregorycu joined #minetest-dev
00:05 red-001 is anyone working on client sided lua?
00:05 cmdskp From my understanding, looking at other parts of the code around it, they used to pass in a version number as a byte in some sections and then determine from that what is needed
00:06 cmdskp or what is expected
00:06 est31 red-001, no, but its a promised bride.
00:06 est31 both hmmm and c55 want to do it, and have done preparations
00:07 * red-001 wonders how hard it would be to do
00:08 sapier est31 me to about a year ago ;-P
00:08 red-001 why not anymore?
00:08 sapier eventually someone will really do it
00:09 red-001 and (hopeful) get their PR merged
00:09 sapier well I was distracted from doing it by some formspec things I merged for someone else ... resulting in me to fix all the "changes"
00:09 est31 red-001, there are many issues client side lua has to do
00:09 est31 fulfill
00:09 est31 like sandboxing
00:10 est31 and other issues are disputed between core devs
00:10 sapier est31 actually all of them are already done
00:10 est31 e.g. whether to run it in the main thread, or in a separate thread
00:10 sapier 90% of it's code is already in
00:10 est31 sapier, you mean SN's sandboxing?
00:10 sapier nope
00:10 est31 its nice, but has to be more strict.
00:10 red-001 then whats stopping it?
00:11 est31 idk
00:11 sapier when I implemented the lua based mainmenu I already did write it in a way you can create lua stacks with minimal content
00:11 red-001 if 90% of the code is done
00:11 est31 everybody wanting to do it and saying "i want to do it" then not doing work on it until sb else says "I do it"
00:11 sapier actually the async worker threads of mainmenu ARE those minimal environments
00:12 sapier My plan by that time was starting the client side lua with some minimal feature set
00:13 sapier e.g. first only supporting prediction of formspec inventory changes
00:13 est31 cmdskp, where exactly do you want to get the version?
00:14 sapier but I'm not very eager to spend my time on work someone else already started
00:14 cmdskp line 1664 of content_cao
00:15 sapier so unless hmmm and celeron don't say "do it" I'm not gonna start working at it again
00:16 cmdskp a test for the client to check if the server is returning protocol version 27, I believe
00:16 est31 cmdskp, can you give me a github link of that line?
00:16 sapier imho first prototype is about one week of work ... assuming most of the preparations I did are still intact
00:16 est31 cmdskp,  click on the line, it will change the url then copy it here
00:16 cmdskp https://github.com/cmdskp/minetest/b​lob/54ede741a9b01b34b8a64cf2cc929398​a5d167d5/src/content_cao.cpp#L1665
00:17 cmdskp ty
00:17 est31 cmdskp, I think you only have to catch the serialisation error here
00:17 est31 using exceptions like if clauses hehe
00:18 est31 but idk, its better than raising protocol version for nothing
00:18 est31 or do you already raise it for some other reason?
00:18 cmdskp ah, that's good to know - I wonder considering the comments below about if sends nothing
00:18 sapier est31 is right catching the serializatiin error is the common way this is done
00:19 sapier protocol version change usual is only done if there's something incompatible
00:19 cmdskp the only reason I can think of is to deny an old client connection to a server running with slippery engaged - it'd be advantageous to the client to not be slipping compared to those up-to-date
00:20 cmdskp on ice, for example
00:20 sapier isn't slippery something like "hunger" ... I guess some ppl wont like it very much
00:20 est31 yeah
00:21 est31 in fact I dont I have to admit
00:21 cmdskp fortunately, I've included a method to override slippery to allow cancelling it out
00:21 est31 still it should be added imo bc some ppl like it
00:21 cmdskp for modders
00:21 cmdskp It's setup in the node groups too, so people can include or not as they wish :)
00:22 sapier to be honest I'd not like it ... but if others do
00:23 cmdskp It won't be active unless people add a slippery=X to a node
00:23 sapier especially as I don't see a good way to implement this for entities or items
00:23 cmdskp entities and items slip too
00:23 sapier e.g. what to do for entities with > 1x1 collision boxes ... calculate average slippery?
00:24 cmdskp they use the node slippery factor to adjust their relative speed
00:24 sapier which one? center? corner?
00:25 est31 well in fact the problem exists for 1x1 collision boxes too
00:25 cmdskp That's handled by Irrlicht I believe, which appears to use the whole bounding box for collisions
00:25 sapier e.g. have a huge whale 3x4 and a single line of ice in the middle  ... shall it slip to the end just becaus it's center is above the ice? ;-)
00:26 sapier est31 yes but there the effect isn't that obvious
00:26 cmdskp Oh, it doesn't keep slipping normally, just for a few nodes (depending on the setting)
00:27 sapier I implemented friction for my monorail mod ages ago ... effects of physical correct implementation sometimes feel strange in minetest environment :-)
00:27 sapier especially if you implement air resistance too :-)
00:27 cmdskp It would be computationally expensive to check the whole area under an entities collision box
00:27 gregorycu Hello all
00:28 est31 hi gregorycu !
00:28 sapier still without calculating inertion the result will be suboptimal
00:28 est31 inertion?
00:28 gregorycu I think I better have a look at some of my PRs
00:28 cmdskp It does raise an interesting option to add - a slip factor on non-player entities to stop the slippery, like I've added for players with physics_override
00:29 sapier so adding slipery somehow results in needing inertion ... needing medium resistance (different in water then in air ;-) )
00:29 est31 i know what an inertial system is, but inertion...
00:29 cmdskp that way you can prevent big entities from slipping at all, if desired and perform a more thorough check inside the entity onstep
00:29 sapier sounds like a feasable workaround ... maybe not even enable slipery for those by default
00:31 sapier but I still don't like it yet I'll try to show you issues it's up to you and those who like the feature to evaluate how important those issues are
00:31 cmdskp that's possible, there's no real reason to have entities slide except for coolness and fun-factor to see thrown items slide across the ice =D  That and expections I guess, if someone has set ice to slip and entities don't by default would seem incongruous :)
00:31 est31 yeah collision box calculations are a place where areastores can profit as well
00:32 cmdskp I'm afriad I must initiate my sleep cycle, but thank you for an interesting discussion.  Food for thought :)
00:32 sapier well I'd wanted a "on_collision" entity callback for some time guess I'm the only one who sees use for it
00:33 cmdskp I was wanting that too for handling momentum on carts
00:33 sapier I guess I have to fix the facedir rotation limit once again ... seems like it always rotates to same direction even if rotation to different direction would be way less change :-/
00:34 * cmdskp exits to sleep cycle :)
00:35 est31 real programmers don't count sheep, they count nops xD
00:35 sapier I hate nop slides
00:37 est31 just use custom microcode for your platform, and patch your compiler
00:38 sapier I've got enough work on working around cpu bugs ... hate to do this .. fixing hardware bugs in software ....
00:39 sapier https://github.com/minetest​/minetest/pull/3454/commits what about merging the non irrlicht updates till irrlicht fixes it's issues?
00:40 est31 sapier, you do embedded stuff, being one of the few devs who use that piece of hw?
00:41 sapier for what I know I'm one of about 3 ppl using that particular part of the asic yes
00:42 sapier well that part is used in other asics too ... yet in different revisions
00:42 sapier but hw bugs seem to be usually added to each new version instead of being fixed
00:44 est31 lol
00:44 est31 sapier, fine with me merge the commits
00:44 est31 but note there is more recent curl version out
00:44 est31 ermm sorry
00:44 est31 its most recent
00:44 est31 openssl too
00:45 sapier I'll check once again ... well it's updating quite some libs not surprising its outdated after some weeks
00:46 sapier but android build server has to be updated using gcc 4.9 fails on old version used there
00:46 est31 yeah suprised to see there still was potential to update
00:46 est31 after all, I'd pushed a similar commit not a long time ago
00:46 sapier I'd wanted to update irrlicht but obviously current dev version has major issues
00:47 est31 2f58e4d9624b4f31c5064dc640e0ad40d8adb9ea
00:48 sapier was there any reason for not using gmp in openssl?
00:49 est31 no
00:49 est31 i just didnt thought of it
00:50 est31 I mostly needed it for other things
00:50 est31 like, cryptography when logging in
00:51 sapier1 joined #minetest-dev
00:53 sapier joined #minetest-dev
00:54 sapier I don't see a more recent openssl version then 1.0.1e
00:58 sapier 1.0.2e of course
01:00 Robby joined #minetest-dev
01:02 gregorycu joined #minetest-dev
01:23 sapier left #minetest-dev
01:30 gregorycu Did c55 do his farblock stuff?
01:34 gregorycu Or farmap, whatever it is called
01:47 Robby joined #minetest-dev
01:54 VanessaE gregorycu: he has yet to merge it
01:54 gregorycu Thanks
01:54 VanessaE but there is a fork you can play with.
01:54 VanessaE https://github.com/celeron5​5/minetest/tree/far_map_wip
01:56 gregorycu How's it look?
02:04 VanessaE https://daconcepts.com/vanessa/hob​bies/minetest/screenshots/random/S​creenshot_2015-11-13_13-03-08.png
02:04 VanessaE https://daconcepts.com/vanessa/hob​bies/minetest/screenshots/random/S​creenshot_2015-11-13_12-59-47.png
02:04 VanessaE it looks like that. :)
02:04 VanessaE in my GPU I can see "terrain" 1 km into the distance, and still with respectable framerates.
02:05 VanessaE those ^^^ screenshots I think have the farmap limit set to around 500m
02:08 gregorycu lol png
02:09 gregorycu Interesting stuff
02:11 VanessaE why wouldn't I use PNG for a screenshot? :)
02:11 VanessaE celeron55: ^^^^
02:18 gregorycu PNGs are lossless, but large size for things like screenshots
02:19 VanessaE yeah I know
02:19 VanessaE but where it comes to screenshots, pixel-accuracy is important to me :)
02:30 gregorycu Yeppers
02:35 gregorycu_ joined #minetest-dev
02:56 younishd_ joined #minetest-dev
02:57 gregorycu Who else does windows dev here?
03:28 gregorycu sofar
03:29 sofar nooooo
03:30 gregorycu Too bad
03:30 gregorycu Found you
03:30 gregorycu So, you've been looking at my PR that "optimises" spreadLight
03:31 gregorycu https://github.com/minetest/minetes​t/pull/3331#issuecomment-167665040
03:31 sofar well, just a casual look.
03:31 gregorycu Yep
03:31 gregorycu Are you able to indicate exactly how you timed spreadLight
03:32 gregorycu TimeTaker was used, as you said
03:32 sofar there's a commented out TimeTaker, that's all I did
03:32 gregorycu I want to make sure I have the same setup as you, to give me the best chance at your results
03:32 sofar I honestly don't even know if that's the right way
03:33 gregorycu Cool, thank you
03:34 gregorycu Hopefully I can speak with paramat too
03:35 sofar map.cpp:                        //TimeTaker timer("spreadLight");
03:35 sofar map.cpp-                        vmanip.spreadLight(bank, light_sources, nodemgr);
03:36 sofar that one
03:36 sofar hey, question for you windowzy person
03:36 sofar why does minetest go nomnoms on system ram and occupy 3gb or memory?
03:37 gregorycu It shouldn't
03:38 sofar perhaps I should update
03:38 sofar resolution is 2560x1600, maybe that matters?
03:38 sofar nvidia hw?
03:38 gregorycu I wouldn't think so, to be honest
03:38 gregorycu ATI
03:39 gregorycu vmanip.spreadLight(bank, light_sources, nodemgr);
03:39 gregorycu That's behind an #if 0
03:40 gregorycu You mean line 831?
03:40 sofar I believe that the macro works out that if you uncomment it, it does an extra recalc pass and times that
03:40 gregorycu etc
03:40 sofar #826-828 yes
03:44 gregorycu Where does info stream get printed?
03:54 cheapie This "AreaStore" thing looks cool, but I can't find any obvious way to save/load them to/from disk. Is that even possible?
04:30 est31 joined #minetest-dev
04:30 gregorycu For me spreadLight is called non-stop
04:30 est31 cheapie, it isn't possible as of now, but this PR will enable it: https://github.com/minetest/minetest/pull/3309
04:31 cheapie est31: OK.
04:31 est31 cheapie, so you'll have to use a table and insert/remove from it while operating on the areastore
04:31 est31 & then save that table
04:32 est31 joined #minetest-dev
04:33 est31 gregorycu, its printed nowhere
04:33 est31 gregorycu, execute minetest with the --info param
04:33 gregorycu I added the info switch
04:33 gregorycu yeah
04:44 gregorycu sofar, for me spreadLight is getting called like 100 times a second for no obvious reason
04:44 gregorycu Did you experience this?
04:50 sofar gregorycu: nope
04:52 asl97 joined #minetest-dev
04:54 Player2 joined #minetest-dev
04:58 yyt16384 joined #minetest-dev
05:50 gregorycu_ joined #minetest-dev
06:38 Hunterz joined #minetest-dev
06:52 gregorycu Does anyone know paramat's time zone?
06:53 gregorycu That dude and I have unfinshed business
06:55 asl97 gregorycu: unless paramat has given false information, just check his/her github profile page: https://github.com/paramat
06:56 gregorycu UK, which is 11 hours behind me... 7am
06:56 gregorycu Thanks, i should fill out that info on my profile too
07:47 Krock joined #minetest-dev
08:04 Darcidride joined #minetest-dev
08:05 gregorycu In minetest parlance, what is a chunk?
08:07 gregorycu vs node vs block
08:08 hmmmmm a chunk is a make-believe cubic region of some mapchunks
08:08 hmmmmm mapblocks i mean
08:08 hmmmmm it's not an actual data structure
08:13 nrzkt joined #minetest-dev
08:31 sofar http://i.imgur.com/Jvoeq7R.png
08:31 sofar ^^ completely client-side drawn connecting nodeboxes
08:31 sofar only a single nodedef
08:33 RealBadAngel that looks promising
08:33 RealBadAngel i could re-do my concrete posts with that
08:34 sofar yes
08:34 sofar plus mesecons wire
08:34 sofar that's probably the best user for it too
08:35 sofar although it'll break the current mesecons wire, since it connects diagonally
08:35 sofar this only connects 6 ways (up, down, n-e-w-s)
08:35 RealBadAngel it has to be done in all directions
08:35 RealBadAngel you know why its diagonally only?
08:36 sofar it's not only diagonally
08:36 RealBadAngel it has separate vertical one
08:36 sofar it's diagonally up
08:36 sofar right
08:36 sofar which is weird since it can also go vertical
08:36 RealBadAngel to get one climbing all possible walls, theres over 1k nodedefs needed
08:36 sofar but that's a different node
08:37 RealBadAngel ask nore, he made such code
08:37 RealBadAngel and thats only for just one type of wire
08:37 sofar well right now this code could be extended to allow specifying how to connect to a diagonally connected face
08:38 sofar but let me make this into a decent PR tomorrow so people have something to test
08:38 RealBadAngel such code is aviable in redpower2 clone, its open source
08:38 sofar I also have a simple "wall" mod to go with it
08:38 RealBadAngel i mean connection rules
08:39 RealBadAngel its called redproject i think
08:39 sofar I gotta hit the hay first. Will put make-up and a dress on this code tomorrow so people can ogle it
08:40 sofar actually, I should convert fences to this code as well
08:40 sofar so that people see that it's relatively flexible
08:40 RealBadAngel indeed
08:40 RealBadAngel that has to be generic
08:41 sofar http://i.imgur.com/SBURS20.png
08:41 RealBadAngel have you checked my posts?
08:41 RealBadAngel from technic modpack
08:43 gregorycu Thanks hmmmmm
08:43 sofar I didn't... I'm fedup with people saying "look at my code" and it's gplv3 ... so I can't merge it back
08:44 RealBadAngel https://github.com/minetest-tech​nic/technic/tree/master/concrete
08:44 sofar it's not rocket science, just reimplement it
08:44 RealBadAngel my code is usually wtfpl ;)
08:44 sofar yeah you're not crazy, like some people maintaining hostile forks
08:45 RealBadAngel technic is actually lgpl v2
08:45 sofar and then claiming I plagiarize them with 6 lines of fucking trivial code
08:45 RealBadAngel yeah i seen that
08:45 sofar YOU KNOW WHAT GRINDS MY GEARS?
08:45 sofar ;^)
08:46 sofar it makes more more angry since he knows other people can't merge it back, but meanwhile he himself pushes code around from and to various gpl trees
08:46 sofar and it's not helping anyone
08:50 RealBadAngel folks doing that are acting like real dicks
08:51 RealBadAngel i dont like such attitude
08:57 sofar ok
08:57 sofar ONE more screenshot
08:57 sofar then I go to bed
08:57 sofar http://i.imgur.com/lNECgTt.png
08:57 sofar tadaa, new fences using this method
08:58 sofar textures used was jungletree and tree (custom texture pack)
08:58 RealBadAngel can i see lua code for that fence?
09:21 enesbil joined #minetest-dev
10:31 OldCoder Is it a known bug that for 0.4.12 clients everything disappears at around 23250 in any direction?
10:31 OldCoder As opposed to 32000
10:45 yyt16384 joined #minetest-dev
10:52 Calinou joined #minetest-dev
11:02 gregorycu lol
11:12 red-001 joined #minetest-dev
11:25 zupoman joined #minetest-dev
11:30 red-001 do I need to add a description of a new privilege some where in the documentation?
11:43 CraigyDavi joined #minetest-dev
11:49 CraigyDavi joined #minetest-dev
11:57 nrzkt red-001, yes
12:03 red-001 where?
12:04 red-001 as in I defined it
12:04 red-001 where to document it?
12:19 yyt16384 Should RemoteClient::SetBlockNotSent and RemoteClient::SetBlocksNotSent set m_nothing_to_send_pause_timer to 0?
12:19 yyt16384 It will not be in a "nothing to send" state after marking the block as not sent.
12:20 yyt16384 This seems to cause modified blocks be sent only every 2 seconds
12:25 rubenwardy joined #minetest-dev
12:27 cmdskp joined #minetest-dev
12:44 celeron55 yyt16384: why is there such a race to modify that part of the codebase now that i have a stalled branch with big rework on it
12:45 celeron55 can someone take far_map_wip and remove the far map stuff from it and merge it upstream instead of fixing the current code that is vastly inerior?
12:46 celeron55 inferior*
12:46 celeron55 it has a reworked map sending implementation that i developed along with the far map stuff
12:49 gregorycu Sorry bro
12:49 gregorycu Everyone converges on the stuff that needs the most work?
12:51 celeron55 i guess so but it's not good
12:51 celeron55 most work will end up being lost
12:51 gregorycu Agreed
12:52 gregorycu How do we avoid this problem
12:53 celeron55 given the circumstances, probably in no way
12:54 gregorycu it's better than nobody wanting to do any work :)
12:54 gregorycu Now where the hell is paramat
12:55 celeron55 i'm now considering trying to rip apart the map sending stuff and the farmap stuff in my branch
12:55 celeron55 this is going to take days i guess but maybe it is worth it
12:55 gregorycu I heard the farmap stuff is pretty good
12:56 celeron55 it can't be done by picking commits; i need to first pick commits, then squash everything that can't be separated based on commits and then edit and delete files until there aren't any unwanted changes left
12:56 gregorycu Hopefully your git-fu is better than mine
12:57 celeron55 the farmap stuff is technically fine but it needs vast usability improvements
12:58 celeron55 like removing the viewing range auto-adjust and changing the configuration UI to make it reasonable to adjust by users, and figuring out sane defaults
12:58 celeron55 lots of work
12:58 celeron55 i'm not sure how to get that work done though
12:58 gregorycu Whats wrong with the autoadjust?
12:59 celeron55 it won't be able to work properly
12:59 gregorycu Won't the far map just "fill in the blanks"?
12:59 celeron55 it can't know how the user wants to balance the far and short ranges
13:00 gregorycu Ahh, I see what you mean
13:00 celeron55 and fps
13:00 gregorycu The two extremes are seeing very far with farmap, or short distance with very little farmap
13:00 gregorycu seeing very far with lots of farmap *
13:01 gregorycu You could just make it a static %
13:01 celeron55 also different computers can render different amounts with the two styles; they don't perform equally related to each other
13:01 celeron55 also farmap takes way more memory
13:01 gregorycu Is that something you can detect? Available memory?
13:02 celeron55 not in a useful way
13:02 celeron55 this was discussed previously with hmmmm
13:02 celeron55 i have already decided this, you can't say anything that would change my mind about the auto-adjust
13:02 gregorycu I don't even have a position on auto-adjust
13:03 celeron55 i made the auto-adjust and i know what it's capable of; it's not capable of this
13:03 gregorycu I was asking what was wrong with it
13:03 celeron55 a thing that also messes up the auto-adjust is that with the map sending changes, the map sending is controlled by the client's view range
13:03 celeron55 not by server's configuration
13:04 celeron55 so then you're controlling both the network and the rendering
13:04 celeron55 based on rendering
13:05 gregorycu I have idea, but you've probably thought of them, so no point speaking them
13:05 celeron55 even now without any of the changes, i think most people want to get rid of the auto-adjust and don't use it at all
13:06 celeron55 personally i certainly don't use it
13:06 gregorycu Doesn't it currently have problems (without your changes?)
13:06 gregorycu Where it see-saws or something
13:08 celeron55 sometimes you hear people complaining about it
13:08 celeron55 i believe it is tuned as well as it can be; if you made it less bouncy, then it would be too slow in other scenarios
13:09 celeron55 and the algorithm itself is as good as i could make it; i certainly won't attempt to make it more clever
13:09 gregorycu I think the issue is that when it increases the range, the first frame updates draw lists etc. which causes the frame rate to take a hit
13:10 gregorycu I was looking at it some time ago, I can't remember
13:10 celeron55 sounds plausible
13:11 celeron55 anyway
13:11 gregorycu I think a more "clever" solution would be to exclude that processing from calculating the current frame rate (because it's a one-off)
13:11 gregorycu But probably not worth it
13:12 celeron55 the way i personally want to control my far map range is in such a way that it is static no matter what happens, and the way i then want to control my normal rendering range is that it is static too, because increasing it for the sake of seeing farther makes no sense when the far map is already doing it
13:13 celeron55 the point of the auto-adjust is to increase the range when the rendering performance allows for it on an empty terrain or a tower or something
13:13 celeron55 that point doesn't exist then anymore
13:13 gregorycu There is another point though
13:14 gregorycu That's when you start to see farmap vs real rendering
13:14 gregorycu another = new
13:14 celeron55 yes, but it's nicer to have more fps than have that point move around
13:16 gregorycu I think different people would want different things (not saying it's worthwhile complicating everything to satisfy everyone)
13:16 gregorycu (Some people use auto-adjust, some do not)
13:17 celeron55 have you tried the far_map_wip branch?
13:17 gregorycu I haven't
13:17 celeron55 it's much easier to think about what people might think about this issue when you have first-hand experience of it
13:18 celeron55 it's more different than you think
13:18 celeron55 (likely)
13:18 gregorycu Yeah, I'll try it soon
13:18 celeron55 (if it's not, then that is useful information too; but you don't know until you try)
13:19 gregorycu I want to get this perf shit sorted with paramat
13:20 gregorycu And I need to consider getting a linux vm happening so I can do these tests myself
13:22 gregorycu I think he did his perf tests with a 121^3 chunk which was not something I considered when optimising code
13:22 celeron55 if others in the core team think that far map with related rework and fixes should be merged with the farmap defaulting to being disabled, i could do that
13:25 gregorycu Seems like a good stepping stone
13:25 celeron55 or, well, i guess even in an enabled form
13:53 behalebabo joined #minetest-dev
14:18 Hunterz joined #minetest-dev
14:25 zat joined #minetest-dev
14:28 gregorycu "i modified one of my mapgen mods to use the non-mapgen object voxelmanip, which uses 'update map' for lighting calculation, which calls 'updateLighting' which calls 'spreadLight'. I made only 1 mapchunk be generated and this mapchunk is identical in each run. Here 'spreadLight' is acting on a 112^3 node volume"
14:28 gregorycu Anyone know what this means? He made a 112^3 "voxelmanip"
14:29 Krock a voxelmanipulator with those area specifications
14:29 Krock 112m in each direction
14:30 gregorycu A voxelmanip is an object to transactionally change the map right?
14:31 Krock yes, it gives you an array with all node IDs. you can change them if you like
14:32 Krock and when you're done, the complete piece gets written to the map
14:32 Krock vm:get_data() and vm:set_data(data) + vm:write_to_map()
14:33 gregorycu So he wrong a 112^ chunk of map, which caused updateLighting to be called on that chunk
14:33 gregorycu Cool, I'll give that a try
14:33 gregorycu s/wrong/wrote/
14:34 Krock oh, that's done with vm:calc_lighting(p1, p2
14:35 Amaz joined #minetest-dev
14:36 gregorycu Do people actually update 112^3 chunks in practice?
14:36 red-001 joined #minetest-dev
14:36 Krock that's a huge amount of nodes, so I doubt that happens frequently
14:37 Krock but there are such operations to fix the bugged light
14:37 celeron55 the map generator uses chunks of roughly that size
14:38 celeron55 not sure what exactly, but that's not very far off
14:38 red-001 what causes the light bugs?
14:38 gregorycu ok
14:39 Krock usually 16*5 in each dimension
14:39 Krock sometimes more.. dunno why
14:39 gregorycu Then the lighting code needs to be able to handle 112^3 nicely
14:40 gregorycu I'm not sure what causes the light bugs
14:40 gregorycu Maybe trees?
14:44 red-001 https://github.com/minetest/minetest_game/pull/602
14:44 red-001 https://github.com/minetest/minetest_game/pull/667
14:51 rubenwardy #602
14:52 red-001 can someone take a look at it
14:52 red-001 ?
14:53 younishd joined #minetest-dev
14:56 gregorycu Must only be called in a mapgen thread...
14:57 gregorycu do
14:57 gregorycu h
14:57 Krock game#602
14:57 Krock ShadowBot!
14:58 gregorycu Classic shadow bot
14:58 red-001 https://github.com/minetest/minetest_game/pull/602
14:59 cmdskp joined #minetest-dev
14:59 red-001 something tells my I will wait a long time for this to be merged
14:59 red-001 me*
14:59 gregorycu Well, brand new concepts generally do
15:00 gregorycu take a long time, that is
15:00 red-001 no that has write access even commented on it
15:02 gregorycu It's the holiday season, people are on holidays/busy
15:03 red-001 yeah I guess your right
15:03 red-001 I made the pr in summer through
15:03 gregorycu This is where github is  alittle weak
15:04 gregorycu Usually before any work is done, people would discuss the concept, come to agreement
15:04 gregorycu Then somebody would go off and code it to spec
15:04 gregorycu PRs only become code style / form reviews
15:04 gregorycu Not a discussion about a feature and how it should work
15:05 red-001 also If someone gets a chance take a look at https://github.com/minetest/minetest_game/pull/667
15:20 rubenwardy Reviewed both, red-001
15:20 red-001 thanks
15:35 gravgun joined #minetest-dev
15:35 hmmmm joined #minetest-dev
15:46 sofar http://i.imgur.com/zaJq8jo.png
15:46 est31 joined #minetest-dev
15:46 est31 can sb have a look at https://github.com/minetest/minetest/pull/3434
15:47 est31 we _really_ have to do a pr merging session where we merge trivial PRs like this
15:47 est31 otherwise contributor after contributor gets annoyed.
16:18 Hunterz joined #minetest-dev
16:22 red-001 joined #minetest-dev
16:39 sofar https://github.com/minetest/minetest/pull/3503
16:49 cmdskp joined #minetest-dev
16:54 fling joined #minetest-dev
17:04 fling joined #minetest-dev
17:33 cmdskp joined #minetest-dev
17:58 DFeniks joined #minetest-dev
18:14 zupoman joined #minetest-dev
18:45 twoelk joined #minetest-dev
19:42 Fixer joined #minetest-dev
19:59 cmdskp joined #minetest-dev
20:36 CraigyDavi joined #minetest-dev
20:44 zat joined #minetest-dev
20:57 Gael-de-Sailly joined #minetest-dev
21:00 red-001 could the old settings tab be returned?
21:00 red-001 The new one is quite confusing
21:01 red-001 It could be used as an advance settings tab
21:03 Gael-de-Sailly some people have proposed it on october but nothing has changed
21:06 Gael-de-Sailly https://forum.minetest.net/vi​ewtopic.php?p=195208#p195208
21:09 red-001 well I might as well open a PR for this
21:09 red-001 as nothing is being do
21:10 Gael-de-Sailly pilzadam said that it will be done
21:10 red-001 ok
21:10 Gael-de-Sailly but it was on october
21:10 red-001 well he probably forgot
21:11 red-001 I doubt It will that this long to add back something so trivial
21:12 Gael-de-Sailly because that's not the priority for the devs
21:12 red-001 well a unusable user interface is a thing that can destroy any game
21:13 red-001 or engine
21:13 Gael-de-Sailly yes, I've said it many times, the "user-friendly" side is very important
21:15 red-001 good old minetest once doesn't merge PR because a setting in minetest.conf is not user friendly
21:15 red-001 and sometimes this happens
21:15 gravgun "user-friendly": term not found in coredevs/vocabulary.txt
21:15 gravgun :P
21:15 Gael-de-Sailly well said
21:15 red-001 user-friendly = nil
21:16 red-001 Is this used on android too?
21:16 red-001 or does android have it's own ui
21:16 gravgun "Remembering PR": term found in coredevs/vocabulary_ignore.txt, ignored
21:17 red-001 :)
21:17 Gael-de-Sailly they think to many things at the same time
21:17 red-001 they think -1 things
21:17 Gael-de-Sailly so they forgot the "less important" ones
21:17 asl97 imo, there wasn't much reason why they can't just include both the overridable function and setting, user could already change the game themselve
21:17 Gael-de-Sailly and generally the "user-friendly" is less important
21:18 Gael-de-Sailly that's the problem
21:21 Gael-de-Sailly the advanced settings it was my idea at the beginning, but it has been done the wrong way
21:21 Gael-de-Sailly The old interface has been removed
21:21 red-001 well it OK for an advance settings tab
21:22 red-001 I wonder what is it with open source projects and complicated settings
21:22 Gael-de-Sailly There should be both tabs with a BUTTON to SWITCH
21:22 red-001 or just two tabs
21:23 red-001 'settings' and 'advance settings'
21:23 Gael-de-Sailly no, you keep the old interface, and add an "advanced settings" button
21:23 red-001 yeah thats what I meant
21:24 red-001 something like this http://postimg.org/image/spj5akqml/
21:24 red-001 which PR megerd this?
21:24 Gael-de-Sailly yes, that's good too
21:25 Gael-de-Sailly merged what?
21:25 Gael-de-Sailly the new settings?
21:26 red-001 yes
21:26 Gael-de-Sailly iirc that was a commit from PilzAdam
21:26 red-001 nvm found it
21:26 * Gael-de-Sailly is checking...
21:26 red-001 https://github.com/minetest/minetest/commi​t/6f2d9de7691c99b833f71c1414230c5d04c42ae0
21:27 Gael-de-Sailly I have this one: https://github.com/minetest/minetest/commi​t/006ef5b4a5e362eed848ca26b411a8059d74b965
21:28 red-001 thanks
21:28 red-001 that seems to be it
21:33 red-001 imo it's time android gets it's own ui
21:35 kaadmy red-001: yes
21:36 Gael-de-Sailly good bye
21:55 Dragonop joined #minetest-dev
21:58 twoelk joined #minetest-dev
22:02 red-001 wow
22:02 sapier joined #minetest-dev
22:03 red-001 an no existing game set as menu_last_game cause minetest to crash
22:04 sapier I've got a quite ugly question: Is minetest really reading mesh from harddrive and loading to irrlicht EVERYTIME any property of an entitie does change?
22:05 * red-001 doesn't know
22:07 red-001 Is that as bad as it sounds?
22:07 celeron55 i'd guess the client doesn't save or load meshes from the hard drive at any other point than the initial cache synchronization
22:07 celeron55 it might re-parse them though
22:08 sapier as irrlicht output isn't cought I'm quite sure at least reparsing is done
22:09 sapier hopefully not reading from disk but as filename is passed to irrlicht I assume that's at best operating systems task
22:09 celeron55 the re-parsing is a workaround due to irrlicht not being able to make copies of a mesh for setting different vertex colors
22:10 sapier argh
22:10 celeron55 the filename is probably just due to irrlicht's weird interface which requires filenames for virtual files
22:10 sapier hmm could be yes
22:10 sapier well I've other issues to fix atm but flooding console with irrlicht messages ain't very good too
22:11 sapier most strange thing is it seems to only happen if you're in mainmenu on a server
22:12 sapier so I'm not even sure it's related to setting properties
22:15 celeron55 i don't see how the entity mesh creation thing would even be triggered from setting properties
22:15 kaadmy sapier: i think it does, but i dunno
22:15 celeron55 it runs once when an entity is created on the client, and never again for the same entity
22:16 sapier mesh is a property you can change it on an existing one too so there has to be some way to do it. Yet I'm not sure if it's really set properties as somehow "pause menu" is involved
22:17 sapier well not exactly "pause" as in multiplayer it ain't pause
22:17 OldCoder Interesting bug came up last night. Tried 0.4.11 and 12, -O2 and -O3, gcc 4.6.3 and 4.9.3, different glibc releases, different Irrlichts
22:17 OldCoder Same issue almost exactly for all combinations. Screen goes blank at exactly a uniform distance from the origin.
22:18 OldCoder When I switched glibc's, the distance changed slightly. It is exact for each test down to 0.1.
22:18 OldCoder In the neighborhood of 23300 or 23253 depending
22:19 OldCoder Has anything like this been observed or reported in recent years?
22:19 celeron55 wait, what
22:19 OldCoder Interesting part is that every changeable component has been changed
22:19 OldCoder Hm? Which part is unclear?
22:19 celeron55 actually yes, the mesh is re-created and the scene node is fully re-created when any property changes
22:19 OldCoder celeron55, are you addressing sapier or me?
22:20 OldCoder Ah, Sapier
22:20 OldCoder Has anybody ever heard of the issue I've described?
22:21 sapier guess I'll have to take care not to write identical settings too often :-(
22:21 celeron55 and that happens because in handling of GENERIC_CMD_SET_PROPERTIES, expireVisuals() is always called
22:21 celeron55 and if the visuals are set expired, the entire scene node is re-created from scratch
22:22 sapier maybe we should only do this if any of the visible properties changes
22:22 celeron55 well... obviously so
22:23 red-001 https://github.com/minetest/minetest/pull/3504
22:23 celeron55 i bet the next bug related to that is going to be a new visual property added after this to which somebody forgets to add the expiration trigger and it gets desynced
22:24 celeron55 unless you come up with a good way of avoiding that
22:25 sapier true that's almost sure gonna happen
22:27 sapier "best" way would be separating visible properties from non visible ones ... but as this can't be done in a compatible way it's no go
22:28 celeron55 OldCoder: how do you go to that distance? if this can't be reproduced (i certainly can't; tried it now), it's not fixable
22:28 celeron55 i tried teleporting in a test world to around 23whatever in some steps and at no point did it go blank
22:28 celeron55 except when i accidentally teleported inside a mountain
22:28 OldCoder I know, celeron55, will work on reproducing. I TP to 22500 and walk the rest of the way
22:29 OldCoder The interesting part is that I've changed *everything* even several years worth of glibc
22:29 OldCoder Not sure what else could be changed: Different optimization, different Irrlicht, different compiler, 2 years worth of MT...
22:29 * OldCoder figured it was an incompatibility with glibc -lm
22:30 OldCoder But it happens with both old and new glibc
22:30 OldCoder Wanted to ask if a limit below 32000 had ever been observed
22:30 * OldCoder takes it the answer is no
22:30 celeron55 tried flying; doesn't happen
22:31 OldCoder Thank you. So, limits below 32000 have not been observed?
22:32 OldCoder I mean, in past bugs?
22:32 OldCoder That is the only question for right now
22:35 celeron55 i have not heard of such
22:35 OldCoder O.K. Thanks. I'll comment further if specifics are identified.
22:35 celeron55 and i can't imagine what could cause it; it's probably something very weird
22:35 OldCoder Yes; hence the description as interesting
22:35 celeron55 are you running 32-bit or 64-bit?
22:36 OldCoder 32 bit
22:36 celeron55 okay, that could explain something
22:36 OldCoder Do most people do 64 these days?
22:36 OldCoder Oh? Tips on directions to look in?
22:36 celeron55 well, practically everyone, including me
22:36 OldCoder So if I find somebody else with 32 bits
22:36 OldCoder And they observe the issue, I have a clue
22:36 OldCoder O.K. That is something to try. Will comment further at a later date. Thank you.
22:37 celeron55 i would do that first, and then go from there
22:37 OldCoder o/
22:37 red-001 how do you set minetest to compile in x64
22:37 red-001 I have 64 bit winodows
22:38 red-001 but minetest compiles in 32 bits
22:38 sapier just flew from 22500 to 24500 no darkening for me too
22:38 celeron55 red-001: on linux it compiles to whatever the system is; on windows you'll have to ask from someone who has made 64-bit builds because nobody else probably has any idea
22:38 OldCoder sapier, thank you!
22:39 OldCoder Will find others who have 32-bit OSes and see if this is an issue
22:39 OldCoder Most possible explanations are ruled out, so need to check this one
22:39 red-001 OldCoder
22:39 red-001 I have 32 bit minetest
22:39 OldCoder red-001, But a 64 bit machine
22:39 OldCoder and OS, right?
22:39 red-001 yes
22:39 celeron55 i'm booting up my 32bit laptop now
22:39 OldCoder Well, thanks; it's a long-shot
22:40 red-001 maybe it's just a bug with 32 bit minetest
22:40 * OldCoder is betting on a glibc math library issue
22:40 OldCoder I should have noticed it years ago but never spent much time on the boundaries
22:41 kaadmy i have a 32 bit OS and MT
22:41 OldCoder Would not have mentioned it here yet but was bemused by the fact that changing MT releases, compilers, Irrlicht releases, optimizations..
22:41 OldCoder kaadmy, could you try this? fly to 23000,100,0
22:41 OldCoder then from there to 24000
22:41 kaadmy ok
22:41 OldCoder ty
22:42 OldCoder BTW encountered this because we're proceeding with stacked realms
22:42 red-001 I made a PR to add back the old settings tab https://github.com/minetest/minetest/pull/3504
22:42 OldCoder Lots of fun possibilities there
22:43 kaadmy wonder if the bug happens on mobile?
22:43 kaadmy checking
22:43 celeron55 i am not seeing the issue on this; keep in mind that this is not an upstream build but instead contains some of my farmap work-in-progress stuff; however i don't think it should have an effect
22:43 OldCoder There is no evidence yet it happens for anybody but me
22:43 OldCoder I thank people for commenting. I will try it under Windows on the same machine
22:44 OldCoder and see just how machine or OS specific it is. If nobody else can reproduce it, it can't be fixed...
22:44 celeron55 (that was 32-bit debian 7 or something like that)
22:44 OldCoder But I trust people agree it is interesting. Time again was appreciated. Talk to you later. celeron55 I will try Windows on the same hardware. Then the same Linux on different hardware.
22:44 * OldCoder is systematic :-)
22:44 OldCoder Ttyl
22:45 red-001 I'm running a debug build of minetest and I'm getting some strange crashes
22:45 OldCoder Crashes happen
22:45 OldCoder Zzz <- napping ttyl
22:45 red-001 when I try to start a singleplayer game
22:46 red-001 http://s000.tinyupload.com/index.​php?file_id=69600612174384718924
22:46 red-001 link to the dump
22:46 kaadmy OldCoder: tried on mobile, doesn't have any problems
22:49 kaadmy what version of mt causes those problems oldcoder?
22:50 OldCoder kaadmy, PM; there is insufficient evidence for group discussion. But all versions. I will research and report evidence if such turns up.
22:50 * OldCoder can narrow it down to either the hardware or the OS and will do so
23:10 red-001 joined #minetest-dev
23:13 cmdskp joined #minetest-dev
23:20 cmdskp @sapier: I found today that entities aren't affected by slippery(only players) and items needed some work from the original PR (which I hadn't included the lua for handling items).  I think it's best to leave other entities to handle slipping themselves, but support for built-in items to support slippery when used by mods on nodes would seem reasonable?
23:21 cmdskp so entities will remain unaffected
23:21 sapier aren't items just entities?
23:21 red-001 and mobs
23:21 red-001 i think
23:22 gregorycu joined #minetest-dev
23:22 cmdskp it appears, items are handled with a specific section in the lua code inside the builtin folder
23:22 cmdskp which handles stopping them when you throw them away
23:23 cmdskp that's the bit that slippery adds to, if its set on a node
23:23 cmdskp the specific file is: builtin\game\item_entity.lua :)
23:25 cmdskp though, as I say, I didn't include this with my merge, I spotted it in the original PR and fixed it today to respect an amount of slip or stop, as by default
23:25 cmdskp had visitors round ^_^ so haven't had much time
23:28 sapier is there any way lua code runs in parallel in current minetest?
23:29 VanessaE sapier: not as yet.
23:29 VanessaE except for that async stuff once used in the main menu I think
23:29 sapier strange because I see a bug which could be explained best by some undesired concurrency
23:31 sapier considering minetest crashes completely I'd not surprised if there is something unwanted in it
23:35 sapier wow :-) that's curious
23:36 sapier best explanation I've got is lua garbage collector is removing a function deleted from a table while it's still in use
23:37 cmdskp now that's a little bit serious, to say the least!
23:38 VanessaE heh
23:38 sapier I have to think more about this that'd be way to bad to not being discovered by now
23:38 sapier yet there's a bug in minetest too I don't get a callstack in gui just a grey screen
23:40 cmdskp Could anyone recommend a nice Lua editor/syntax checker (preferably for Windows)?
23:40 red-001 notepad c++ is nice
23:41 red-001 not really a  lua editor
23:41 kaadmy https://www.gnu.org/software/emacs/
23:41 sapier eclipse ldt ... there's even debugger support ... but there's some regression in latest version (at least on linux) but eclipse usually is way less stable on limux
23:41 sapier linux
23:41 Calinou http://atom.io :)
23:41 red-001 kaadmy you mean echo?
23:41 Calinou there's a Lua linter available as a package
23:42 cmdskp "else if" catches me way too often with a complete crash out... =)  I'd forgotten about Eclipse IDE having Lua support potential!
23:42 red-001 https://notepad-plus-plus.org/
23:42 cmdskp ty
23:42 sapier well eclipse is a little bit overkill true but if you're doing c++ like me you're used to it anyway
23:43 red-001 or vs
23:43 red-001 Visual studio
23:43 sapier well for vs you need to have an microsoft account
23:43 sapier but if you've got win10 you probably have one anyway
23:43 sapier including security copy of your hard drive crypto key at nsa servers :-)
23:44 Calinou you can use Windows 10 with local account
23:44 cmdskp atom's interesting looking - net heard of it before.  But I'm fairly comfortable to Eclipse too for Java - choices choice! :D  I do like smaller editors for scripting...I'll try each!
23:44 red-001 It's not exactly hard to make a Microsoft account
23:44 cmdskp I'm currently using Visual Studio for Minetest compiling
23:44 sapier Calinou: you can use android without google account too :-)
23:44 Calinou Atom suits any language really
23:44 kaadmy trying atom right now ;)
23:44 Calinou it shines in Web development, but can do C++, Java, Lua, Rust… just fine
23:44 Calinou it's a bit slow to load, and uses quite a bit of RAM, you are warned
23:44 sapier imho win10 is for pc what android is for mobiles
23:44 kaadmy and i have yet to try using Rust
23:44 Calinou but if you edit files < 100 KB, speed should be good
23:45 paramat joined #minetest-dev
23:45 Calinou sapier, Windows 10 is pretty usable with local account, unlike Android
23:45 kaadmy but imo terminal-based text editors are the best
23:45 cmdskp awe, shame Atom is for Win7 and later - I have Win7 but my video drivers are not very stable compared to XP (ironic that! =D)
23:45 Calinou there is Slap for the command-line: https://github.com/slap-editor/slap
23:46 red-001 cmdskp
23:46 red-001 notepad++ runs in xp iirc
23:46 gregorycu paramat
23:46 gregorycu So good to see you
23:46 cmdskp does it offer syntax checking?  Not just colouring?
23:46 red-001 hi paramat
23:46 gregorycu He's mine, leave him alone
23:46 cmdskp I would like to catch some of my little idiosyncracies when switching from C++ to Lua =D
23:47 red-001 what do you mean?
23:47 Calinou no, out of the box, Notepad++ does only colouring
23:47 Calinou I doubt there's a plugin for Lua linting
23:47 paramat Fixer, yes permanent fire has no sound that's intentional
23:47 red-001 https://github.com/minetest/minetest_game/pull/667
23:47 red-001 https://github.com/minetest/minetest_game/pull/602
23:47 paramat heh
23:47 Fixer paramat, ok
23:47 kaadmy cmdskp: just build from source, that might work.
23:48 kaadmy <offtopic> does anybody here play DooM?
23:49 Calinou try the non-dev channel, kaadmy
23:49 Calinou you're in -dev :p
23:49 sapier Calinou: I use android without google account ;-) But once I did look at adb debug output what this os does even do without account ... terrible ... windows doesn't even provide a way to see it :-)
23:49 paramat apologies to all, mtgame devs are very elusive at the moment, and i am super busy
23:52 gregorycu cmdskp: Why did you branch from my repo, out of interest, as opposed to the master branch
23:52 paramat will merge https://github.com/minetest/minetest/pull/3499 'Minimal: Add mapgen alias for air' in a moment
23:53 cmdskp gregorycu: I'm not sure what pulled my interest - but I had troubles after trying to branch to the original Minetest github, the site seemed to keep redirecting me to the fork I had made from yours.
23:54 gregorycu Interesting
23:54 cmdskp I'm sure it was something I had found a couple weeks ago from a mod in the forum or perhaps via google's results
23:54 cmdskp I'm afraid I'm new to github's wonders
23:54 gregorycu Me too
23:54 gregorycu And lua
23:54 paramat now merging 3499
23:54 gregorycu Well, somewhat new
23:56 cmdskp Lua's interesting, I do question some of the language design, but I think it's quite useful for teaching with, esp. inside Minetest
23:57 red-001 github for windows is just pain awful
23:57 gregorycu I prefer python
23:57 cmdskp python's not something I've used in earnest yet
23:58 paramat merged
23:58 cmdskp I'm more a Java/C++(and way long ago, Assembly Language) guy ^_^  But I have a fond spot for BASICs and Lua reminds me of them a bit.
23:58 gregorycu Java lol
23:58 gregorycu I'm proficient in C++, C#, and python
23:59 Calinou Python would not be suitable for a game like Minetest, where the KISS principle is still quite important
23:59 paramat will merge game 763 in a moment 'Default: Add mapgen alias for air' again

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