IRC log for #minetest-dev, 2013-02-20

Time	Nick	Message
02:58		ShadowNinja joined #minetest-dev
03:04		exceed joined #minetest-dev
05:27		ffoxin joined #minetest-dev
09:28		Calinou joined #minetest-dev
10:10		jin_xi joined #minetest-dev
11:19		rubenwardy joined #minetest-dev
11:30		darkrose joined #minetest-dev
11:30		darkrose joined #minetest-dev
11:52	proller	/announce
11:52	proller	ups sorry
12:18		troller joined #minetest-dev
12:26		proller joined #minetest-dev
13:40		hmmmm joined #minetest-dev
13:50	* rubenwardy	has no idea how to make the grey bar at the top: http://ompldr.org/vaGl2ZA/mtext_mockup1.png
13:57	rubenwardy	(ping: celeron55)
14:09	thexyz	looks really weird
14:10	rubenwardy	the mock up?
14:10	rubenwardy	this is what I have done so far: http://multa.bugs3.com/minetest/forum/
14:10	thexyz	grey doesn't fit there
14:11	rubenwardy	no it does not
14:12	rubenwardy	do you prefer http://multa.bugs3.com/minetest/forum/ or http://ompldr.org/vaGl2ZA/mtext_mockup1.png
14:13	rubenwardy	(thexyz)
14:21		proller joined #minetest-dev
14:25	thexyz	why not add another bar under the logo?
14:25	thexyz	not inside it
14:26	thexyz	I'm not sure whether this conversation is appropriate for -dev
14:26	celeron55	the gradients of current mt-ext look horrible at least
14:26	celeron55	and boxes everywhere
14:27	celeron55	i at least approve website development in here
14:29	rubenwardy	Should I delete the gradients, of make them softer?
14:29		proller joined #minetest-dev
14:29	celeron55	delete
14:30	celeron55	or, well
14:30	celeron55	i don't think rubenwardy can ever design anything out of his head that would aesthetically please me at all
14:30	celeron55	so i don't really know what to do with these discussions :P
14:30	rubenwardy	gradients removed
14:31	celeron55	i'm a bad artist too, but "trying everything" sometimes tends to gain sufficient results...
14:32	celeron55	where are all the web designers
14:32	celeron55	i guess they don't play sandbox games with bad graphics
14:34	rubenwardy	what about this?
14:35	rubenwardy	ah
14:35	rubenwardy	working
14:35	celeron55	let me quote the AVGN Game Glitches episode: "Not gonna work. Not gonna work. What time is it? Not gonna work o'clock!"
14:36	rubenwardy	I have aligned to right
14:37		proller joined #minetest-dev
14:40	rubenwardy	(2:30:12 PM) celeron55: i don't think rubenwardy can...
14:40	rubenwardy	I doesn't help that you are fairly "fussy" about what you want
14:44	rubenwardy	The reason I made the side bar, is because it is an effective way to display data on this page: http://multa.bugs3.com/minetest/forum/viewmod.php?id=2
14:45	jin_xi	rubenwardy: when scrolling down the carousel thing goes over the top bar
14:45	rubenwardy	yeah, I know
14:45	thexyz	oh, well
14:46	thexyz	you deny everything designers do
14:46	rubenwardy	me?
14:46	VanessaE	who?
14:47	thexyz	c55 does
14:47	thexyz	http://irc.minetest.ru/minetest-dev/2013-02-19#i_2882565
14:48	rubenwardy	You replied to the jungle topic saying: http://forum.minetest.net/viewtopic.php?pid=71506#p71506
14:48	rubenwardy	and this explains it: http://forum.minetest.net/viewtopic.php?pid=71562#p71562
14:48	rubenwardy	(celeron55)
14:49	thexyz	they weren't removed just because c55 thought they were boring
14:49	thexyz	they weren't readded because of that
14:50	rubenwardy	yeah, well I was told that. Sorry.
14:52	rubenwardy	thexyz: should I keep the second bar like celeron55 suggested: http://multa.bugs3.com/minetest/forum/index.php
14:52	rubenwardy	or should I make it a new full width bar like you said?
14:52	rubenwardy	(2:25:04 PM) thexyz: why not add another bar under the logo?
14:52	thexyz	no idea
14:52	thexyz	but as for now it looks like crap
14:53	rubenwardy	celeron55: I am open minded, and I will implement your design to see what it looks like
14:55	rubenwardy	I think i have pissed c55 off
14:56	VanessaE	celeron55: regarding people re-implementing jungles...well, you can look at the results of moretrees. The jungle biome settings aren't right, but well there they are.
14:56	VanessaE	(I know you don't like bas080's mod)
15:12	VanessaE	it would be trivial to adjust the biome settings to get the old jungles back via moretrees, but that would probably crowd out the other trees.
15:32		iqualfragile joined #minetest-dev
15:32	rubenwardy	Is this what you had in mind, celeron55? http://multa.bugs3.com/minetest/forum/index.php
15:34		PilzAdam joined #minetest-dev
15:40	rubenwardy	celeron55...
15:40	rubenwardy	someone repost this, i think i am on ignore
15:43	rubenwardy	(VanessaE)
15:44	VanessaE	oh sure, and get him doubly pissed off at me? :-)
15:44	rubenwardy	nm then
15:45	VanessaE	besides, he reads the logs, too
15:45	celeron55	eh what? i am not allowed to be away?
15:45	celeron55	...
15:46	celeron55	i think i need to be more away then
15:46		PotatoBrain joined #minetest-dev
15:46	VanessaE	celeron55: your timing just sucks :)
15:55	proller	celeron55, https://github.com/celeron55/minetest/pull/492
15:58	celeron55	rubenwardy: just do what you can and listen to other people; i really can't give you any magic opinions or solutions
15:59	rubenwardy	ok
15:59	rubenwardy	thanks
16:00	celeron55	now does somebody find issues in proller's pull request?
16:01	celeron55	i guess the issue of who should be responsible for running the auto-announceable servers.minetest.net should be tacked down
16:02	celeron55	proller: if you're left with the job for now, how long do you expect to stick around providing it?
16:02	VanessaE	why aren't we just piping this all through thexyz's server monitoring thing?
16:02	celeron55	or should it be transferred to thexyz or so
16:02	VanessaE	this seems like some duplication of effort here
16:03	PilzAdam	VanessaE, proller's pull requests makes the things that are done by thexyz currently automatically
16:03	celeron55	i guess thexyz could implement auto-announcements to his service, but i can't force that
16:03	VanessaE	PilzAdam: right, I figured that much
16:03	proller	celeron55, as necessary
16:03	VanessaE	but I mean if thexyz is already running a server that does this job, why should we need a second similar service? could not both functions be combined into one?
16:04	VanessaE	(assuming he is up to the task of adding the extra hooks/data storage that would be needed)
16:04	F00	Monitoring page gripe: Why is time little-endian, and why not do ping in ms?
16:05	VanessaE	F00: because people think in "it's up x% of the time", not "it took xxxx ms to respond last time it was pinged"?
16:05	VanessaE	think of the MT userbase, most folks don't seem to be technically-inclined enough to understand a ping response, specifically whether a given time is good or not.
16:06	F00	Er, it says "ping 0.160" for the sole listed server, that's not uptime
16:06	F00	uptime is next to that
16:06	celeron55	ping response is irrelevant because it's just relative to the master server
16:06	VanessaE	F00: what list are you looking at?
16:06	VanessaE	http://servers.minetest.ru/
16:06	VanessaE	I see no ping times here.
16:07	F00	servers.minetest.net, the default URL used in proller's current pull request
16:07	VanessaE	oh, I'm looking at the one everyone is already using
16:07	proller	F00, what is "time little-endian"
16:07	F00	"minutes hours days" rather than "days hours minutes"
16:08	proller	http://servers.minetest.net/
16:08	F00	Also if you're doing fractional hours it seems weird to also do fractional days
16:08	VanessaE	proller: he means you should write "1.2d, 5.5h", not "5.5h 1.2d"
16:08	VanessaE	(the fractions aside)
16:08	proller	there is 2 times
16:08	proller	last update and start time
16:08	F00	1.5d 12h == 2 days; 1d 12h = 1.5 days
16:08	VanessaE	oh right
16:09	VanessaE	then put a slash between them or use separate columns
16:09	F00	Oh. That's two times?
16:09	proller	and! server page is like beta and there is what to do
16:09	F00	Yeah, confusing
16:09	VanessaE	I read "updated/started" as if they're supposed to be synonyms.
16:09	F00	Aye
16:09	proller	slash, ok
16:10	proller	and better to delete update time
16:10	proller	its useful only for debugging
16:10		Calinou joined #minetest-dev
16:11	celeron55	i'll suggest waiting thexyz's input on this
16:11	VanessaE	actually it might be useful for others to see "hrm, this server does not respond... oh, it hasn't announced in several weeks now."
16:11	celeron55	umm what
16:11	F00	If they haven't done anything in 2*default announce delay or something, I'd just delist them
16:12	celeron55	there should be two columns, one for "updated", one for "started"
16:12	celeron55	what kind of nonsense is that what it has now
16:12	celeron55	8D
16:12	VanessaE	celeron55: you forgot to call it "random" or "horrible" :-)
16:12	proller	VanessaE, server hides after 900s
16:12	proller	of inactivity
16:12	F00	That seems adequate
16:12	VanessaE	900s? that's a strange amount of time?
16:13	celeron55	it's a strange amount of time also called 15 minutes
16:13	proller	no time_alive => 650,
16:13	proller	its 2 lost announces
16:13	proller	announces every 300s or 5 min
16:14	VanessaE	once an hour or less often is usually considered more appropriate, fyi
16:14	F00	Er, for game servers? Getting a semi-accurate player count is useful, and that can change drastically +/- 60 minutes.
16:14	VanessaE	consider the recipient of these announcements, which might perhaps be getting hit hundreds of times every few minutes. this will drive up bandwidth usage, which may be undesirable.
16:15	* Calinou	is tempted to tell people to do a master server the sauerbraten way
16:15	Calinou	check whether the server is up or not every 60 mins and calculate ping/players in real time or so
16:15	VanessaE	then update only when a player joins/leaves, there's no reason for such a frequent "keepalive" announcement
16:15	F00	That can potentially be much more often than every 5 minutes
16:16	celeron55	so let's make some realistic calculations, shall we?
16:16	F00	People have texture issues, get stuck on someone on join, sometimes end up joining and leaving 3-4 times
16:16	celeron55	how large is one announcement, on average?
16:16	VanessaE	ok then take an average - if the last announcement was less than 5 minutes ago, withhold it. Otherwise announce it, but don't do periodic "just because I am here" announcements at anything less than 1 hour intervals.
16:17	celeron55	i assume one announcement is about 200 bytes
16:17	celeron55	and i think we can expect an active server count of roughly 100 in the near future
16:17	F00	I see frequent announces as useful because an alarming number of servers are run by people on their home computers and are up/down like crazy
16:17	proller	celeron55, may be 1000bytes
16:18	celeron55	proller: what is the average?
16:19	celeron55	1000 bytes is quite a chunk of json, i wouldn't guess it would be that much
16:19	F00	Even if it's a kilobyte, times a hundred servers every five minutes, that's less than a gigabyte per month.
16:21	celeron55	kilobyte once in 5 minutes for 100 servers for a month makes... 860 megabytes per month
16:21	Calinou	every 15 minutes then?
16:21	celeron55	and that's a very maxed out estimate
16:21	Calinou	3Ã— less megabytes :P
16:21	F00	servers.minetest.net is hosted by... Hetzner?
16:21	F00	VPS?
16:22	PilzAdam	thexyz said minetest.ru has a limit of 10 TB per month
16:22	F00	It looks like Hetzner's bottom-end VPSes include a TB of traffic a month. Their dedicated servers are unmetered 100 Mbps
16:23	celeron55	a gigabyte is pretty close to nothing in terms of hosting
16:23	F00	Aye
16:25	celeron55	we can tune it in the future though - most servers will run on default config anyway
16:25	F00	Also, some people may restart their servers via cronjob or something
16:26	F00	Perhaps it would be good to add a random initial announce delay
16:26	F00	To avoid clumps of servers announcing at roughly the same time
16:26	proller	celeron55, announce size can be reduced to 200-400 bytes
16:27	F00	proller: Why set counter to 0.01 instead of just 0 when you're adding dtime to it afterwards?
16:27	proller	for first announce
16:27	proller	if !counter -> first
16:28	F00	Yes, but it would go from 0 to non-zero because dtime should never be zero unless the game is running faster than float precision
16:28	F00	Either way, it's very minor
16:28	F00	I just dislike random constants
16:28	celeron55	i'd use -1 for a special value of an otherwise positive float
16:29	celeron55	but anyway, i don't care
16:36	F00	Also, as far as traffic on the master server goes, the server announces will likely be greatly outpaced by clients requesting info from it
16:47	proller	celeron55, if all counters will be inited to 0.0 and one counter to -1 -> there will be a question
16:47	F00	proller: There is one thing I'm concerned about, namely unscrupulous server owners
16:48	F00	Inflating player count to look more active, etc.
16:48	proller	server script support blacklist
16:49	F00	So, just play whack-a-mole and hope the bad server has a static IP?
16:49	proller	its possible to ban by regexp
16:49	proller	all net
16:50	proller	and ipv4 now more and more valuable
16:51	celeron55	actually, that's a real concern, now that you mentioned
16:53	celeron55	how does openttd do it?
16:53	celeron55	i can't even see a "report server" button
16:53	celeron55	i guess people are just nice
16:54	F00	I play a bit of TF2 and there was kind of a scumbag arms race for a while
16:55	F00	Just takes one "entreprenurial" jerk and then you've got dozens of bad servers faking player counts so they can serve ads to people via HTML MOTD
16:55	celeron55	i think we can start without anything, and if it ends up too bad, we'll figure out something
16:55	celeron55	rate limiting per ip should be done though
16:56	celeron55	or something like that
16:56	celeron55	so that flooding the list isn't 100% trivial
16:56	F00	Ooh.
16:56	proller	server must reply on his port by udp
16:57	F00	Ah, so when announcing the master server polls back?
16:57	proller	yes
16:57	proller	and show only pingable
16:57	F00	To what extent?
16:57	F00	Wait wait wait. That's not a good idea, necessarily.
16:59	F00	redcrab.suret.net, perhaps the most popular public server, is run on Amazon EC2
16:59	F00	Which, by default, drops ICMP echoes
17:00	celeron55	UDP != ICMP
17:00	F00	When people say 'ping' I assume ICMP echo.
17:14	rubenwardy	Should give version code names: 0.3.1 Dungeon Master
17:14	rubenwardy	like Icecream Sandwich
17:14	rubenwardy	lol
17:32	proller	and https://github.com/celeron55/minetest/pull/482 is good
17:47		jin_xi joined #minetest-dev
19:08		Kray joined #minetest-dev
19:15		sapier1 joined #minetest-dev
19:16	sapier1	celeron55 what do you think about implementing security this way: https://github.com/celeron55/minetest/pull/495
19:17		doserj joined #minetest-dev
19:17	VanessaE	doserj: your change that added the mod manager broke dependency reporting on failure. can you please fix this?
19:18	doserj	VanessaE: https://github.com/celeron55/minetest/pull/500
19:18	VanessaE	hah!
19:18	VanessaE	ok :-)
19:18	doserj	:P
19:19	thexyz	servers.minetest.net is hosted by me ATM
19:19	doserj	VanessaE: If you use the gui, you should never actually have an inconsistent configuration, so I didn't care that much about error reporting...
19:20	VanessaE	doserj: I'm used to doing things manually :-)
19:20	sapier1	<< normaly punches everyone saying "didn't care ... about error reporting"
19:21	doserj	sapier: the error was still reported, but just not in that much detail as before
19:22	sapier	not really better ;)
19:52		deivid joined #minetest-dev
19:59	deivid	hi. I'm trying to make the blocks draw partially based on distance, but I don't know how to do it
20:00	deivid	currently the renderMap gets all the meshes from the block and draws them
20:22		david___ joined #minetest-dev
20:34		troller joined #minetest-dev
21:03		proller__ joined #minetest-dev
21:21	F00	celeron55: Any further thoughts regarding physics?
21:22	F00	I don't like repeatedly dredging it up, but I'm concerned about the various implementation flaws (flying 10 blocks off the top of ladders, taking 3 or 6 damage when fast-moving downwards in water or on ladders, respectively)
21:22	F00	Makes me wonder whether non-falling vertical inertia is worthwhile
21:27	sapier	hmm I've implemented inertia for monorail ... sometimes physicaly correct calculation gives really strange results
21:27	* F00	thinks realism isn't generally a good goal in games
21:28	sapier	true as there are some parameters missing for inertion
21:28	sapier	you'd need friction as well as air resistance too ... both require parameters generally not available within minetest
21:30	F00	Yeah
21:31	F00	A problem I have with vertical inertia on ladders/etc. is that fall speed can't be predicted
21:32	sapier	it's not complicated at all it can even be done in lua but I'm using default values for those parameters ... wich work fine as long as it's a cart moving in air upon tracks
21:32	sapier	but as soon as a cart should move e.g. on dirt, stone .. etc it's getting complicated
21:32	sapier	it can
21:33	sapier	by the way you don't need to do anything special if you change speed by using acceleration
21:33	F00	You can't predict it because of block loading
21:34	sapier	ohh so no physical problem but lack of environment information
21:34	F00	After falling 100 nodes, someone might be falling at a speed so slow as to not incur any fall damage, if the block positioning is convenient
21:34	sapier	no no thats no problem if you add air resistance max speed is automaticaly limited
21:35	F00	I object to terminal velocity for selfish reasons
21:35	sapier	as air resistance results in acceleration against gravity acceleration
21:42	F00	Anyhow, if you had a terminal velocity you'd only be able to estimate the maximum speed for a given fall
21:46	sapier	you need speed on collision... why not read it at that moment?
21:56	F00	That's how fall damage works, but you don't take damage when hitting unloaded block edges
21:56	F00	So somebody may be reset an indeterminate number of times
21:56	sapier	hmm so problem is blocks not being loaded fast enough
22:03	F00	Yeah, which is more or less unsolvable unless the server and client are able to communicate at very high rates with extremely low latency
22:05	sapier	which won't happen ... but damage has to be calculated on server in order to avoid clients cheating so client shouldn't be involved at all
22:10	F00	That would be nasty, though.
22:14	F00	Server-side movement works well in games where the client has the whole map loaded
22:14	F00	It would be goofy to have the client fall off a tall structure, repeatedly get stopped by block loading
22:14	F00	And then instantly die when they finally hit the ground at low speed
22:14	sapier	as far as I know atm it's some sort of dual calculation
22:15	sapier	client predicts and calculates positions while server sends updates to correct clients prediction errors
22:15	sapier	ohhh ok now i understand your problem
22:16	sapier	as long as server would be able to load blocks fast enough player most likely wouldn't hit at low speed ... still player would stop in between several times
22:17	F00	It really depends on where the terrain is inside that final block
22:17		mrtux joined #minetest-dev
22:18	sapier	yes .. i don't see a solution with current minetest architecture
22:20	F00	Yeah, hence why I'm not fond of inertia when climbing/etc.
22:20	F00	Even with exponential (yay!) deceleration, you still wouldn't be able to stop people at a precise block with ladders/water
22:21	sapier	I don't quite understand what inertia has to do with ladders ?
22:23	F00	Prior to the new physics being introduced, ladders stopped you instantly (as did water) with no damage
22:24	F00	Currently, ladders only decel you as fast as you accelerate while falling
22:24	sapier	when you drop on them?
22:24	F00	Yeah.
22:24	sapier	ok now i understand
22:25	F00	So falling 50 nodes = you need 50 nodes of ladders to stop
22:25	sapier	a interesting problem indeed
22:48		sapier left #minetest-dev
23:00	hmmmm	what is the whole thing about jungles
23:05	VanessaE	beats me, I thought I had the concept pretty well covered in moretrees (especially after today's update)
23:10	doserj	F00: I don't have a problem with the player flying up from ladders at the speed they climb them, that is just basic physics at work. But the fast move speed should be reduced on ladders (and in water), that would alleviate the problem a bit. (I think fast move speed is too fast overall, anyways...). I agree that falling speed should be slowed down more on ladders (similar to in water?)
23:21	F00	To me, having fast move climbing equate to probable fall damage when you reach the top is... bad
23:22	VanessaE	agreed.
23:23	F00	I see fast move climbing as... normal climbing, but faster
23:23	F00	I.e. hypothetical hands and legs moving up one rung at a time
23:23	F00	(Even though the climb rate is completely smooth)
23:23	VanessaE	the way I look at it, you can climb as fast as your muscles allow on a real ladder, but no matter how fast you're going, you still have to pull yourself up with eqach rung
23:23	VanessaE	each*
23:23	VanessaE	so you shouldn't overshoot the top by more than a meter or two imho
23:24	VanessaE	right, what you said :D
23:34	F00	I also think that way on the way down.
23:34	VanessaE	well,
23:35	F00	I.e. you stop as soon as you grab a rung
23:35	VanessaE	on the way down from a ladder, you can just slide along the rails
23:35	VanessaE	if you're going fast enough, you have no chance of grabbing a rung - it'll just pull out of your hand from your inertia
23:36	F00	For me that's a collision between convenience and reality where the former wins out
23:37	VanessaE	of course.
23:37	VanessaE	just pointing it out
23:37	F00	The issue with not being able to stop immediately (wihout sneak and a ledge, anyhow, which is another bug) is that you can't stop at a particular place in a tall structure
23:37	F00	Have to hold fast move all the way and wiggle up and down to get where you want
23:37	F00	And it's active, rather than passive
23:37	VanessaE	in the game, if you're falling fast enough, I think it should take a lot of ladders or water to stop you, but not as much as it might take in real life
23:38	VanessaE	20 ladders for terminal velocity seems about right
23:38	VanessaE	(sicne you fall at something like 200 blocks/second at terminal velocity)
23:38	F00	There is no terminal velocity
23:38	VanessaE	yes there is.
23:38	VanessaE	sorta
23:38	VanessaE	thing is the mapgen usually can't keep up at those speeds
23:39	VanessaE	but it maxes out at about that amount
23:39	F00	I'm fairly sure it doesn't unless the new physics added that
23:39	VanessaE	I've tested it before.
23:40	F00	From what I recall, you end up getting so fast that FPS drops to <1
23:40	VanessaE	back when we had the stresstest server at minetest.ru:30001, there was a hole called "Hell's Jump" that went down 7km. Jump down that and well, you get the idea
23:40	F00	With that (collision detection?) vs your CPU speed being the limiting factor
23:40	VanessaE	Once it got past around 3 or 4km deep, your falling speed maxed out
23:42	VanessaE	of course technically, even if your falling speed just keeps increasing, assuming no map loader lag, you still have a maximum speed you can possibly hit, by virtue of the map having a ~62000 node height.
23:45	F00	I think you were observing something else there. The gravity is just: speed.Y -= 9.81 * BS * dtime_part * 2;
23:46	F00	No clamping from what I can see
23:46	VanessaE	interesting
23:47	VanessaE	I wonder where the effect was coming from then
23:51	VanessaE	in that case, it should be clamped to something reasonable then

← Previous day | Channels | #minetest-dev index | Today | Next day → | Google Search | Plaintext