Minetest logo

IRC log for #minetest-dev, 2012-12-11

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

All times shown according to UTC.

Time Nick Message
01:02 SitDog joined #minetest-dev
01:43 doserj left #minetest-dev
05:26 SitDog anyone having trouble getting to sourceforge.net?
06:32 cy1 joined #minetest-dev
06:32 cy1 I gave a shot at adding IPv6 support.
06:33 cy1 Doesn't work, but I'd appreciate anyone who could look at it see if you can figure out what's wrong.
06:33 cy1 https://github.com/cyisfor/minetest/commit/f96ede0420a1ba697686fe2f2339ea73658f5652
06:33 cy1 l8r
09:57 rsiska joined #minetest-dev
12:04 celeron55 lol wtf is void setCPlusPlusSucks(Address address)
12:06 celeron55 also, the change to Address::serializeString() to output the port will break bans
12:07 celeron55 those are what i can see by quickly looking at the code
13:14 Taoki joined #minetest-dev
14:27 PilzAdam joined #minetest-dev
14:57 rsiska joined #minetest-dev
15:22 thexyz can anything bad happen if i https://gist.github.com/3358184e1824cfe7db7e ?
15:25 celeron55 can, because the object system is still buggy and can multiplicate objects
15:25 celeron55 remember the DM times? 8)
15:26 nyuszika7h joined #minetest-dev
15:26 doserj joined #minetest-dev
15:27 thexyz so it's not safe to rely on entity existance (like "painting" mod does)
15:32 celeron55 not really, until it is fixed
15:32 nyuszika7h joined #minetest-dev
15:33 celeron55 altough, i wouldn't say it is safe to rely on minetest on anything 8)
15:36 PilzAdam I think we can merge this: https://github.com/celeron55/minetest/pull/345
15:40 RealBadAngel joined #minetest-dev
15:40 RealBadAngel celeron55, got a question
15:40 RealBadAngel about LUA entities
15:41 celeron55 PilzAdam: if it works, i guess so
15:41 RealBadAngel they suffer lags, heavy
15:41 RealBadAngel any suggestion where i should start to get rid of the lags?
15:41 celeron55 are you talking about the lags that map generation in lua causes?
15:42 RealBadAngel not only
15:42 RealBadAngel theyre seems  to be out of main draw loop
15:42 RealBadAngel entity going one direction, sometimes seen to flew away
15:43 PilzAdam celeron55, ok, i merge it
15:43 RealBadAngel and a few seconds after it wakes up
15:43 RealBadAngel i mean control of it
15:43 celeron55 i don't think there are much else that would make them laggy, except bad network conditions
15:43 RealBadAngel it happens also on local
15:43 celeron55 local has map generation too, and mods can jam the server at other times too
15:43 RealBadAngel even spawnin of them is delayed sometimes
15:44 celeron55 all the time that code is executing in lua, nothing is updated
15:44 RealBadAngel this is crucial issue
15:44 celeron55 (that is how event-driven systems work)
15:44 RealBadAngel because of it everything that moves suffer
15:45 RealBadAngel any suggestions?
15:45 celeron55 i request you to do test without additional mods, determining if there are any other cases than map generation
15:45 celeron55 -do
15:45 RealBadAngel ok
15:45 RealBadAngel i will try
15:46 RealBadAngel i was thinkin bout it a bit, and thought somethin like we use with 3d players could be of help
15:46 RealBadAngel let clients do the calculations of the movement
15:47 RealBadAngel and report postition to the server?
15:48 celeron55 which client does the server trust? 8)
15:48 doserj the best you could do is to make client-side predictions, and update when you get the real data from the server
15:48 celeron55 the way it works (and will work) is that the server updates the clients with authoritative information, and the clients figure out stuff at times when the server says nothing
15:49 celeron55 the thing is, the client has to know what to do at those times
15:49 celeron55 currently, for lua entities, it just applies the basic physics
15:49 RealBadAngel but thanks to it we have chopping animations, carts going straight to nowhere
15:49 RealBadAngel all the animations suffer
15:50 celeron55 17:46:19 < celeron55> i request you to test without additional mods, determining if there are any other cases than map generation
15:50 RealBadAngel server is not obliged to do all the calculations
15:50 RealBadAngel cloud idea is present and alive
15:51 celeron55 good luck with that one 8)
15:51 RealBadAngel i havent said is an easy piece of code :)
15:51 RealBadAngel but lettin server do all the job, just fails
15:52 RealBadAngel look, by now we do have it calculated server side
15:52 RealBadAngel so if an update occurs
15:53 RealBadAngel client suffers, visually
15:53 celeron55 now you're being just unrealistic
15:53 RealBadAngel if it could be made client side, we can have smooth animations
15:53 celeron55 the problem is not in the central server model, it's in the bad implementation of it
15:54 RealBadAngel and updates to the server data, not resultin in graphics glithches
15:54 thexyz dafuq am i reading?
15:54 celeron55 also adding more knowledge to the client is doable too
15:54 celeron55 but it's just an addition
15:54 thexyz celeron55: anything else about my pull request?
15:54 RealBadAngel celeron55, Jeija implemented client side LUA
15:55 RealBadAngel maybe some stuff should be sent by the server
15:55 celeron55 i've implemented client side lua before server side lua, that's nothing new
15:55 RealBadAngel but executed locally
15:55 RealBadAngel i believe, but this could solve trust problem
15:56 celeron55 is there a documentation of the client-side API made by jeija somewhere?
15:56 RealBadAngel client side executin only server sent contens
15:57 RealBadAngel dunno, just read a topic bout it with proof of concept
15:57 RealBadAngel kinda hello world
15:59 RealBadAngel i think limiting client side LUA to server sent scripts could work
16:00 RealBadAngel no trust problems, most heavy calculations passed to the client
16:00 PilzAdam this seems to be rather trivial: https://github.com/celeron55/minetest/pull/348
16:01 RealBadAngel and most vital ones, like animations done locally
16:01 RealBadAngel not to mention such trivial ones like music
16:02 celeron55 "no trust problems" ...
16:02 celeron55 it needs to be heavily sandboxed
16:02 celeron55 otherwise the server will just send rm -rf / and all is gone
16:02 RealBadAngel just dont allow execute anythin which is not from server
16:03 celeron55 happily lua allows to make a very strict sandbox; it just disables some metaprogramming stuff like metatables
16:03 celeron55 but if the server sends code, there are no other options
16:03 RealBadAngel we can build better sandbox
16:04 RealBadAngel server sends code, client periodically responds
16:04 RealBadAngel if client doesnt have rights to respond
16:05 RealBadAngel server rejects and shuts the client
16:05 celeron55 code sent by server will be media just like images and sounds currently are
16:05 RealBadAngel sure
16:05 RealBadAngel but the client is logged in
16:05 RealBadAngel and can be given rights
16:05 celeron55 ...wat.
16:06 celeron55 stop talking about things you clearly don't understand anything about
16:06 RealBadAngel maybe i dont
16:07 celeron55 or alternatively try to form full understandable sentences
16:07 RealBadAngel im thinkin out loud now
16:07 celeron55 well, okay; i'll not be reading it then
16:07 RealBadAngel :)
16:08 celeron55 somebody note me when it's over so i can enable activity indicators for this channel again...
16:09 PilzAdam RealBadAngel, why not PM yourself? :-p
16:09 RealBadAngel oh, cmon. ok, just stopped
16:14 RealBadAngel i will just raise one question, about formspec fields. you were against changing it to be non-exit on enter pressing
16:15 RealBadAngel then please explain why there are two methods of fields
16:15 RealBadAngel one full, with coords etc
16:15 RealBadAngel and one shortened with automatic added "proceed" button
16:17 RealBadAngel imho the last one was thought to be fast exit one
16:17 RealBadAngel not the full one
16:18 celeron55 i don't care about anything else than that existing mods will work as expected by their creators by reading the existing lua_api.txt
16:19 RealBadAngel nothin will be broken with the change, signs will work as before
16:19 PilzAdam +1 for change it
16:20 PilzAdam I dont know any mod that uses the fields with coords and expect them to close the formspec
16:21 RealBadAngel search phrases are worst
16:21 RealBadAngel user types, press enter, then nothin ;)
16:22 RealBadAngel there was also another mod with multiple lines sign
16:22 RealBadAngel user had to switch from field to field with care
16:23 RealBadAngel instead of givin focus on enter to next field for example
16:23 PilzAdam I cant imagine any case where the current way is usefull
16:23 RealBadAngel single line sign
16:24 RealBadAngel enter and gone, that was the only instace before so thats why propably it was done so
16:25 RealBadAngel but even if so, why adding a proceed button? :)
16:26 celeron55 there are many reasons
16:26 celeron55 what if somebody opens it without having a keyboard?
16:26 celeron55 8)
16:26 celeron55 or enter
16:26 RealBadAngel yup, and wrote the contenst usin telekinesis
16:27 celeron55 also it is good to have the thing visible as a button that will happen if enter is pressed on the field
16:27 RealBadAngel ok, i agree
16:27 celeron55 (if it's not completely obvious... might be in that case though)
16:27 PilzAdam but thats only for fields without coords
16:28 RealBadAngel but still, that instance is ok, its shortened for such occasions
16:28 RealBadAngel but theres also a full one
16:28 RealBadAngel which is treated the very same way
16:28 RealBadAngel i asked to change this one
16:29 RealBadAngel so signs etc wont be affected
16:30 RealBadAngel exactly, you just admited the one with coords is bugged :)
16:30 RealBadAngel or just not finished rather
16:48 Calinou joined #minetest-dev
16:57 hmmmm joined #minetest-dev
16:57 sstrandberg joined #minetest-dev
17:26 hmmmm joined #minetest-dev
17:34 ssstrandberg joined #minetest-dev
18:07 PilzAdam an addition to the noclip commit: https://github.com/celeron55/minetest/pull/350
18:43 ssstrandberg joined #minetest-dev
19:54 Penguin joined #minetest-dev
19:55 Penguin left #minetest-dev
19:58 rubenwardy joined #minetest-dev
20:01 rubenwardy left #minetest-dev
20:12 SpeedProg joined #minetest-dev
21:07 sema4 joined #minetest-dev
21:20 ssstrandberg left #minetest-dev
21:36 VanessaE joined #minetest-dev
21:47 sema4 joined #minetest-dev
21:50 serengeor joined #minetest-dev
23:13 rsiska joined #minetest-dev
23:23 celeron55 joined #minetest-dev
23:26 rsiska joined #minetest-dev

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