| Time |
Nick |
Message |
| 00:01 |
sapier1 |
maybe ... if signs_lib abuses some mod mechanisms in a way they aren't meant to be used, e.g. by creating thousands of entities ;-) that'd be a mod problem |
| 00:02 |
VanessaE |
try tens of entities. |
| 00:02 |
VanessaE |
one per sign and only a few dozen signs around a typical spawn area at most |
| 00:03 |
kaeza |
VanessaE, I was reading that just now |
| 00:03 |
VanessaE |
one entity per sign, one final texture per entity, composited together from up to around 150-200 characters (if the sign is full). |
| 00:04 |
VanessaE |
kaeza: ok |
| 00:04 |
sapier1 |
how is composition done? |
| 00:04 |
kaeza |
I don't think compositing could cause FPS drops, as it's a single texture that gets modified at load time |
| 00:05 |
sapier1 |
maybe still loading those entities is the reason for fps drop ... whatever is caused by it |
| 00:05 |
kaeza |
so, "foo.png^bar.png^baz.png" causes the same FPS drop as just "foo.png" |
| 00:05 |
sapier1 |
not sure about that |
| 00:05 |
sapier1 |
you add a lot of new textures |
| 00:06 |
sapier1 |
meshes are sorted into lists by their materials |
| 00:06 |
sapier1 |
so you add a new list for each of those textures |
| 00:07 |
kaeza |
sapier1, https://github.com/minetest/minetest/blob/master/src/tile.cpp#L952 |
| 00:07 |
sapier1 |
that's not what really causes lag |
| 00:07 |
sapier1 |
I think more about the meshbuffer creation |
| 00:07 |
sapier1 |
there's a list for each material |
| 00:08 |
sapier1 |
each meshbuffer (node) is added to the list containing it's material |
| 00:08 |
sapier1 |
think about a scenario where e.g. dirt is at the end of that list but the tile textures at the beginning |
| 00:08 |
sapier1 |
sign textures |
| 00:08 |
sapier1 |
the list is traversed to the end for each dirt node |
| 00:09 |
kaeza |
sooo it is an engine problem? |
| 00:10 |
sapier1 |
partial |
| 00:10 |
sapier1 |
you can always say allowing silly things is an engine problem ;) |
| 00:10 |
VanessaE |
entirely. |
| 00:11 |
VanessaE |
if I send a thousand images through the compositor, I expect one final image on the output. that's all the engine should be dealing with in the end. |
| 00:11 |
VanessaE |
if it doesn't composite it down to one, it's doing its job wrong :P |
| 00:11 |
sapier1 |
but if you say so forceload entities will never have a chance to be added because you can easyly kill engine by abusing them |
| 00:11 |
VanessaE |
and if it does it, but it does it more than once, it's also doing it wrong. |
| 00:12 |
VanessaE |
well sure, but these aren't force-loaded. |
| 00:12 |
VanessaE |
different problem entirely. |
| 00:12 |
sapier1 |
that's just been an example for abusing ;-) |
| 00:12 |
VanessaE |
of course |
| 00:12 |
kaeza |
PS: feature abuse should rest on the modders'/server maintainer shoulders, not in the engine developers |
| 00:13 |
VanessaE |
but as long as we're gonna discuss abusing the engine :P let's discuss this specific form of abuse and how it shows the engine doing it wrong |
| 00:13 |
sapier1 |
I agree with engine shouldn't create a new scene nod on each step but that's how it is ... |
| 00:14 |
VanessaE |
kaeza: generally yes, as long as it's because the modder is just an idiot and doing it wrong to begin with |
| 00:15 |
sapier1 |
most likely there wont be a solution within a couple of days so moders should try to avoid the problematic things for the time being |
| 00:15 |
VanessaE |
sapier1: of course not. |
| 00:23 |
|
Piggybear87 joined #minetest-dev |
| 00:41 |
kaeza |
re: https://github.com/minetest/minetest/pull/583#issuecomment-15694987, https://github.com/minetest/minetest/pull/1075 |
| 00:41 |
kaeza |
fire at will |
| 00:41 |
VanessaE |
BANG! |
| 00:42 |
VanessaE |
(why did I just fire at Will? will he survive?) |
| 00:42 |
kaeza |
hah |
| 00:56 |
OldCoder |
Hi. I'm seeing requests for actual media files such as GET /media/curl/robin2.ogg instead of the hashes sfan5 had me make. |
| 00:56 |
OldCoder |
Was he correct or incorrect that only hashes are used now? |
| 01:03 |
|
Taoki joined #minetest-dev |
| 01:08 |
OldCoder |
Taoki, you here? |
| 01:14 |
Piggybear87 |
Can anyone help? I cannot connect to servers since they updated. |
| 01:14 |
asl |
which? |
| 01:15 |
asl |
landrush? |
| 01:15 |
Piggybear87 |
Landrush and Minerealms. |
| 01:15 |
asl |
!up minetest.org 30011 |
| 01:15 |
ShadowBot |
asl: minetest.org port 30011 is down |
| 01:15 |
Piggybear87 |
!up minetest.org 30012 |
| 01:15 |
ShadowBot |
Piggybear87: minetest.org port 30012 is down |
| 01:15 |
asl |
it's down atm so i guess there nothing we could do |
| 01:17 |
OldCoder |
Folks, 30011 is migrating to LevelDB |
| 01:17 |
OldCoder |
That will take hours. |
| 01:17 |
OldCoder |
The other servers are experiencing network problems that seem to be in the code |
| 01:17 |
OldCoder |
I have installed curl but what sfan5 told me is not true |
| 01:17 |
OldCoder |
Vanessa is not clear about what he meant |
| 01:17 |
OldCoder |
Wait; 30012 should not be down |
| 01:18 |
Piggybear87 |
It is. |
| 01:18 |
OldCoder |
Nope (tm) |
| 01:18 |
OldCoder |
People are on it right now |
| 01:18 |
Piggybear87 |
!up minetest.org 30012 |
| 01:18 |
OldCoder |
!up minetest.org 30012 |
| 01:18 |
ShadowBot |
Piggybear87: minetest.org port 30012 is down |
| 01:18 |
ShadowBot |
OldCoder: minetest.org port 30012 is down |
| 01:18 |
OldCoder |
Interesting |
| 01:18 |
OldCoder |
Miles Dyson is hard at work |
| 01:19 |
OldCoder |
Right this moment |
| 01:19 |
OldCoder |
Is this a bug in the server's network code? |
| 01:19 |
OldCoder |
And not related to media at all? |
| 01:19 |
asl |
curl is just a program to download stuff using the http protocol, something like wget, it shouldn't really effect the way minetest works |
| 01:19 |
OldCoder |
I can try to revert the git code to a few days ago |
| 01:19 |
asl |
~up minetest.org 30012 |
| 01:19 |
ShadowBot |
asl: minetest.org port 30012 is down |
| 01:19 |
OldCoder |
But it is up right now |
| 01:20 |
OldCoder |
Checking git log |
| 01:21 |
OldCoder |
Interesting. I show few commits recently. |
| 01:21 |
OldCoder |
Who can tell me how to do a revert? |
| 01:21 |
asl |
git revert? |
| 01:21 |
asl |
git reset --hard HEAD^ |
| 01:22 |
asl |
two ^ for back 2 commits, three for 3 commits, and so on |
| 01:24 |
|
cg72 joined #minetest-dev |
| 01:25 |
|
CraigyDavi_ joined #minetest-dev |
| 01:33 |
|
VanessaE joined #minetest-dev |
| 01:36 |
OldCoder |
asl I shall try that |
| 01:38 |
|
VanessaE joined #minetest-dev |
| 01:40 |
OldCoder |
Piggybear87, when you press Escape does it say USE_CURL=1 ? |
| 01:40 |
Piggybear87 |
Yes, yes it does. |
| 01:40 |
OldCoder |
kk |
| 01:41 |
OldCoder |
Piggybear87, I can't really do anything unless devs will talk to me. I am building a server with recent changes removed. |
| 01:41 |
sapier1 |
I'd suggest not usin curl but for what I know you can't disable it on runtime |
| 01:41 |
OldCoder |
sapier1, without curl they can't connect |
| 01:42 |
sapier1 |
? |
| 01:42 |
OldCoder |
sapier1, suddenly nobody could connect |
| 01:42 |
OldCoder |
To any world |
| 01:42 |
OldCoder |
It stopped halfway thru media |
| 01:42 |
OldCoder |
People told me |
| 01:42 |
sapier1 |
what versions? |
| 01:42 |
Piggybear87 |
I'm trying the Windows version of stable to see if that works, AFTER Wine installs. If that works, it's the latest git that's the problem. |
| 01:42 |
OldCoder |
MT no longer works without curl |
| 01:42 |
OldCoder |
sfan5 said to run his script to make hashes |
| 01:42 |
sapier1 |
stable? |
| 01:42 |
Piggybear87 |
0.4.9-stable |
| 01:42 |
sapier1 |
come on stable is broken for ages |
| 01:43 |
OldCoder |
They just want to play |
| 01:43 |
asl |
piggybear, try building from the older commit from the git, the one i am using should work |
| 01:43 |
sapier1 |
stable was broken on transmission of textures for quite some time, I fixed this about 3-4 months ago, as of that time non curl is almost as fast as curl mode |
| 01:44 |
OldCoder |
Well, today they stopped being able to connect |
| 01:44 |
OldCoder |
sfan5 and others |
| 01:44 |
OldCoder |
said to set up curl |
| 01:44 |
OldCoder |
It just stops halfway through media |
| 01:44 |
OldCoder |
Suggestions are welcome |
| 01:44 |
OldCoder |
I am reverting the server back a month |
| 01:45 |
sapier1 |
don't do that |
| 01:45 |
OldCoder |
What do you suggest? |
| 01:45 |
sapier1 |
first of all find out what version those ppl use |
| 01:45 |
OldCoder |
Piggybear87, ^ |
| 01:45 |
Piggybear87 |
The newest PPA for me. |
| 01:45 |
OldCoder |
he says "latest PPA" |
| 01:45 |
sapier1 |
that's as saying "I use something" |
| 01:45 |
OldCoder |
Heh |
| 01:45 |
OldCoder |
Go on |
| 01:45 |
Piggybear87 |
The PPA that was released TODAY. |
| 01:46 |
sapier1 |
minetest doesn't release ANY ppa so that version is built by someone else |
| 01:46 |
sapier1 |
based on some version that one decided to use |
| 01:46 |
OldCoder |
sapier1, what client should he use? |
| 01:47 |
sapier1 |
you should see the git hash on upper left corner can you post it |
| 01:47 |
sapier1 |
? |
| 01:47 |
sapier1 |
if your server is on latest master users should use latest master ... this is quit inconveniant I know but that's the way to go on bleeding edge code |
| 01:48 |
sapier1 |
still if there was some change in behaviour we should find out why it did change |
| 01:48 |
sapier1 |
did you do an update lately oldcoder? |
| 01:48 |
OldCoder |
Probably rebuilt from git |
| 01:48 |
OldCoder |
sapier1, I'll try building him a client myself |
| 01:49 |
OldCoder |
Piggybear87, are you Windows? Ubuntu? |
| 01:49 |
Piggybear87 |
Lubuntu. |
| 01:49 |
OldCoder |
Hm; sorry. I don't think I can cross-build. |
| 01:49 |
sapier1 |
can you provide a link to the source of your ppa? |
| 01:49 |
OldCoder |
Just for Windows. |
| 01:49 |
Piggybear87 |
One sec sapier1. |
| 01:50 |
Piggybear87 |
sapier1, https://code.launchpad.net/~minetestdevs/+archive/daily-builds/+packages?field.name_filter=&field.status_filter=published&field.series_filter=trusty |
| 01:50 |
Piggybear87 |
11 hours ago. |
| 01:52 |
sapier1 |
ok that seems to be latest master |
| 01:52 |
Piggybear87 |
Like I was saying... |
| 01:52 |
sapier1 |
ok then next question what's been the last version to work for you? |
| 01:53 |
Piggybear87 |
It started with an "f" |
| 01:53 |
OldCoder |
f? |
| 01:53 |
OldCoder |
Um; the hash? |
| 01:54 |
Piggybear87 |
minetest-minetest-f<something> |
| 01:54 |
OldCoder |
Is anybody else having problems ATM with my world 30012? |
| 01:55 |
sapier1 |
well it might be f70e ... I suggest getting that version and check it ... if oldcoder changes his server and you change your client you'll never find the bug |
| 01:55 |
OldCoder |
sapier1, I restarted git-latest |
| 01:55 |
Piggybear87 |
I'm not self compiling anymore, I can't get an older client. |
| 01:57 |
sapier1 |
well that's a problem |
| 01:57 |
sapier1 |
do others have same issue you have? |
| 01:58 |
Piggybear87 |
Yes, quite a few from what I've heard. |
| 01:58 |
OldCoder |
Piggybear87, "what I've heard" is anecdotal |
| 01:58 |
OldCoder |
Is *anybody* else having problems? |
| 01:58 |
Piggybear87 |
What I've read. |
| 01:59 |
OldCoder |
Which is basically Vazon? |
| 01:59 |
OldCoder |
Repeating what you said? |
| 01:59 |
sapier1 |
old client used to have the hang issue |
| 01:59 |
OldCoder |
No offense |
| 01:59 |
OldCoder |
Hm |
| 01:59 |
OldCoder |
Piggybear87, I'd like to see you build a new client |
| 01:59 |
sapier1 |
so there's a big difference if we're talking about recent clients or old ones e.g. buildcraft ones |
| 02:00 |
Piggybear87 |
If I have the latest master, then, making a client would net me the same thing I have... |
| 02:01 |
OldCoder |
Not necessarily. sapier1 what might you suggest? |
| 02:01 |
sapier1 |
oldcoder minetest.org 30012? |
| 02:01 |
OldCoder |
Yes |
| 02:02 |
OldCoder |
Running git-current |
| 02:02 |
OldCoder |
AND curlk |
| 02:02 |
OldCoder |
AND curl * |
| 02:02 |
OldCoder |
though sfan5 instructions do not match up with what clients do |
| 02:04 |
OldCoder |
Trying it myself with renamed cache |
| 02:05 |
OldCoder |
sapier1, it seems pretty slow |
| 02:05 |
Piggybear87 |
For some reason I can't login as root OR sudo... |
| 02:05 |
OldCoder |
Not stopped, though |
| 02:05 |
OldCoder |
Login to what? |
| 02:05 |
Piggybear87 |
Terminal, so I can make a client. |
| 02:05 |
sapier1 |
hmm I got all textures ... need to disable preload |
| 02:05 |
OldCoder |
Is there a setting? |
| 02:06 |
Piggybear87 |
For? |
| 02:06 |
OldCoder |
Disable preload, sapier1 |
| 02:06 |
sapier1 |
preload_item_visuals = false |
| 02:06 |
OldCoder |
Is this the problem? |
| 02:07 |
OldCoder |
And is that a server side setting? |
| 02:07 |
sapier1 |
no |
| 02:07 |
OldCoder |
Then no fix :( |
| 02:07 |
OldCoder |
Is this the problem? |
| 02:07 |
asl |
preloading is always a bad idea on 512mb of ram |
| 02:07 |
sapier1 |
it's a client setting and I just tried an old client to be sure |
| 02:08 |
OldCoder |
Is there a patch I can make to the server code? |
| 02:08 |
sapier1 |
now latest one |
| 02:08 |
sapier1 |
for what I see nothing related to network code changed recently |
| 02:08 |
OldCoder |
Is the preload issue the problem? |
| 02:08 |
OldCoder |
What is happening? |
| 02:09 |
Piggybear87 |
Brb |
| 02:09 |
sapier1 |
sad to tell but works for me |
| 02:09 |
OldCoder |
kk |
| 02:09 |
asl |
preloading double/triple the loading time |
| 02:09 |
sapier1 |
I can connect using latest as well as old clients |
| 02:09 |
OldCoder |
Clarify? Are all clients preloading? |
| 02:10 |
OldCoder |
Is this the problem? |
| 02:10 |
sapier1 |
no it isn't |
| 02:10 |
OldCoder |
You said; got all textures? |
| 02:10 |
sapier1 |
it's just annoying slow on startup |
| 02:10 |
OldCoder |
He claims that he and GF waited a long time |
| 02:10 |
sapier1 |
yes I got all textures even on using curl |
| 02:10 |
OldCoder |
So this *is* a problem |
| 02:10 |
OldCoder |
Is there a patch I can make on the server side? |
| 02:10 |
OldCoder |
To disable preloading? |
| 02:10 |
sapier1 |
we don't even know what version she used |
| 02:11 |
sapier1 |
no it's a client only setting |
| 02:11 |
sapier1 |
no way to fix it server side |
| 02:11 |
OldCoder |
I mean... |
| 02:11 |
OldCoder |
Ah |
| 02:11 |
OldCoder |
So even if I modify the code? |
| 02:11 |
OldCoder |
So, no solution |
| 02:11 |
sapier1 |
you can't do anything as of that issue |
| 02:11 |
OldCoder |
Too bad |
| 02:11 |
|
Piggybear87 joined #minetest-dev |
| 02:11 |
OldCoder |
Why did it get worse? |
| 02:11 |
sapier1 |
if preloading items was the issue (I doubt it was) you can't do anything |
| 02:11 |
OldCoder |
It was perceived to have done so |
| 02:12 |
sapier1 |
but I'm not sure it was |
| 02:12 |
asl |
cause he remove the 256mb of ram is my guess |
| 02:12 |
OldCoder |
Piggybear87, exactly how long did you wait? |
| 02:12 |
OldCoder |
And asl what do you mean? |
| 02:12 |
OldCoder |
Did he modify his system? |
| 02:13 |
sapier1 |
possible, e.g. firewall settings, proxy configuration ... things like that |
| 02:13 |
OldCoder |
Hm. So this could be a genuine and serious issue. Or anecdotal rumor. |
| 02:13 |
asl |
originally piggybear said he got 768mb of ram (256+512), then someone i think is VanessaE said to try to remove the 256mb of ram |
| 02:13 |
Piggybear87 |
I waited until the machine slept, !10 minutes. |
| 02:13 |
sapier1 |
possible but without further information no chance to decide |
| 02:13 |
asl |
and see what happen |
| 02:13 |
OldCoder |
Multiple people claimed that clients were simply locking up suddenly halfway through. But no way to be sure. |
| 02:14 |
OldCoder |
Very well. Can do nothing without more people reporting. |
| 02:14 |
OldCoder |
asl, try it without a cache when you can. |
| 02:14 |
asl |
i just did |
| 02:14 |
sapier1 |
OldCoder I know old clients to behave that way on not using curl |
| 02:14 |
OldCoder |
Yes |
| 02:14 |
OldCoder |
Aiming for new ones |
| 02:14 |
Piggybear87 |
And asl, I never took out the RAM, I forgot about it. |
| 02:15 |
sapier1 |
I've never seen a new one to lockup that way ... a lot of ppl claimed but any one I had chance to track down was something completely different or even an old one |
| 02:15 |
OldCoder |
Piggybear87, try the disable setting |
| 02:15 |
OldCoder |
listed above |
| 02:15 |
sapier1 |
but that doesn't proove new one doesn't have that bug |
| 02:15 |
OldCoder |
Did you get it? |
| 02:15 |
OldCoder |
sapier1, agreed that more information will be needed |
| 02:16 |
OldCoder |
preload_item_visuals = false |
| 02:16 |
OldCoder |
^ Piggybear87 |
| 02:16 |
Piggybear87 |
The preload? I turn that off before I launch the game the first time. My computer can't handle it. |
| 02:16 |
asl |
i bet most of those people didn't wait pass 10 mins |
| 02:16 |
OldCoder |
Very well |
| 02:16 |
OldCoder |
asl, who exactly has waited 10 minutes? |
| 02:16 |
OldCoder |
How long *should* it take? |
| 02:16 |
|
werwerwer joined #minetest-dev |
| 02:16 |
asl |
it took i think 15 mins for me to join after renaming the cache |
| 02:16 |
OldCoder |
So this is a known and highly regrettable issue |
| 02:17 |
asl |
landrush took 20mins iirc when i first join |
| 02:17 |
OldCoder |
But it was only 3 minutes for me just now |
| 02:17 |
asl |
you are closer to your server i think |
| 02:17 |
OldCoder |
But didn't Piggybear87 get in successfully before? |
| 02:17 |
OldCoder |
So what happened to him? |
| 02:17 |
sapier1 |
oldcoder what's uplink bandwidth of your server? |
| 02:17 |
OldCoder |
Pretty fast. I don't recall. It's a cloud services level system. |
| 02:17 |
asl |
internet is a unstable thing |
| 02:18 |
OldCoder |
asl, he shouldn't have locked up if he had a cache |
| 02:18 |
Piggybear87 |
asl, not me. LAndrush started right away for me the first time. |
| 02:18 |
asl |
somedays i just can't join, others it work fine |
| 02:18 |
OldCoder |
Folks, I can't spend more time on this today |
| 02:18 |
OldCoder |
Need to move on |
| 02:18 |
OldCoder |
Exhausted after a lively few months |
| 02:18 |
OldCoder |
I'll finish migrating Landrush to LevelDB |
| 02:19 |
asl |
finally :D |
| 02:19 |
OldCoder |
I suggest Piggybear87 try a different PC, more RAM, and building a client |
| 02:19 |
OldCoder |
It's at 90% asl |
| 02:19 |
asl |
oh |
| 02:19 |
OldCoder |
Um; 90% is good? |
| 02:19 |
OldCoder |
Hopefully just a little longer. If it broke might be down all night but need to do this. |
| 02:20 |
OldCoder |
Speed is good |
| 02:20 |
Piggybear87 |
OldCoder, Gonna send me money to buy a new machine? That's the only way I can try a different PC. |
| 02:20 |
OldCoder |
Piggybear87, Gonna send me money to host this thing? It actually does cost money and I have none. |
| 02:20 |
* OldCoder |
did not intend to sound sharp |
| 02:20 |
asl |
i would chose lighting instead of leveldb/sql if i could |
| 02:20 |
OldCoder |
Is that an option? |
| 02:20 |
OldCoder |
What are pros and cons |
| 02:20 |
asl |
nope |
| 02:21 |
asl |
lighting is the fastest db or so some people claim |
| 02:21 |
sapier1 |
hmm I just tried using non curl mode too ... no issue |
| 02:21 |
OldCoder |
asl, it depends on purpose |
| 02:21 |
OldCoder |
sapier1, thank you for your time. It was really slow for me without cache even with curl. |
| 02:21 |
Piggybear87 |
Well you suggested that I try a differnt PC, I cannot do that without money, I flat broke and have a 4 month old... |
| 02:21 |
OldCoder |
Not locked up, though. This ought to be the #1 development priority. |
| 02:22 |
OldCoder |
Piggybear87, I am sorry to hear it but if you knew of my life; never mind |
| 02:22 |
sapier1 |
well I felt it beeing faster without curl, but I'm biased as I fixed that code |
| 02:22 |
OldCoder |
sapier1, thank you for your time. Is there any chance of making the long delays of 20+ minutes for some people the #1 priority? |
| 02:22 |
OldCoder |
This *has* been a problem since I joined the project |
| 02:23 |
OldCoder |
2 years now |
| 02:23 |
OldCoder |
With or without curl |
| 02:23 |
sapier1 |
for initial join I don't see a way to fix it |
| 02:23 |
OldCoder |
The protocol can't be fixed up? |
| 02:23 |
OldCoder |
Should it take 10-20 minutes? |
| 02:23 |
sapier1 |
no because any high speed sending will cause lag for other players |
| 02:23 |
OldCoder |
How about medium speed? |
| 02:23 |
OldCoder |
And default to no preload on client side? |
| 02:23 |
sapier1 |
well it's quite fast for most players |
| 02:24 |
OldCoder |
I have never seen loads fast |
| 02:24 |
sapier1 |
I cleared cache and joined within less then 5 min |
| 02:24 |
OldCoder |
Hm |
| 02:24 |
OldCoder |
So it comes down to the time that is acceptable. Perhaps this could be more clearly documented; load screen should warn people. |
| 02:24 |
sapier1 |
I know about packet loss on your line causing a download to become really really slow |
| 02:24 |
OldCoder |
Also preload should be off by default; is this reasonable? |
| 02:24 |
sapier1 |
but that's hard do fix |
| 02:25 |
OldCoder |
I do have a technical suggestion |
| 02:25 |
sapier1 |
imho yes preload should be off I don't actually know why it's still on |
| 02:25 |
OldCoder |
Send files as bundles. File by file for such small files in some cases is really slow; correct? |
| 02:26 |
sapier1 |
in theory yes but in practice that's not the issue |
| 02:26 |
Piggybear87 |
I got this in Terminal and it wont go away |
| 02:26 |
Piggybear87 |
END-USER LICENSE AGREEMENT FOR MICROSOFT SOFTWARE |
| 02:26 |
Piggybear87 |
Ideas? |
| 02:26 |
OldCoder |
sapier1, I see this as the make or break issue in competition with MC. |
| 02:26 |
OldCoder |
??? |
| 02:26 |
OldCoder |
WTF? You are Lubuntu, right? |
| 02:26 |
asl |
lol, that's why i don't use MICROSOFT stuff |
| 02:26 |
Piggybear87 |
Yes, I tried to install Wine and it froze. |
| 02:26 |
OldCoder |
sapier1, you have been generous with your time |
| 02:26 |
asl |
they lock up stuff |
| 02:26 |
OldCoder |
sapier1, no further questions for now except this: |
| 02:26 |
asl |
if anyone want to know more about lighting, look here: http://symas.com/mdb/ |
| 02:27 |
OldCoder |
sapier1, (a) I suggest preload off by default (b) I suggest load screen say "FIRST CONNECT MAY TAKE 15 MINUTES". These two changes would help a lot. |
| 02:27 |
Piggybear87 |
I cannot use Terminal now because of this. If I have to reinstall Lubuntu I am pouring soda all over the motherboard. |
| 02:27 |
OldCoder |
Piggybear87, this is a topic for #minetest and not #minetest-dev |
| 02:28 |
sapier1 |
a reasonable suggestion oldcoder true |
| 02:28 |
Piggybear87 |
I'm trying to get a new client like you said, I need help from this channel to help see if old clients work better... |
| 02:28 |
OldCoder |
I would like humbly to suggest both (a) and (b) to the devs |
| 02:28 |
OldCoder |
Piggybear87, I will join you over in #minetest. I should warn you that this channel is finicky about off-topic. |
| 02:29 |
OldCoder |
|
| 02:29 |
Piggybear87 |
Then I will not help this little experiment... |
| 02:29 |
OldCoder |
Piggybear87, do not be distressed; I will PM shortly. |
| 02:29 |
OldCoder |
sapier1, thank you again! |
| 02:30 |
|
sapier1 left #minetest-dev |
| 02:44 |
|
ShadowNinja joined #minetest-dev |
| 02:44 |
RealBadAngel |
http://pastebin.com/hNHrgRvh |
| 02:44 |
RealBadAngel |
ooops |
| 02:52 |
OldCoder |
RealBadAngel, Hi |
| 03:26 |
|
zat joined #minetest-dev |
| 03:27 |
|
zat1 joined #minetest-dev |
| 03:32 |
|
CraigyDavi__ joined #minetest-dev |
| 04:03 |
|
CraigyDavi joined #minetest-dev |
| 04:16 |
|
CraigyDavi_ joined #minetest-dev |
| 04:37 |
|
Miner_48er joined #minetest-dev |
| 04:52 |
Miner_48er |
will sqlite3 remain supported in future versions? |
| 05:14 |
|
nore joined #minetest-dev |
| 05:35 |
|
CraigyDavi joined #minetest-dev |
| 05:39 |
|
Hunterz joined #minetest-dev |
| 06:20 |
nore |
Can someone check that please? https://github.com/Novatux/minetest/commit/e17d29ccd608a654c3038bf35f5ca710e5ef0db8 |
| 06:26 |
|
darkrose joined #minetest-dev |
| 06:26 |
|
darkrose joined #minetest-dev |
| 06:36 |
|
Weedy joined #minetest-dev |
| 07:07 |
nore |
Merging https://github.com/Novatux/minetest/commit/e17d29ccd608a654c3038bf35f5ca710e5ef0db8 in one hour if no one disagrees |
| 07:08 |
sfan5 |
that looks fine |
| 07:08 |
sfan5 |
but I don't know anything of the clouds stuff |
| 07:08 |
sfan5 |
so I can't verify that it is correct (w/o testing) |
| 07:09 |
nore |
Well it works, and the source of the bug was that the bounding box for clouds was not changed when camera offset was changed |
| 07:09 |
sfan5 |
if it works go ahead |
| 07:09 |
nore |
ok |
| 07:11 |
nore |
done |
| 07:58 |
|
SmugLeaf joined #minetest-dev |
| 07:58 |
|
SmugLeaf joined #minetest-dev |
| 07:58 |
|
Weedy joined #minetest-dev |
| 07:58 |
|
Weedy joined #minetest-dev |
| 07:59 |
|
Weedy_ joined #minetest-dev |
| 08:00 |
|
SmugLeaf joined #minetest-dev |
| 08:00 |
|
SmugLeaf joined #minetest-dev |
| 08:01 |
|
Fresh_me_ joined #minetest-dev |
| 08:03 |
|
SmugLeaf joined #minetest-dev |
| 08:03 |
|
SmugLeaf joined #minetest-dev |
| 08:04 |
|
Weedy joined #minetest-dev |
| 08:04 |
|
Weedy joined #minetest-dev |
| 08:08 |
|
dsimon joined #minetest-dev |
| 08:18 |
|
diemartin joined #minetest-dev |
| 08:30 |
|
ligeti joined #minetest-dev |
| 08:35 |
|
restcoser joined #minetest-dev |
| 08:50 |
|
sapier joined #minetest-dev |
| 08:53 |
|
Calinou joined #minetest-dev |
| 08:57 |
|
Amaz_ joined #minetest-dev |
| 09:06 |
|
ligeti joined #minetest-dev |
| 09:08 |
|
PenguinDad joined #minetest-dev |
| 09:36 |
|
Garmine joined #minetest-dev |
| 09:43 |
sfan5 |
~tell ShadowNinja why do we have an RFC tag? |
| 09:43 |
ShadowBot |
sfan5: O.K. |
| 10:18 |
|
proller joined #minetest-dev |
| 10:36 |
|
PilzAdam joined #minetest-dev |
| 11:15 |
|
jin_xi joined #minetest-dev |
| 11:16 |
|
smoke_fumus joined #minetest-dev |
| 11:20 |
|
ImQ009 joined #minetest-dev |
| 11:31 |
sapier |
celeron55 I can't find any location except of addArea using the NOT_LOADED flag ... and there it seems to be used to skip unloaded areas only ... I'm not exactly sure but wouldn't it be better to just copy unloaded areas if it's faster then skipping them? |
| 11:33 |
celeron55 |
umm... dunno |
| 11:33 |
celeron55 |
http://i.imgur.com/PsoQC48.png |
| 11:33 |
celeron55 |
i generated this when trying doxygen |
| 11:33 |
sapier |
of course in worst case scenario lower left node is set and upper right one for a voxelmanip beeing 10000x10000x1000 copying everything is wors then skipping |
| 11:33 |
celeron55 |
it sheds some light to this issue; looks like VM::addArea is used by two other things too |
| 11:34 |
sapier |
yes I had a look at it |
| 11:34 |
celeron55 |
(the MeshMakeData ones) |
| 11:36 |
celeron55 |
the original use case for VoxelManipulator and MapVoxelManipulator was an initial water implementation |
| 11:37 |
sapier |
is it used any longer? |
| 11:37 |
celeron55 |
it was a finite one which needed speedy access for weirdly propagating air bubbles through water and that's why it used automatic emerge too |
| 11:37 |
celeron55 |
no, it was used only for like two months many years ago |
| 11:37 |
|
vifino joined #minetest-dev |
| 11:38 |
sapier |
hmmm ... for what I understand the approach of looping twice through a possibly big 3d array doesn't seem to be very wise to me ... but I don't know if there's another way to do it |
| 11:39 |
sapier |
especially as everything is set by assignment operators in this loop instead of memcpy which usually is highly optimized |
| 11:40 |
sapier |
I haven't found any issue, do you have a hint what to test in order to find possible bugs? |
| 11:41 |
celeron55 |
have you invoked all the things that are visible in that graph |
| 11:42 |
sapier |
hard to tell :-) ... hmm placing breakpoints in there could tell |
| 11:42 |
|
Megaf joined #minetest-dev |
| 11:42 |
|
CraigyDavi_ joined #minetest-dev |
| 11:42 |
celeron55 |
most of that gets done if you start a new world and start a singleplayer game in it |
| 11:43 |
celeron55 |
will you be removing ManualVoxelManipulator? |
| 11:43 |
celeron55 |
wait |
| 11:43 |
celeron55 |
i mean MapVoxelManipulator |
| 11:43 |
sapier |
I haven't removed it yet by now I just removed emerge ... no use for a function calling another function |
| 11:44 |
celeron55 |
i don't mean that one, that's in ManualMapVoxelManipulator |
| 11:44 |
sapier |
creating a new world is my regular testcase for this optimizations |
| 11:45 |
sapier |
I always use same seed too to be at least somehow compareable ... yet random player pos is an issue |
| 11:48 |
celeron55 |
there could be some kind of issue related to schematics and treegen, in some corner case |
| 11:48 |
celeron55 |
that would be serious as the end result would get saved on disk |
| 11:49 |
celeron55 |
the client stuff is not a concern at this point |
| 11:49 |
celeron55 |
also lighting might still screw up in some weird case? |
| 11:49 |
celeron55 |
which is saved too |
| 11:50 |
celeron55 |
not that it would be terribly reliable anyway, but 8) |
| 11:51 |
celeron55 |
whatever; if you're ready to quickly fix any problems that arise, i'm fine with it |
| 11:51 |
sapier |
well I don't know how bad it would be but switching from a slow glitchy solution to a fast (equaly) glitchy one sounds like an option to me |
| 11:52 |
sapier |
of course if new one is way more bad that's completely different |
| 11:55 |
Taoki |
OldCoder: Back now. Hi |
| 12:18 |
sapier |
removing MapVoxelManipulator isn't a big deal, the only one using it is ManualVoxelManipulator |
| 12:18 |
celeron55 |
wait, what |
| 12:19 |
celeron55 |
in that case don't remove it |
| 12:19 |
sapier |
ManualVoxelManipulator is derived from it but only uses it's map |
| 12:19 |
sapier |
everything else is reimplemented |
| 12:19 |
celeron55 |
oh 8) |
| 12:20 |
sapier |
oh clear isn't reimplemented too ... calling map clear |
| 12:22 |
sapier |
lol I just removed emerge from ManualVoxelManipulator to make compiler tell me where it's used ... guess what ... it isn't even used |
| 12:23 |
celeron55 |
yeah? that's that the graph shows |
| 12:24 |
sapier |
I wonder how much dead code is in there we could remove :-) |
| 12:24 |
celeron55 |
not very much |
| 12:24 |
sapier |
within whole minetest? ;-) |
| 12:24 |
celeron55 |
you actually can find most of it by browsing through the doxygen diagrams |
| 12:25 |
celeron55 |
that's what i noticed when trying the 1.5 years old pull request that adds doxygen to cmake 8) |
| 12:25 |
sapier |
maybe I should do this next time |
| 12:27 |
sapier |
maybe we should encourage ppl to use doxygen ;-) |
| 12:32 |
sapier |
I replaced the perf pull by #1359 ... last two commits are new, later one isn't really a performance fix, I may put it separate |
| 12:32 |
ShadowBot |
https://github.com/minetest/minetest/issues/1359 -- Perf fixes june 2014 2 by sapier |
| 12:34 |
proller |
removing whole *VoxelManipulator* - good idea ;) |
| 12:37 |
celeron55 |
removing proller - good idea |
| 12:39 |
|
book` joined #minetest-dev |
| 12:48 |
|
eeew joined #minetest-dev |
| 12:56 |
|
hmmmm joined #minetest-dev |
| 12:57 |
proller |
celeron55, and it will increase performance ? |
| 13:00 |
celeron55 |
it will improve performance of these irc channels |
| 13:00 |
celeron55 |
anyway, what would you replace VMs with? |
| 13:01 |
celeron55 |
they quite clearly do what they are intended to do, which is store temporary voxel data and accelerate access speed to it |
| 13:01 |
celeron55 |
and rather considerably too |
| 13:06 |
proller |
no 8) |
| 13:07 |
Megaf |
Hi all |
| 13:07 |
Megaf |
some players on my server and me when using the windows build get the error "Access Violation". Why? |
| 13:09 |
Megaf |
Maybe we are running out of memory? |
| 13:42 |
|
CraigyDavi joined #minetest-dev |
| 13:58 |
OldCoder |
Taoki, hi |
| 13:58 |
Taoki |
hi :) |
| 14:20 |
Megaf |
sapier: |
| 14:20 |
Megaf |
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND |
| 14:20 |
Megaf |
5365 pi 20 0 84368 25m 5404 S 6.5 11.1 1:21.79 ServerThread |
| 14:20 |
Megaf |
5363 pi 20 0 84368 25m 5404 S 1.3 11.1 0:23.53 ConnectionSend |
| 14:20 |
Megaf |
That's on my Raspberry Pi |
| 14:20 |
Megaf |
2 or 3 players online |
| 14:20 |
Megaf |
not bad uh? |
| 14:21 |
Megaf |
celeron55: I managed to make Minetest Server run great on my Raspberry Pi and ODroid U3 (1 x 630 MHz armv6 and 4 x 2,1(or 2,1, I'm not sure) GHz armv7) |
| 14:23 |
celeron55 |
so, what did you do then? |
| 14:23 |
Megaf |
Here is the full conf, http://paste.debian.net/104114/ |
| 14:24 |
Megaf |
ANd the following settings are key for good performance |
| 14:24 |
Megaf |
max_simultaneous_block_sends_per_client = 1 |
| 14:24 |
Megaf |
max_simultaneous_block_sends_server_total = 10 |
| 14:24 |
Megaf |
max_simultaneous_block_sends_per_client = 1 |
| 14:24 |
Megaf |
max_simultaneous_block_sends_server_total = 10 |
| 14:24 |
Megaf |
emergequeue_limit_total = 10 |
| 14:24 |
Megaf |
emergequeue_limit_diskonly = 4 |
| 14:24 |
Megaf |
emergequeue_limit_generate = 4 |
| 14:24 |
Megaf |
max_forceloaded_blocks = 1 |
| 14:25 |
Megaf |
(Those are good settings for up to 10 players) |
| 14:25 |
celeron55 |
it's going to pretty much hang every 5 minutes to save the map, but if that is fine, whatever |
| 14:26 |
Megaf |
I tried using short save maps intervals, but it will hang for the same amount of time or even longer |
| 14:26 |
Megaf |
tried 1 second, 5 seconds, 30 seconds, 60 seconds |
| 14:26 |
celeron55 |
if nothing changes in the world, then it's fine |
| 14:26 |
celeron55 |
but if things keep changing all the time, then accumulating all that and saving at 5 minute interval will be bad |
| 14:27 |
Megaf |
I think this may cause some load to the CPU server_unload_unused_data_timeout = if set to very high numbers, like an hour |
| 14:27 |
celeron55 |
i guess your game is quite lightweight so that doesn't happen |
| 14:27 |
|
LemonLake joined #minetest-dev |
| 14:28 |
Megaf |
Yep, the game on the Raspberry Pi is lightweight, but the one on the ODroid is heavy thing, pipeworks, homedecor, mesecons, mesegates... |
| 14:28 |
Megaf |
moretrees, plantlife |
| 14:29 |
Megaf |
it slowed down a lot when I first added moretrees, but than I could get a good performance with that config |
| 14:29 |
celeron55 |
maybe post those settings on the server settings topic |
| 14:29 |
sapier |
maybe we should add some preconfigs for different usecases in util? settings perfect for a pi will be nonsense for a big machine like vanessaE's server and the other way round |
| 14:30 |
sapier |
but I'm not sure if ppl would find them there |
| 14:31 |
Megaf |
I'm quite sure those setting will be fast on VanessaE's server too |
| 14:34 |
sapier |
well and I am sure those settings will cause players to not get any map data on servers with more then hmm about 10 playes online |
| 14:35 |
Megaf |
yep, so just change max_simultaneous_block_sends_server_total = to (numer of players * 1.2 |
| 14:35 |
Megaf |
or even times 2 |
| 14:35 |
sapier |
I once suggested a change like that but it's most likely not that easy |
| 14:36 |
Megaf |
well, my settings relie/rely on fast read access/speed |
| 14:36 |
sapier |
that's why those are settings so everyone can adjust them to her/his usecase |
| 14:37 |
sapier |
most time wrong settings don't cause any major issues but for edge cases like your pi you need the perfect setup |
| 14:38 |
|
Piggybear87 joined #minetest-dev |
| 15:02 |
|
asl joined #minetest-dev |
| 15:13 |
Megaf_ |
sapier; actually I begun changing settings back in the day when my server ran on octo core XEON CPUs |
| 15:13 |
Megaf_ |
with 100 mbps of upink |
| 15:14 |
Megaf_ |
uplink* |
| 15:15 |
Megaf_ |
250 mpbs actually |
| 15:25 |
|
NakedFury joined #minetest-dev |
| 15:37 |
|
Jordach joined #minetest-dev |
| 16:07 |
|
Hunterz joined #minetest-dev |
| 16:08 |
|
Calinou joined #minetest-dev |
| 16:26 |
|
khonkhortisan joined #minetest-dev |
| 16:33 |
|
Anchakor joined #minetest-dev |
| 16:42 |
|
rubenwardy joined #minetest-dev |
| 17:18 |
SoniEx2 |
make world saving/loading scripted |
| 17:19 |
SoniEx2 |
if it isn't already |
| 17:19 |
SoniEx2 |
that way you can test your algorithms and stuff |
| 17:32 |
Tesseract |
sfan5: Because I added it. Why? |
| 18:06 |
|
zat joined #minetest-dev |
| 18:47 |
|
Amaz_ joined #minetest-dev |
| 19:05 |
|
Piggybear87 joined #minetest-dev |
| 19:12 |
|
dsimon left #minetest-dev |
| 20:38 |
|
GhostDoge joined #minetest-dev |
| 20:52 |
|
Jordach joined #minetest-dev |
| 21:20 |
|
proller joined #minetest-dev |
| 21:40 |
|
FlyingPandas_ joined #minetest-dev |
| 21:41 |
|
ImQ009 joined #minetest-dev |
| 21:53 |
|
eeew joined #minetest-dev |
| 21:58 |
|
Eater4 joined #minetest-dev |
| 22:40 |
|
zat joined #minetest-dev |
| 23:04 |
|
Exio4 joined #minetest-dev |
| 23:10 |
|
VanessaE joined #minetest-dev |
| 23:12 |
|
VanessaE joined #minetest-dev |
| 23:37 |
SoniEx2 |
anyone? |
| 23:40 |
RealBadAngel |
whats up? |
| 23:56 |
NakedFury |
not much, chilling with some E3 show |
| 23:56 |
NakedFury |
what about you? |
