Minetest logo

IRC log for #minetest-dev, 2012-08-23

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

All times shown according to UTC.

Time Nick Message
01:04 hmmmm why was 0.05 chosen for the default dedicated_server_step?
01:11 hmmmm what are the packets with lengths 42 and 61?
01:30 hmmmm nevermind that... my minetest experience is immensely better with dedicated_server_step increased to 0.35... there are still lots of problems with sluggishness
01:30 hmmmm celeron, i think that you underestimate how much overhead network traffic has
01:31 hmmmm i think that each message should be more of an event
01:31 hmmmm and the method to transfer that event is chosen based on the location of the client in relation to the server
01:32 hmmmm that is, if it's localhost, bypass the network layer completely
01:32 hmmmm this is how starcraft works pretty much anyway :-)
05:06 VanessaE_ joined #minetest-dev
06:26 jin_xi joined #minetest-dev
06:31 EdB joined #minetest-dev
08:57 EdB joined #minetest-dev
10:14 PilzAdam joined #minetest-dev
11:47 Calinou joined #minetest-dev
13:33 cornernote joined #minetest-dev
13:33 cornernote hey, i am trying to use minetest.set_player_privs(), but it doesnt seem to save
13:34 cornernote i checked all the other mods i could see, and none use that
13:34 cornernote think its a bug ?
13:35 cornernote do i need to call auth_reload() after ?
13:50 cornernote yep ,that worked !  =)
14:19 ArcticCoolMan joined #minetest-dev
14:21 ArcticCoolMan Minetest is cool but i think making mobs three-dimensional models with hitbox shouldn't take much time. By the way block textures in inventory are misplaced eg. mese looks like leaves.
14:21 ArcticCoolMan left #minetest-dev
14:54 Calinou joined #minetest-dev
16:09 VanessaE_ celeron55: you around?
16:10 celeron55 kinda
16:10 VanessaE_ I'm about to file an issue, but I wanted to ask you about it first.
16:11 VanessaE_ recall our discussion about image formats and what irrlicht/game can load..  and in the API where it says how this works (line 142)
16:12 VanessaE_ Is there a known situation, i.e. some corner case, where this is not supposed to work?
16:12 celeron55 in terms of what you can do from the lua side of things?
16:12 VanessaE_ yeah
16:13 celeron55 not that i am aware of
16:13 VanessaE_ ok
16:13 celeron55 i don't probably remember everything though; could be
16:15 VanessaE_ well the short version:  if you don't give a filename extension in your mod's tiles={} declaration (or others probably), or the extension is wrong, the game doesn't load the image.
16:16 celeron55 where DOES it work then?
16:16 VanessaE_ so tiles={"mymod_somefile.jpg"} works, but tiles={"mymod_somefile"} does not
16:16 VanessaE_ but, in a texture pack it *does* work
16:16 VanessaE_ if the mod specifies "mymod_somefile.png" and I provide "mymod_somefile.jpg" in the texture pack, the game *does* load the jpg.
16:17 celeron55 i know that to work always
16:17 VanessaE_ yes, that's what I though.
16:17 VanessaE_ thought.
16:17 VanessaE_ but within the confines of the mod itself, if you specify a wrong filename, even if the API says you're right, you're still wrong :D
16:19 celeron55 is there a place where it works without an extension?
16:19 VanessaE_ I haven't tried it out extensively yet
16:19 celeron55 it kind of should, but i have never tested it
16:19 VanessaE_ just some initial fiddling to try to fix a bug in the mod
16:20 celeron55 but as it tends to be, nothing works if it isn't tested 8)
16:21 VanessaE_ of course
16:21 VanessaE_ but then again, I'm not sure what else there is to test outside of images specified in the tiles, wield_image, and inventory_image lines
16:22 VanessaE_ (and the aforementioned texture pack of course)
16:22 PilzAdam it dont work for inventory_image and textures of tiles
16:22 celeron55 then there is the question why not
16:22 PilzAdam *entities
16:24 VanessaE_ shall I go ahead and file this issue for later reference?
16:27 celeron55 i am actually investigating this now
16:27 VanessaE_ ok
16:28 celeron55 this is one of those things that appear to work even with some debugging, but at the end it just doesn't end up like it should
16:29 VanessaE_ like lua with it's fucking commas ;)
16:30 VanessaE_ "strange, this should work.  the code at line 3846 is right..  OH, missed a comma on line 345."
16:30 celeron55 that's how you pay for flexibility and dynamism!
16:30 VanessaE_ haha
16:31 celeron55 if you just replaced code at 3846, it could end up as completely valid but doing something different 8)
16:33 VanessaE_ [offtopic]you seen my latest abuse of nodeboxes by the way? :-)[/offtopic]
16:35 celeron55 [offtopic]no, and i don't care for now[/offtopic]
16:35 VanessaE_ heh ok
16:36 celeron55 i found the problem, and it is kind of odd
16:36 VanessaE_ maybe something to leave for later then
16:36 celeron55 the checking of different file extensions is done at the filesystem level, not at cache level (in which the stuff from the server ends up)
16:37 VanessaE_ ahhh
17:30 celeron55 VanessaE_: file the issue, i'm not going to fix it now
17:30 VanessaE_ ok
17:30 VanessaE_ filed.
17:30 celeron55 put that last line i said there
17:31 VanessaE_ done :-)
17:32 VanessaE_ I had a feeling it would be a bit non-trivial, but since there's a workaround, it doesn't need addressed now
18:23 VanessaE_ celeron55: I suppose issue #207 can be closed now btw.
18:24 VanessaE_ or not; that patch you had us try isn't in upstream yet :-)
18:38 celeron55 joined #minetest-dev
18:47 celeron55_ joined #minetest-dev
18:55 sapier joined #minetest-dev
18:58 sapier Hello, is minetest supposed to handle 39+ active objects at once at a (single core) 3.2GHz Phenom II? I ask because I seem to be in a situation where emerge thread is locked out by onStep throwing lots of exceptions because of unloaded blocks. If I understand emerge threads purpose it shall load those blocks that cause the exceptions.
19:07 celeron55 joined #minetest-dev
19:36 hmmmmmm joined #minetest-dev
19:51 ironzorg gravity should really apply to apples as well
19:51 ironzorg I don't like the way they float in the air once leaf decay did its job
20:51 EdB joined #minetest-dev
21:06 sapier ok if someone is interested I've an answer to my question. Minetest ain't capable of handle collisions of 39 active objects moving around in unloaded areas. The exception handling triggered is far to expensive.
21:06 sapier left #minetest-dev
21:12 celeron55 ehm
21:18 saschaheylik joined #minetest-dev
21:34 Octupus joined #minetest-dev
21:42 celeron55 joined #minetest-dev
22:13 hmmmmmm celeron?
22:13 hmmmmmm you never answered my question from yesterday, why is the default server step time so low?
22:32 saschaheylik hmmmmmm: why do you think it should be higher?
22:59 saschaheylik_ joined #minetest-dev
23:24 VanessaE joined #minetest-dev
23:37 hmmmmmm because at 50 milliseconds, it consumes too much cpu time
23:37 hmmmmmm in fact, there's a constant 35-40% cpu time being eaten up in kernel mode
23:38 hmmmmmm which is reduced to more like 2-3% if the server step time is increased to 350ms
23:38 hmmmmmm but the choice of 50ms seems arbitrary and pointless, unless there was a specific reason for such a low value...

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