| Time |
Nick |
Message |
| 00:42 |
|
XtremeHacker joined #minetest-dev |
| 01:19 |
|
torgdor joined #minetest-dev |
| 01:23 |
|
torgdor joined #minetest-dev |
| 02:29 |
|
Grandolf joined #minetest-dev |
| 02:33 |
|
luk3yx joined #minetest-dev |
| 03:01 |
|
cheapie joined #minetest-dev |
| 03:12 |
|
twoelk left #minetest-dev |
| 03:16 |
|
Hunterz joined #minetest-dev |
| 03:20 |
|
DI3HARD139 joined #minetest-dev |
| 04:39 |
|
paramat joined #minetest-dev |
| 04:40 |
paramat |
hmm reading hub -logs, i think we should drop android support, we're halfway to that already |
| 04:40 |
paramat |
then the apps have to work for their money instead of just copying our code |
| 04:41 |
* paramat |
is doing |
| 04:45 |
paramat |
it would also reduce our workload |
| 04:47 |
paramat |
the apps would continue but would have to do their own android code, this would weed out the lazy ones and leave the more serious commited ones, like multicraft |
| 04:49 |
paramat |
i made a post on the forum 11 months ago asking for new devs to help with android code, nothing has come of that |
| 04:57 |
paramat |
this would actually help android players as their choice of apps is narrowed to higher quality ones |
| 05:03 |
paramat |
and hopefully narrowed to the ones less likely to have adds and payment scams |
| 05:07 |
paramat |
0.5 is next release so a good time to do this |
| 05:07 |
luk3yx |
0.5? |
| 05:09 |
paramat |
0.5.0 |
| 05:13 |
|
dax left #minetest-dev |
| 05:16 |
|
Hunterz joined #minetest-dev |
| 05:18 |
paramat |
so it has been at least a year now with android being neglected by devs and hoping someone would appear to maintain it, long enough. i remember a year ago we talked about possibly dropping android but waiting until we had spent some time waiting for an enthusiastic android dev to appear, or for sapier to return |
| 05:22 |
Hijiri |
are there any android minetests other than the official one without proprietary components? |
| 05:22 |
Hijiri |
not that I really play on android though, since I don't like the controls |
| 05:27 |
paramat |
i dont know |
| 05:28 |
paramat |
we could perhaps team up with a particularly good app and declare it the 'officially recommended' one |
| 05:29 |
paramat |
maybe multicraft, the author has helped out here a few times |
| 05:49 |
|
Raven262 joined #minetest-dev |
| 06:26 |
paramat |
~tell nerzhul for #5798 "Mgv6 mudflow: Remove decoration if 'dirt with grass' below flows away" testing is quite complex, this had been tested by 2 people, including me, so please could you approve on code alone and trust me it is tested? |
| 06:26 |
ShadowBot |
paramat: O.K. |
| 06:31 |
|
DS-minetest joined #minetest-dev |
| 07:39 |
|
nerzhul joined #minetest-dev |
| 07:57 |
|
DS-minetest joined #minetest-dev |
| 09:19 |
|
nerzhul joined #minetest-dev |
| 09:54 |
|
YuGiOhJCJ joined #minetest-dev |
| 10:20 |
|
Fixer joined #minetest-dev |
| 10:25 |
nerzhul |
Hijiri, there is minetest on playstore i publish at each release |
| 10:34 |
nerzhul |
merging #5811 & #5798 in ~10 mins |
| 10:34 |
ShadowBot |
https://github.com/minetest/minetest/issues/5811 -- Close formspec on client shutdown. by red-001 |
| 10:34 |
ShadowBot |
https://github.com/minetest/minetest/issues/5798 -- Mgv6 mudflow: Remove decoration if 'dirt with grass' below flows away by paramat |
| 10:34 |
nerzhul |
seconds need to be pushed fast to servers and tested to ensure we are okay |
| 10:34 |
nerzhul |
nice shot red-001 for the 5811 :) |
| 10:45 |
nerzhul |
wtf ! |
| 10:47 |
nerzhul |
!tell paramat is this normal ? https://lut.im/L8AoeDXpXh/0GYLGgXwe4EpUAN7.png |
| 10:47 |
ShadowBot |
nerzhul: O.K. |
| 10:47 |
nerzhul |
not very realistic |
| 10:47 |
|
paramat joined #minetest-dev |
| 10:47 |
nerzhul |
paramat, is this normal: https://lut.im/L8AoeDXpXh/0GYLGgXwe4EpUAN7.png ? |
| 10:48 |
paramat |
hehe yes in mgv7 |
| 10:48 |
paramat |
3D noise mountain terrain |
| 10:48 |
paramat |
thanks for the merge |
| 10:48 |
|
DS-minetest joined #minetest-dev |
| 10:49 |
paramat |
as many have observed, mgv7 is stunning and epic, and occasionally does something very weird |
| 10:50 |
nerzhul |
yeah, gravity seems to have no sense at this point :( |
| 10:51 |
Raven262 |
But: http://i.imgur.com/Ey6muVK.jpg |
| 10:53 |
paramat |
hehe |
| 10:53 |
nerzhul |
lava fire is disabled by default ? |
| 10:53 |
nerzhul |
i put lava bloc on a tree, no fire |
| 10:53 |
paramat |
erm |
| 10:54 |
Raven262 |
but you could cause a forest fire like that. |
| 10:54 |
paramat |
singleplayer should have fire spread enabled |
| 10:54 |
nerzhul |
okay |
| 10:54 |
nerzhul |
it's not the case atm |
| 10:54 |
nerzhul |
oh |
| 10:54 |
nerzhul |
it was triggered now |
| 10:54 |
nerzhul |
very strange, the first tree doesn't burnt but when lava hit second tree it burns |
| 10:55 |
Raven262 |
First one must have been fire resistant. |
| 10:55 |
Raven262 |
Lag, maybe? |
| 10:55 |
nerzhul |
i have a good cpu and it's regular MT game on my local devel client :p |
| 10:56 |
paramat |
the chance of ignition is deliberately low because fire spread speeds up exponentially |
| 10:56 |
nerzhul |
i tested a performance enhancement for ABM |
| 10:56 |
paramat |
you can place fire at the base of a tree and sometimes it will go out before spreading |
| 10:56 |
Raven262 |
Ah, so not every tree will ignite. |
| 10:56 |
paramat |
yes they will with enough tries |
| 10:56 |
* DS-minetest |
doesn't like mgv7 caves that go deep |
| 10:57 |
* Raven262 |
likes v7 caves that go deep. |
| 10:57 |
paramat |
place fire in leaves and the chance of catching is higher |
| 10:57 |
Raven262 |
No, really caves are fun. |
| 10:57 |
* DS-minetest |
thinks, they destroy gameplay |
| 10:57 |
Raven262 |
Ah, how so? |
| 10:57 |
paramat |
which caves? the never ending tunnels or large blobby caves |
| 10:57 |
DS-minetest |
yep |
| 10:58 |
DS-minetest |
the 1st option |
| 10:58 |
paramat |
ok |
| 10:58 |
Raven262 |
I think tunnels make mining much easier. |
| 10:58 |
paramat |
underground would be boring without long tunnels |
| 10:59 |
DS-minetest |
imo the tunnels are a bit boring |
| 10:59 |
paramat |
how do they destroy gameplay? i'm interested :] |
| 10:59 |
DS-minetest |
the player doesn't have to dig much and already gets deep ores |
| 11:00 |
DS-minetest |
just by falling down |
| 11:00 |
Raven262 |
I think it would be better then if ores were lowered. |
| 11:01 |
nerzhul |
sfan5, rubenwardy #5815 is fairly trivial and requires your approval, can you have a look ? |
| 11:01 |
ShadowBot |
https://github.com/minetest/minetest/issues/5815 -- Enhance ABM performance a little bit by removing two std::set copy by nerzhul |
| 11:01 |
DS-minetest |
also they are so big that they are not cudly |
| 11:01 |
|
Krock joined #minetest-dev |
| 11:01 |
|
Krock joined #minetest-dev |
| 11:01 |
nerzhul |
Krock, ^ |
| 11:01 |
nerzhul |
hi Krock |
| 11:01 |
* Krock |
looks logs |
| 11:01 |
DS-minetest |
so big holes are also unrealistic |
| 11:01 |
rubenwardy |
feature freeze, nerzhul |
| 11:02 |
rubenwardy |
applies to performance fixes :P |
| 11:02 |
nerzhul |
it's not a feature |
| 11:02 |
nerzhul |
it's a fix |
| 11:02 |
rubenwardy |
what bug does it fix? |
| 11:02 |
nerzhul |
feature freeze doesn't mean a trivial performance "fix" should be blocked, it doesn't change the app behaviour |
| 11:03 |
nerzhul |
it's why it's "feature" freeze, and not "bugfix except freeze" |
| 11:03 |
Krock |
hide all features behind a solid wall of "bugfixes" - new PR strategy |
| 11:03 |
rubenwardy |
this has been discussed before by other core devs |
| 11:03 |
nerzhul |
if it was a major refactor i agreed with you as it's not feature but refactor |
| 11:03 |
nerzhul |
Krock, lol |
| 11:03 |
nerzhul |
here it's trivial |
| 11:04 |
Krock |
it doesn't change anything but perhaps the execution speed |
| 11:04 |
nerzhul |
yeah, it improves our ABM speed with a trivial change |
| 11:04 |
nerzhul |
http://dev.minetest.net/Releasing_Minetest doesn't talk about performance fix freeze, just new feature are not accepted |
| 11:05 |
Krock |
trivial and lgtm |
| 11:05 |
nerzhul |
ty Krock , i will insert it later today i'm looking at some other problems |
| 11:05 |
nerzhul |
maybe we should update http://dev.minetest.net/Releasing_Minetest, Blockmen doesn't do release now :p |
| 11:06 |
Krock |
nerzhul, only one tab missing: https://github.com/minetest/minetest/pull/5815/files#diff-59f5585fbd782ec57b38b0ee73fa885cR750 |
| 11:06 |
nerzhul |
who did it for 0.4.14 and 0.4.15 ? Krock or sfan5 ? |
| 11:06 |
Krock |
sfan5, most likely |
| 11:06 |
rubenwardy |
paramat |
| 11:06 |
rubenwardy |
he did it the last two times |
| 11:07 |
nerzhul |
for windows ? |
| 11:07 |
nerzhul |
i mean release, not managing release process |
| 11:07 |
Krock |
if "leveldb" wasn't on that list I could do it aswell |
| 11:07 |
nerzhul |
what is the problem with leveldb ? |
| 11:07 |
paramat |
looking |
| 11:08 |
Krock |
no matter what I'm trying - either it lacks of MSVC support or there are linker errors |
| 11:08 |
paramat |
cuddly tunnels :] |
| 11:08 |
DS-minetest |
yep, there was a d missing >_< |
| 11:09 |
nerzhul |
Krock, MSVC builds doesn't work with leveldb but mingw yes ? |
| 11:09 |
Krock |
I think so |
| 11:09 |
Krock |
haven't tried |
| 11:09 |
nerzhul |
okay i update to set you as MSVS |
| 11:09 |
nerzhul |
MSVC |
| 11:10 |
Krock |
no, I think sfan5 should be "elected" as he's also able to provide 64-bit builds |
| 11:10 |
nerzhul |
we have both builds in travis |
| 11:11 |
Krock |
mmh.. right. |
| 11:11 |
nerzhul |
maybe we should update www.minetest.net to announce we will release a version no ? |
| 11:11 |
nerzhul |
it's our portal, not GH/IRC :) |
| 11:12 |
rubenwardy |
nerzhul, forum is for news |
| 11:12 |
nerzhul |
ok |
| 11:16 |
Krock |
there's one more thing that should be done somewhen: http://dev.minetest.net/Changelog |
| 11:20 |
nerzhul |
wtf no changelog for 0.4.15 |
| 11:21 |
Krock |
well, it's more or less just copying the important changes from Fixer's post. He saves a lot of time |
| 11:22 |
rubenwardy |
I did the 0.4.14 changelog, was possible the most boring thing ever |
| 11:23 |
nerzhul |
yeah, for 0.4.16 it's on the forum |
| 11:23 |
nerzhul |
but i don't wrote all commits, just the most import for end users |
| 11:25 |
paramat |
the last engine changelog ended up pretty much a list of commit messages |
| 11:25 |
nerzhul |
you mean 0.4.15 paramat ? |
| 11:26 |
paramat |
yes. not complaining though |
| 11:26 |
nerzhul |
yeah :) |
| 11:26 |
nerzhul |
i will try to maintain a changelog on the forum every 10-15 days |
| 11:27 |
rubenwardy |
why not the wiki? |
| 11:27 |
paramat |
auto-generate commit messages? some are long though, and technical |
| 11:27 |
rubenwardy |
we used to do 0.4.X -> ? for dev versions |
| 11:27 |
nerzhul |
both can be done, i can do it on wiki but seems related to released version no ? |
| 11:27 |
rubenwardy |
that stopped though as people started adding commits before ones before had been added |
| 11:27 |
nerzhul |
paramat, auto generate is not good for feature teasing :p |
| 11:28 |
nerzhul |
i can prepare 0.4.16 changelog on wiki if someone does 0.4.15 lol |
| 11:29 |
paramat |
ugh :] |
| 11:29 |
nerzhul |
i cannot do the mt_game changelog for both |
| 11:29 |
paramat |
many commits can only be explained simply by someone who understands them |
| 11:30 |
paramat |
it would be best if i did the mapgen ones for example |
| 11:30 |
nerzhul |
i just don't get time to read mtg changelog, if someone wants i prepared changelog scheme on http://dev.minetest.net/Changelog |
| 11:32 |
paramat |
maybe if each time an author has a commit merged they add a simple description somewhere to be used in a changelog. but meh |
| 11:38 |
|
twoelk joined #minetest-dev |
| 11:52 |
sfan5 |
Krock: we can have both msvc and mingw builds again |
| 11:52 |
Krock |
ah right, we already had both build types in earlier versions |
| 11:56 |
nerzhul |
http://dev.minetest.net/Changelog#0.4.15_.E2.86.92_0.4.16_.28not_released_yet.29 |
| 11:56 |
nerzhul |
i transformed the forum changelog in our official changelog |
| 12:13 |
rubenwardy |
you're meant to include authors :P |
| 12:17 |
nerzhul |
no time to re-take the whole history, do it yourself, and someone should complete with mtg |
| 12:20 |
|
RobbieF joined #minetest-dev |
| 12:39 |
|
shivajiva joined #minetest-dev |
| 12:43 |
Fixer |
Calinou posted a site that has a good point: changelog is needed to present most significant and important changes for the player |
| 12:43 |
Fixer |
and mod developer |
| 12:44 |
Fixer |
what you need is just run git log and remove minor stuff from it |
| 12:44 |
nerzhul |
no git log is not intended to be readable for most of our users |
| 12:44 |
nerzhul |
but we need to do some changelogs updates periodically, like i did on forum |
| 12:44 |
nerzhul |
i can do updates when i get time, and motivates, every 10 or 15 days, but i cannot take all changelog, mtg should have same |
| 12:45 |
Fixer |
nerzhul: you can format the output better way, like this https://pastebin.com/raw/maTFqmKh and just remove minor changes from it |
| 12:45 |
Fixer |
and rephrase/merge some other changes |
| 12:46 |
nerzhul |
sorry but it's neither user friendly nor condensed |
| 12:46 |
nerzhul |
http://dev.minetest.net/Changelog#0.4.15_.E2.86.92_0.4.16_.28not_released_yet.29 |
| 12:46 |
nerzhul |
look at this :p |
| 12:46 |
Fixer |
you can use git log as a base for your changelog |
| 12:47 |
nerzhul |
it's what i did but i merged many commits into 1 feature or fix :p |
| 12:47 |
Fixer |
"Hacked clients cannot ignore node damages anymore (fire, lava, cactus...). " |
| 12:47 |
Fixer |
do we have cactus damage? o_0 |
| 12:48 |
nerzhul |
i will split our page changelog in two parts, it's too big, i will move before 0.4.10 versions to OldChangelog page |
| 12:48 |
paramat |
would i be allowed to merge tin stair/slab for release? it's sort of a bug. rubenwardy sfan5 Krock nore |
| 12:48 |
paramat |
no cactus damage but there are PRs |
| 12:49 |
Krock |
that's just a missing part - doesn't mean it's a bug |
| 12:50 |
nerzhul |
oh sorry then for cactus example, it's wrong in MTG context :p |
| 12:50 |
nerzhul |
it's good for cactus enabled server but not mtg :p |
| 12:50 |
nerzhul |
i moved all changelog from 0.4.4 to 0.4.10 here: http://dev.minetest.net/OldChangelog |
| 12:50 |
nerzhul |
the page is very huge to load and it doesn't help google ranking to have so many content :) |
| 12:52 |
paramat |
hmm well i won't bother with tin then, yet, but it was an omission and inconsistent with all other common metals |
| 12:59 |
sfan5 |
omission are not a bug, a feature freeze is a feature freeze |
| 12:59 |
nerzhul |
:) |
| 12:59 |
VanessaE |
oh just merge it, it's a trivial change I'm sure |
| 13:00 |
VanessaE |
sfan5: users will see minor omissions like that as a bug, actually. |
| 13:01 |
nerzhul |
it's just in mtg ? no core engine change to perform ? |
| 13:01 |
paramat |
well i'm not going to now, i'm happy i asked :] |
| 13:02 |
paramat |
just a mtg stairs mod change |
| 13:05 |
VanessaE |
then change it |
| 13:05 |
VanessaE |
don't leave something like that hanging that users will perceive as a bug or oversight, if possible. |
| 13:06 |
nerzhul |
sorry travis i trigger many builds because you are just shit :p |
| 13:07 |
nerzhul |
gitlab is far more better than travis... it permit to have pipeline and doesn't only check the build, you can build in every dsitro environment because you can use docker containers :p |
| 13:07 |
nerzhul |
but anyway, i'm crying because travis windows build is... complicated :p |
| 13:10 |
Wayward_One |
not sure if it's an issue with my setup, but a debug build fails with: CMakeFiles/minetest.dir/threading/event.cpp.o: file not recognized: File truncated |
| 13:10 |
|
kilbith joined #minetest-dev |
| 13:10 |
nerzhul |
this problem is your setup, generally a FS full |
| 13:11 |
Wayward_One |
ah, okay. |
| 13:17 |
|
lisac joined #minetest-dev |
| 13:25 |
|
Lejo_ joined #minetest-dev |
| 13:40 |
|
DS-minetest joined #minetest-dev |
| 13:41 |
|
octacian joined #minetest-dev |
| 13:50 |
|
Hunterz joined #minetest-dev |
| 14:00 |
nerzhul |
merging #5815 in ~10 mins |
| 14:00 |
ShadowBot |
https://github.com/minetest/minetest/issues/5815 -- Enhance ABM performance a little bit by removing two std::set copy by nerzhul |
| 14:01 |
nerzhul |
VanessaE, you last comment show me UDP and Linux implementation checksuming and many network implementation triggers bad network results on minetest on IPv6 |
| 14:02 |
nerzhul |
i think many packets are lost and mientest networking code is just waiting for those lost packets |
| 14:02 |
nerzhul |
i think you should remove IPv6 from your server unless we fixed the network implementation |
| 14:03 |
nerzhul |
this pr #5767 needs approvals for a bugfix, if somes can look at |
| 14:03 |
ShadowBot |
https://github.com/minetest/minetest/issues/5767 -- Don't add damage flash while punch texture modifier is active by stujones11 |
| 14:03 |
VanessaE |
nerzhul: but, I don't have ipv6 at home, so why would it time out for me? |
| 14:04 |
nerzhul |
because of the other guy making timeout on yoyur server i think |
| 14:04 |
VanessaE |
ok. |
| 14:05 |
nerzhul |
as i see we are using select listener instead of epoll/kqueue and i think this doesn't help if behind the networking code waits for missing packet segments... if i get time, i think 0.5 is time to rewrite the whole networking protocol using better code and multiple threads, better than now |
| 14:05 |
nerzhul |
but it's not simple, it's very tricky |
| 14:05 |
VanessaE |
ok, ipv6 disabled. |
| 14:07 |
VanessaE |
ok, that didn't fix it. |
| 14:07 |
VanessaE |
still getting timeouts |
| 14:08 |
nerzhul |
i think it doesn't stop all, but tell me if it enhance your experience |
| 14:08 |
VanessaE |
it doesn't. |
| 14:08 |
VanessaE |
however, |
| 14:08 |
nerzhul |
mobile devices connecting to minetest is a pain, remember that, if you connect using 2G with minetest you will jsut trigger high latency in the server... |
| 14:08 |
|
cx384 joined #minetest-dev |
| 14:08 |
VanessaE |
wait, there is something here. |
| 14:09 |
VanessaE |
when it times out for one client, other clients see map loading lag |
| 14:09 |
VanessaE |
no they don;'t |
| 14:09 |
VanessaE |
yeah, definitely not the server end |
| 14:09 |
VanessaE |
I just ran a good 500m down a road that I know for sure wasn't loaded yet, while watching another client instance time out |
| 14:09 |
VanessaE |
there was no lag at all |
| 14:10 |
nerzhul |
VanessaE, yes, this timeout on one client triggering lag for all every occurs since... ever |
| 14:10 |
VanessaE |
(that other one was probably actually server lag) |
| 14:10 |
nerzhul |
it's possible to DoS any minetest server using 1 client and very very slow packet sending... :( |
| 14:10 |
VanessaE |
well that surely needs to be fixed |
| 14:10 |
VanessaE |
before someone uses it as an attack |
| 14:12 |
VanessaE |
some people have been speculating that CSM could be a cause |
| 14:12 |
VanessaE |
but I don't see how. |
| 14:13 |
nerzhul |
it's ridiculous :p |
| 14:13 |
VanessaE |
so what do I do now>? |
| 14:13 |
nerzhul |
pray ... |
| 14:13 |
VanessaE |
I was having huge popularity with this server just a week ago |
| 14:13 |
nerzhul |
i don't know how to fix that |
| 14:13 |
VanessaE |
then suddenly it just ain't working right |
| 14:14 |
VanessaE |
then I suggest 0.4.16 not be released until it's fixed. |
| 14:14 |
VanessaE |
honestly |
| 14:14 |
nerzhul |
it's not possible i think i'm looking for the code i just think we can trash the code and rewrite it |
| 14:16 |
nerzhul |
if there is a problem it's in "void ConnectionReceiveThread::receive()" |
| 14:16 |
nerzhul |
but this is a very sensible loop |
| 14:17 |
nerzhul |
i think selects waits too many time |
| 14:17 |
nerzhul |
50 ms |
| 14:18 |
nerzhul |
the server loop is 100 ms... it's very long |
| 14:18 |
VanessaE |
one thing I know for sure is it was working great last week, then I updated, and in so doing I went newer than that last map version bump, so now I can't roll back to an older version to test it |
| 14:18 |
nerzhul |
yeah it's anoying, can you reproduce on a test server where map is not important ? |
| 14:18 |
VanessaE |
I cannot. |
| 14:19 |
VanessaE |
remember, I can't reproduce the issue with any of my other servers, it's only affecting the one heavy server |
| 14:19 |
VanessaE |
(so by extension, I won't be able to reproduce it with a test instance either :P ) |
| 14:19 |
|
Fritigern joined #minetest-dev |
| 14:19 |
VanessaE |
the odd thing is, TWO of my servers are "heavy", in the sense of mod content, but only one of them is being affected by thids |
| 14:20 |
VanessaE |
this* |
| 14:20 |
nerzhul |
it's why i said it's targetted |
| 14:20 |
VanessaE |
yeah |
| 14:20 |
VanessaE |
ok let's try it |
| 14:21 |
VanessaE |
I'm kicking everyone off of the affected server |
| 14:21 |
VanessaE |
so it'll just be me, and you know I have a fast pipe and a good PC |
| 14:21 |
VanessaE |
so this won't be "slow" |
| 14:23 |
VanessaE |
nope. |
| 14:23 |
VanessaE |
that's not it either. |
| 14:24 |
VanessaE |
I can't get more than 3 or 4 client instances to stay signed into the affected server, even when I'm the only one there |
| 14:24 |
nerzhul |
i think i understand where the problem can be |
| 14:24 |
VanessaE |
wait, not done |
| 14:25 |
nerzhul |
but not sure |
| 14:25 |
VanessaE |
(had to kick someone off) |
| 14:25 |
nerzhul |
yeah i understand the problem but it's not fixeable without risking major network break |
| 14:26 |
nerzhul |
if one client lag, ti's ... normal... with current code to make everybody lag |
| 14:26 |
VanessaE |
ok. and yeah, still timeouts even with just me for sure. I get all the way to a grey screen, black HUD elements, and some red text at the top telling me I timed out |
| 14:26 |
nerzhul |
because on each packet reassemble we try to read data for everboy |
| 14:26 |
nerzhul |
everybody |
| 14:27 |
nerzhul |
if someone is laggy we wait for him |
| 14:27 |
VanessaE |
yeah but this was all me this time. no one online but me. no way am I lagging if I can also do the same on the other heavy server |
| 14:28 |
nerzhul |
if you lag alone, maybe it's you the lag problematic client |
| 14:28 |
nerzhul |
for somewhat reason |
| 14:28 |
VanessaE |
nope |
| 14:28 |
VanessaE |
because others get the problem when I'm not online as well |
| 14:28 |
nerzhul |
unfortunately, viewing the current code i cannot fix it |
| 14:29 |
nerzhul |
it needs to rewrite all the underlying network to fix it |
| 14:29 |
VanessaE |
but why was it working last week? if this was some targeted attack, wouldn't kicking everyone off have stopped it? |
| 14:29 |
nerzhul |
you kick sessin |
| 14:29 |
nerzhul |
session* |
| 14:29 |
nerzhul |
the problem is not on player session it's on socket listener |
| 14:29 |
VanessaE |
mmmh |
| 14:29 |
nerzhul |
it's not in minetest protocol itself, it's in the way we read the socket |
| 14:29 |
VanessaE |
ok |
| 14:30 |
nerzhul |
we have one thread which listens and reassemble packet 1 client per 1 cleint as i understand, then if one client is very slow, minetest packet reassemble will be very slow and packet are not sent to server thread |
| 14:30 |
nerzhul |
after the C++11 switch i will work on it i think ,because i don't like our session system, and our network model can be simplified |
| 14:31 |
VanessaE |
so you're saying I'd need to reboot the server instance to test this? |
| 14:31 |
nerzhul |
no no |
| 14:31 |
VanessaE |
(to ensure all sockets are closed) |
| 14:31 |
nerzhul |
hmmm yes, you can reboot the service for cleaning up listeners |
| 14:31 |
nerzhul |
but it's udp, there is no client state |
| 14:32 |
nerzhul |
but maybe it can help you to purge a laggy client if you wait 10sec it should take a connection reset and his very slow client wil lbe discarded |
| 14:32 |
VanessaE |
already did that. |
| 14:32 |
nerzhul |
(it's why i like TCP, kernel manage that and present us a valid socket...) |
| 14:32 |
nerzhul |
erf |
| 14:32 |
VanessaE |
I had all users sign off, save one, which I kicked off. and it took me way longer than 10s to start my test |
| 14:32 |
nerzhul |
then i got no solution for release sorry |
| 14:33 |
VanessaE |
tell me something, why was the map format bump done again? |
| 14:33 |
nerzhul |
ask sfan5 |
| 14:33 |
nerzhul |
it's for private meta if i remember |
| 14:33 |
sfan5 |
yes |
| 14:33 |
VanessaE |
why can't the map just load without the feature? |
| 14:33 |
VanessaE |
I mean, literally no mods use it yet |
| 14:34 |
VanessaE |
i.e. why isn't there some way for me to roll back to a known good commit if only for testing? |
| 14:34 |
VanessaE |
(the only commits I am sure of predate the map version bump) |
| 14:34 |
sfan5 |
because that is how version bumps have always worked |
| 14:34 |
VanessaE |
yes, I get that |
| 14:34 |
sfan5 |
you can patch the older commit to understand the new map format if you want |
| 14:35 |
VanessaE |
but basically now I can't bisect this |
| 14:35 |
sfan5 |
unfortunate, but that's how it is |
| 14:36 |
VanessaE |
I see. |
| 14:36 |
sfan5 |
i'll make you a patch to allow reading new maps in older versions so you can bisect this |
| 14:36 |
VanessaE |
no, don't bother |
| 14:37 |
sfan5 |
why not? |
| 14:37 |
VanessaE |
because imho the proper way would be for one of you guys to download my world and run tests locally |
| 14:37 |
VanessaE |
I'm certain to break something irreparably if I do it. |
| 14:38 |
VanessaE |
I know the file is huge, but there's really no other way |
| 14:38 |
sfan5 |
there http://sprunge.us/IUTE?diff |
| 14:39 |
|
est31 joined #minetest-dev |
| 14:39 |
VanessaE |
ehm, can't copy&paste that. line numbers. |
| 14:40 |
est31 |
so given this PR, it seems I can turn off ubuntu precise nightly builders? |
| 14:40 |
est31 |
https://github.com/minetest/minetest/pull/5814 |
| 14:40 |
sfan5 |
remove the ?diff from the URL |
| 14:40 |
VanessaE |
right. |
| 14:41 |
sfan5 |
est31: when it's merged, yes |
| 14:41 |
est31 |
i'm not watching the minetest project any more so closely |
| 14:42 |
est31 |
I dont want to check often until the pr is merged |
| 14:42 |
sfan5 |
someone else should have access to travis too |
| 14:42 |
sfan5 |
then not yet |
| 14:42 |
est31 |
why do you hang so much on launchpad |
| 14:42 |
est31 |
12.04 is EOL |
| 14:42 |
est31 |
officially |
| 14:42 |
est31 |
if you are using it you are like an XP user |
| 14:42 |
est31 |
(sorry Krock) |
| 14:43 |
sfan5 |
??? |
| 14:43 |
est31 |
s/launchpad/precise/ |
| 14:43 |
sfan5 |
I am not saying that we should keep using precise |
| 14:43 |
est31 |
well in any case, if you dont know how to disable it, you can just pm me via irc once the pr is merged |
| 14:44 |
sfan5 |
why does this pr even relate to it |
| 14:44 |
est31 |
c55 has access |
| 14:44 |
est31 |
shadow ninja too |
| 14:44 |
est31 |
sfan5: c++11?? |
| 14:44 |
est31 |
the gcc on ubuntu precise doesnt support it |
| 14:44 |
est31 |
not fully |
| 14:44 |
sfan5 |
is that a problem? 12.04 is EOL |
| 14:44 |
* DS-minetest |
found something meh |
| 14:45 |
est31 |
sfan5: so I can remove it then? |
| 14:45 |
DS-minetest |
https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L3850 and https://github.com/minetest/minetest/blob/master/builtin/game/item.lua#L218 |
| 14:45 |
sfan5 |
yes, disregard what i said before i misunderstood |
| 14:45 |
est31 |
ok |
| 14:45 |
est31 |
fine, doing it then |
| 14:45 |
nerzhul |
est31, no problem i can pr you, but it's not for now, this PR is for next release and not finished, i'm waiting for new leveldb library snapshot to make windows cross build works |
| 14:45 |
VanessaE |
wait a minute, |
| 14:45 |
VanessaE |
nerzhul: |
| 14:45 |
VanessaE |
Client: Don't send `TOSERVER_RECEIVED_MEDIA` since it's not used anymore |
| 14:45 |
VanessaE |
Server: Handle `TOSERVER_RECEIVED_MEDIA` using `Server::handleCommand_Deprecated` |
| 14:46 |
VanessaE |
commit ff1ef67dcc |
| 14:46 |
nerzhul |
? |
| 14:46 |
VanessaE |
could this be why? it came after a known, working commit |
| 14:46 |
DS-minetest |
it says placer can be any objectref or nil but builtin needs placer to be a player :P |
| 14:48 |
nerzhul |
this packet was removed on MT in 2014 |
| 14:48 |
VanessaE |
ok. |
| 14:48 |
est31 |
so I've disabled ubuntu precise builders now |
| 14:48 |
nerzhul |
ty est31 |
| 14:49 |
nerzhul |
VanessaE, diff show the client now doesn't send a useless packet server ignore since a long time |
| 14:49 |
nerzhul |
and red-001 fix the server handler from not handled to deprecated |
| 14:49 |
|
est31 left #minetest-dev |
| 14:49 |
nerzhul |
then a warning should be shown in your logs if you receive this packet |
| 14:49 |
VanessaE |
just checking. I'm skimming the commit log to make sure I'm not about to waste a bunch of time |
| 14:51 |
VanessaE |
I really wish git log would show me the date a commit was actually pushed, rather than when it was started |
| 15:08 |
red-001 |
not warning the log level for it is set to info iirc |
| 15:10 |
nerzhul |
red-001, sounds better :p |
| 15:11 |
VanessaE |
sfan5, nerzhul: without having tried to bisect yet, I am certain it worked at commit 73de17af (Apr. 16), probably also worked fine at 57eaf62c, but broke some time between those and 9b5effff. The private metadata commit happened on May 10 |
| 15:12 |
nerzhul |
network doesn't change in this interval :( |
| 15:12 |
VanessaE |
(9b5... was May 17) |
| 15:15 |
VanessaE |
I know 73d... worked because that's when I last did a clean clone. didn't touch it again for a month, when it pulled to 9b5... then I rolled it back to 57e... for whatever reason, then went forward from there. `git log -g` is a little confusing |
| 15:16 |
VanessaE |
73d... predates the map version bump, so I'd need to use that patch sfan5 made. but I feel the need to understand wtf is really going on first. |
| 15:20 |
|
DS-minetest joined #minetest-dev |
| 15:28 |
|
Fixer joined #minetest-dev |
| 15:35 |
VanessaE |
nerzhul: --> github |
| 15:36 |
|
Grandolf joined #minetest-dev |
| 15:40 |
|
cx384 joined #minetest-dev |
| 15:40 |
* VanessaE |
applies sfan5's patch and rolls back to the old commit |
| 15:40 |
VanessaE |
(well, in opposite order :P ) |
| 15:41 |
|
cx384 joined #minetest-dev |
| 15:41 |
VanessaE |
fuck |
| 15:41 |
VanessaE |
can't build it |
| 15:42 |
VanessaE |
sfan5: https://pastebin.com/QFV18Rs6 |
| 15:42 |
VanessaE |
I applied your patch over 73de17af |
| 15:43 |
sfan5 |
oh yea forgot about those |
| 15:43 |
sfan5 |
change it to meta->deSerialize(is, 1); |
| 15:44 |
VanessaE |
NodeMetadata *data = new NodeMetadata(item_def_mgr); |
| 15:44 |
VanessaE |
- data->deSerialize(is); |
| 15:44 |
VanessaE |
+ meta->deSerialize(is, 1); |
| 15:48 |
VanessaE |
nope. that ain't it either. |
| 15:49 |
VanessaE |
https://pastebin.com/2N5YkhPi |
| 15:49 |
sfan5 |
VanessaE: in rollback_interface.cpp not nodemetadata.cpp |
| 15:50 |
sfan5 |
lemme make you a new patch |
| 15:50 |
VanessaE |
thanks. |
| 15:52 |
sfan5 |
VanessaE: http://sprunge.us/HYIP |
| 15:52 |
|
halt_ joined #minetest-dev |
| 15:53 |
VanessaE |
trying... |
| 15:54 |
VanessaE |
well it built. |
| 15:54 |
* VanessaE |
runs it.... |
| 15:55 |
VanessaE |
2017-05-25 17:55:27: ERROR[Main]: ServerError: AsyncErr: environment_Step: Runtime error from mod 'pipeworks' in callback environment_Step(): ERROR: MapBlock format not supported |
| 15:56 |
VanessaE |
(no clue why it shows that mod at fault, maybe because it force-loads blocks or something) |
| 15:56 |
kilbith |
had the same error |
| 15:56 |
sfan5 |
derp |
| 15:56 |
sfan5 |
lemme make you yet another patch |
| 15:56 |
VanessaE |
ok |
| 16:02 |
sfan5 |
VanessaE: here http://sprunge.us/IAMH |
| 16:02 |
sfan5 |
if that still doesn't work i'll make another one and actually test it before |
| 16:04 |
VanessaE |
applied and building |
| 16:06 |
VanessaE |
builds ok |
| 16:06 |
* VanessaE |
runs it... |
| 16:06 |
|
Raven262 joined #minetest-dev |
| 16:13 |
VanessaE |
sfan5, nerzhul I think I can safely say it is working properly at 73de17afa82 with sfan5's last patch |
| 16:13 |
VanessaE |
I now have EIGHT client instances of my own online with zero issue, and cheapie has three more on |
| 16:13 |
VanessaE |
and there's more coming |
| 16:14 |
nerzhul |
okay, now you can bisect and we can find the problem, i hope it's not a coincidence |
| 16:15 |
VanessaE |
mmmmh |
| 16:17 |
VanessaE |
one person is saying he can't connect with your latest build, sfan5 |
| 16:18 |
sfan5 |
sounds like someone broke something clientside |
| 16:21 |
VanessaE |
one of the guys says he can't even get in with 0.4.15-release, but I think he's having a local problem |
| 16:21 |
VanessaE |
care to try it? |
| 16:21 |
VanessaE |
daconcepts.com 30001, expect ~40 MB of medioa |
| 16:21 |
VanessaE |
media* |
| 16:21 |
|
QwertyDragon joined #minetest-dev |
| 16:21 |
nerzhul |
VanessaE, due to the gap between last and patched version, just bisect to help us find which commit |
| 16:22 |
VanessaE |
nerzhul: I am not sure that I can bisect given how much resources it takes to test :-/ |
| 16:24 |
|
QwertyDragon_ joined #minetest-dev |
| 16:25 |
|
QwertyDragon_ joined #minetest-dev |
| 16:25 |
nerzhul |
bisect permit to have a commit range, it just take time :) |
| 16:26 |
nerzhul |
seems you are the only server owner to report this problem ti's difficult |
| 16:30 |
VanessaE |
ok the connection issue for those guys was just ipv6 being disabled |
| 16:30 |
VanessaE |
so ignore that part. |
| 16:31 |
VanessaE |
I turned it back on and they can connect okay now. |
| 16:31 |
VanessaE |
the timeouts are definitely fixed by using this older commit |
| 16:32 |
nerzhul |
and it seems to be linked with server, it's easier to track |
| 16:32 |
VanessaE |
can you try connecting please? |
| 16:33 |
VanessaE |
maybe you can ... I dunno.....wireshark it? |
| 16:33 |
nerzhul |
on which commit are you ? |
| 16:34 |
VanessaE |
73de17af plus sfan5's patch to make it load my "too new" world |
| 16:34 |
nerzhul |
erf 342 files changed :p |
| 16:34 |
nerzhul |
236 files in src/ |
| 16:35 |
nerzhul |
bisecting is really required |
| 16:35 |
VanessaE |
bisecting may not be possible here without help |
| 16:35 |
VanessaE |
(as in, someone who can connect lots of clients at once, so my machine doesn't melt down) |
| 16:36 |
nerzhul |
first revert sfan5 patch |
| 16:36 |
nerzhul |
if it's not a git patch a single git reset --hard on your tree is good |
| 16:36 |
nerzhul |
then go to current master: git fetch && git checkout origin/master |
| 16:36 |
nerzhul |
then when you are sure you are on master |
| 16:37 |
nerzhul |
git bisect start && git bisect bad HEAD && git bisect good <commitid_good_you_found> |
| 16:37 |
VanessaE |
nono |
| 16:38 |
VanessaE |
sfan5's patch is necessary here, without it, my map is "too new" to run the older MT commits, because it was altered by the map version bump that came with the "private metadata" commit |
| 16:38 |
nerzhul |
how do you applied it ? |
| 16:38 |
nerzhul |
by hand ? |
| 16:38 |
VanessaE |
yeah, patch -p1 < file.patch |
| 16:39 |
nerzhul |
okay then |
| 16:39 |
nerzhul |
instead of reset, type git stash |
| 16:39 |
VanessaE |
so not by hand, but not a git commit either :P |
| 16:39 |
nerzhul |
it will save the diff |
| 16:39 |
nerzhul |
we can apply it later |
| 16:39 |
nerzhul |
and then start bisect |
| 16:39 |
nerzhul |
git will place your tree in bisect commit, if this commti is before sfan5 change, type git stash apply, else jsut comiple |
| 16:40 |
VanessaE |
ok that sounds fair |
| 16:41 |
nerzhul |
git is a nice tool heh, it have all you need for debugging |
| 16:41 |
VanessaE |
I really wish I could get you to try this on your local box though |
| 16:41 |
nerzhul |
i cannot reproduce it |
| 16:41 |
VanessaE |
download the world, run a server with it, connect as many clients as you can at once until it barfs |
| 16:42 |
nerzhul |
but bisect will help us to find the problem when we have at least a little commit range, and better the cmmit |
| 16:42 |
VanessaE |
ok |
| 16:42 |
nerzhul |
i'm on freebsd maybe i will not be affected |
| 16:42 |
VanessaE |
good point |
| 16:42 |
VanessaE |
could be specific to linux on the server side |
| 16:42 |
VanessaE |
I'm still having some of the users beat the crap out of it right now though |
| 16:43 |
VanessaE |
I'll try to bisec shortly |
| 16:43 |
nerzhul |
yeah take your time, after the bisect and the test |
| 16:43 |
nerzhul |
type git bisect bad if the commit is bad, else git commit good |
| 16:43 |
nerzhul |
and continue while you are on the problematic commit :) |
| 16:45 |
nerzhul |
s/while/until/g |
| 16:46 |
VanessaE |
ok, starting to bisest |
| 16:46 |
VanessaE |
bisect* |
| 16:47 |
|
paramat joined #minetest-dev |
| 16:49 |
VanessaE |
sfan5: can I just always apply your patch while I'm doing this? |
| 16:50 |
nerzhul |
i think you can unless current commit is after the sfan5 merge |
| 16:50 |
sfan5 |
yes |
| 16:50 |
VanessaE |
ok |
| 16:50 |
nerzhul |
type git stash apply |
| 16:50 |
nerzhul |
it's in the git-stash :) |
| 16:50 |
VanessaE |
eh, it doesn't like that. |
| 16:51 |
nerzhul |
it depend when you are in the history just check :) |
| 16:51 |
VanessaE |
https://pastebin.com/gNcXE96m |
| 16:52 |
VanessaE |
screw it, I'll just `patch` it in. |
| 16:53 |
nerzhul |
seems good |
| 16:53 |
nerzhul |
it patched without conflict |
| 16:53 |
VanessaE |
well you sure wouldn't know it from the bold red text. |
| 16:53 |
VanessaE |
ok |
| 16:53 |
VanessaE |
I'll do it your way then |
| 16:53 |
nerzhul |
before doing git bisect good/bad don't forget to git stash |
| 16:54 |
VanessaE |
ok |
| 17:05 |
VanessaE |
nerzhul: root daconcepts:/home/minetest/minetest_core# git bisect bad ... git stash apply... error: Your local changes to the following files would be overwritten by merge: (the files sfan5's patch modifies) |
| 17:06 |
sfan5 |
forgot the git stash before `git bisect bad`? |
| 17:06 |
sfan5 |
in any case git reset HEAD --hard && git stash apply should work |
| 17:06 |
VanessaE |
yep, that got it |
| 17:06 |
nerzhul |
sfan5, yes |
| 17:15 |
VanessaE |
it seems I have to do that with each cycle, but whatever. as long as it works. |
| 17:21 |
nerzhul |
yeah it's the bisect principle |
| 17:21 |
nerzhul |
but at each loop step you divide commit range by 2 |
| 17:22 |
VanessaE |
I mean the `reset --hard HEAD` part. |
| 17:22 |
VanessaE |
meh, as long as the code is right and it works, idc :P |
| 17:22 |
VanessaE |
git is all black magic |
| 17:23 |
nerzhul |
white magic :D |
| 17:24 |
Fixer |
greyish magic |
| 17:25 |
|
Fritigern joined #minetest-dev |
| 17:26 |
Krock |
black magic with poisoned injections called "patches" and blood extracts named "diffs". We're all witches here. |
| 17:28 |
VanessaE |
heh |
| 17:31 |
VanessaE |
I'm about 4 cycles into the bisect, fwiw |
| 17:33 |
nerzhul |
4 cycles ? you divide commit range by 8, it's nice :) |
| 17:33 |
nerzhul |
16 * |
| 17:33 |
nerzhul |
2^4 |
| 17:33 |
nerzhul |
what is the current bad commit ? |
| 17:33 |
VanessaE |
moment |
| 17:34 |
VanessaE |
https://pastebin.com/cpQbAJZf |
| 17:35 |
paramat |
with copyright credits at the top of cpp files, is the second year always updated to the current year even if that dev has not contributed to the file in that year? |
| 17:35 |
VanessaE |
and I'm about to mark the current commit as bad also |
| 17:35 |
VanessaE |
paramat: traditionally it just has to reflect the last date the file was modified, but it's a good idea to always keep it the year of release anyway |
| 17:36 |
paramat |
thanks |
| 17:36 |
nerzhul |
190 files changes between 73de17afa821ccea84a119096b267e05a77db3ff ad9fcf859ec2347325830e09504ae96968b51ea8 |
| 17:36 |
nerzhul |
erf :p |
| 17:36 |
VanessaE |
yeah, still whittling it down |
| 17:37 |
nerzhul |
at least 12 days of development |
| 17:37 |
nerzhul |
:) |
| 17:37 |
|
Grandolf joined #minetest-dev |
| 17:41 |
nerzhul |
in this interval there is plenty of commits |
| 17:41 |
VanessaE |
https://pastebin.com/6i0x9CaY |
| 17:41 |
VanessaE |
there it is. |
| 17:41 |
VanessaE |
or so says bisect anyway |
| 17:41 |
nerzhul |
you mean this commit is the bad commit ? |
| 17:42 |
VanessaE |
that's what bisect tells me |
| 17:42 |
VanessaE |
https://pastebin.com/ZqAJeN7W |
| 17:42 |
VanessaE |
there's the full log |
| 17:42 |
nerzhul |
ok |
| 17:42 |
Grandolf |
Fix jitter when walking against the sneak limits. |
| 17:42 |
Grandolf |
Fix damage evading on sneak ladders. |
| 17:42 |
nerzhul |
i see what has been removed |
| 17:42 |
VanessaE |
(well git log followed by the output of the final "bad") |
| 17:42 |
Grandolf |
r u removing the "sneak glitch"? |
| 17:42 |
VanessaE |
Grandolf: old issue, please don't. |
| 17:42 |
nerzhul |
the player got teleported flag was removed |
| 17:42 |
nerzhul |
what is the side effet now |
| 17:42 |
VanessaE |
for the love of all that is hold and decent, don't ask about sneak |
| 17:43 |
nerzhul |
it's purely client side |
| 17:43 |
nerzhul |
you are recompiling client or server ? |
| 17:43 |
VanessaE |
server. |
| 17:43 |
VanessaE |
it can't be client-side unless it's something that started with 0.4.15-release maybe |
| 17:43 |
nerzhul |
then there is something wrong :p |
| 17:43 |
nerzhul |
this commit is purely clietn side |
| 17:44 |
VanessaE |
lemme roll back to one commit before it and see what happenbs. |
| 17:44 |
VanessaE |
-b |
| 17:45 |
nerzhul |
maybe you set a coommit as good whereas it wasn't |
| 17:45 |
VanessaE |
I'm sure I didn't. |
| 17:46 |
nerzhul |
to be sure, you can restart bisecting, but declare f9fdb48dc85e53253ef93972355d3c7bc9d0ffb0 as first bad, and the previous known good commit as good |
| 17:46 |
VanessaE |
ok, after this current test |
| 17:47 |
VanessaE |
I'm testing the server at bd921a79 just to see what it does. |
| 17:47 |
nerzhul |
should be bad too, as the git log seems to indicate this is not related :p |
| 17:49 |
VanessaE |
ok, bd921a79 is indeed also bad.\ |
| 17:49 |
VanessaE |
mmmh |
| 17:51 |
VanessaE |
ok, restarting the bisect from ^^^ your commit as bad, and the known good one as...good. :) |
| 17:52 |
red-001 |
Grandolf, old news |
| 17:53 |
nerzhul |
VanessaE, at least there is less commits to bisect |
| 17:53 |
VanessaE |
truew |
| 17:53 |
Grandolf |
thought it had already been added |
| 17:53 |
Grandolf |
red-001 ^ |
| 17:53 |
VanessaE |
Grandolf: please in the name of all that is good and/or holy, don't talk about sneak |
| 17:54 |
VanessaE |
the first rule of sneak is, don't talk about sneak |
| 17:54 |
* Grandolf |
sneaks outa chat |
| 17:54 |
VanessaE |
wiseass :P |
| 17:54 |
red-001 |
we already had a flame war between the devs and the one that shall not be mentioned over this |
| 17:55 |
red-001 |
Grandolf, lets move this to #minetest |
| 17:55 |
red-001 |
it's not ontopic here |
| 17:55 |
Grandolf |
ok |
| 18:10 |
nerzhul |
VanessaE, news? |
| 18:11 |
VanessaE |
still working on it |
| 18:12 |
VanessaE |
fwiw, each test I get about 20 clients online at a time, if the commit is good |
| 18:12 |
VanessaE |
if bad, around 4. |
| 18:12 |
VanessaE |
(6 of my own, several from other players who are helping me test it) |
| 18:23 |
Fixer |
i can help you too |
| 18:23 |
Fixer |
right now |
| 18:23 |
Fixer |
where to connect? |
| 18:24 |
Fixer |
or nevermind %) |
| 18:29 |
VanessaE |
nerzhul: https://pastebin.com/0YrMUwQK |
| 18:30 |
nerzhul |
lol |
| 18:30 |
Fixer |
o.O |
| 18:30 |
VanessaE |
at least that's got changes in connection.cpp |
| 18:30 |
VanessaE |
and other places |
| 18:30 |
nerzhul |
yeah |
| 18:30 |
nerzhul |
letme a minute |
| 18:30 |
VanessaE |
ok |
| 18:31 |
nerzhul |
it seems this PR is correct, the problematic PR should be before |
| 18:31 |
VanessaE |
PR? or commit? |
| 18:31 |
VanessaE |
what shall I do here? |
| 18:31 |
nerzhul |
checkout this |
| 18:31 |
nerzhul |
b662a4577d692329b9ca83525e6039f2ddcd1ac1 |
| 18:32 |
nerzhul |
and tell me if it is problematic |
| 18:32 |
nerzhul |
both commits are related |
| 18:32 |
VanessaE |
ok, checked out, and the patch is already applied at that commit. building.... |
| 18:33 |
nerzhul |
hopefully this kind of commit can be removed after release std::chrono is far more better, but if it's the problem we should find and fix it |
| 18:34 |
VanessaE |
running it... |
| 18:35 |
nerzhul |
i don't think it's the problem the commit as it doesn't seems to be called by connetion |
| 18:35 |
nerzhul |
but second uses it |
| 18:35 |
nerzhul |
i reviewed the commit anotehr time and it seems exactly same, just pass u32 to u64 and fix indents |
| 18:36 |
nerzhul |
i don't see why Krock type change can cause the problem, except if it works due to a integer overflow ! |
| 18:37 |
nerzhul |
and if it's the case... PAN |
| 18:37 |
nerzhul |
if the current test works, please test 9762650f978cc7bae78861b70a051b26cc5e2dc6 and ensure me it works or not on this |
| 18:37 |
VanessaE |
testing it |
| 18:37 |
VanessaE |
er testing b66... I mean |
| 18:38 |
nerzhul |
y |
| 18:38 |
VanessaE |
b662a457 appears to be good. |
| 18:38 |
VanessaE |
I'll try 9762650f now? |
| 18:41 |
nerzhul |
y |
| 18:41 |
VanessaE |
ok. |
| 18:41 |
nerzhul |
if last is good it confirms we have a big problem :p |
| 18:41 |
VanessaE |
building... |
| 18:43 |
VanessaE |
running it... |
| 18:43 |
|
RobbieF left #minetest-dev |
| 19:00 |
|
H-H-H joined #minetest-dev |
| 19:00 |
VanessaE |
nerzhul: 9762650f appears to be *mostly* good. 23 client instances online, 8 of which are mine (others are from a few players doing multiple instances also). One user reports that he can't get more than just 1 to connect (which is among those 23). could be on his end. |
| 19:01 |
H-H-H |
hey guys quick question :) is it possible to add a key for use in a mod from a mod or would it need to be done in the c++ ? |
| 19:01 |
nerzhul |
VanessaE, okay then... |
| 19:02 |
VanessaE |
nerzhul: bad news? :) |
| 19:02 |
nerzhul |
3db66b453152c3610b858aa1650e1ab3f545a430 |
| 19:02 |
nerzhul |
sorry |
| 19:02 |
nerzhul |
no yes it's the good commit :p |
| 19:02 |
|
nerzhul left #minetest-dev |
| 19:03 |
|
nerzhul joined #minetest-dev |
| 19:03 |
nerzhul |
f727f54192644f6427ac1b2c86df8c64c7c5fdf0 seems to break but don't know why atm |
| 19:03 |
nerzhul |
can you just test 3db66b453152c3610b858aa1650e1ab3f545a430 to confirm it's broken |
| 19:03 |
VanessaE |
nooooooooooooo |
| 19:04 |
VanessaE |
I've got 26 clients online, don't make me spoil it when I'm so close to filling the server to its configured limit :) |
| 19:06 |
red-001 |
any luck finding the offending commit? |
| 19:07 |
VanessaE |
see above, red-001 |
| 19:07 |
nerzhul |
red-001, f727f54192644f6427ac1b2c86df8c64c7c5fdf0 seems to be faulty and ... tricky |
| 19:07 |
nerzhul |
but related |
| 19:08 |
red-001 |
huh |
| 19:10 |
VanessaE |
ok nerzhul, setting up to try 3db66b453152c3610b858aa1650e1ab3f545a430 |
| 19:11 |
VanessaE |
ok, that's built. running it... |
| 19:15 |
VanessaE |
nerzhul: 3db66b45 produces timeouts, or occasionally "init packet not recognized by server" |
| 19:15 |
nerzhul |
okay |
| 19:15 |
VanessaE |
this is good news>? :) |
| 19:15 |
nerzhul |
then f727f54192644f6427ac1b2c86df8c64c7c5fdf0 is faulty |
| 19:16 |
VanessaE |
ok |
| 19:16 |
nerzhul |
Krock, ^ |
| 19:16 |
Krock |
pong |
| 19:16 |
nerzhul |
it's your commit :) |
| 19:17 |
VanessaE |
nerzhul: ok, so I guess I should put the server at 9762650f for now? |
| 19:17 |
Krock |
that's interesting. |
| 19:17 |
nerzhul |
you can at least make your server working on 9762650f as it seems yes |
| 19:18 |
VanessaE |
ok, willdo |
| 19:18 |
nerzhul |
or keep upstream master and revert f727f54192644f6427ac1b2c86df8c64c7c5fdf0 |
| 19:18 |
nerzhul |
it will not produce conflict |
| 19:18 |
VanessaE |
ok, let me try THAT |
| 19:18 |
nerzhul |
OKAY |
| 19:19 |
Krock |
I bet someone gotta explain me how >more< space for a number can cause new problems |
| 19:20 |
VanessaE |
f727f541 reverted okay. building... |
| 19:20 |
VanessaE |
Krock: unless some value was over/underflowing and it just happened to work okay before? |
| 19:23 |
nerzhul |
Krock, as i said before youi connect, i hope network just don't work because there is overflow on an integer... |
| 19:24 |
Krock |
overflow detections are done by checking if the previous number is more than the new computed one. That's a generic check that works on all integer types |
| 19:24 |
nerzhul |
if VanessaE confirms current master works without f727f54192644f6427ac1b2c86df8c64c7c5fdf0 we should study it carefully and find what append |
| 19:24 |
nerzhul |
it's done when you did it, we are in C/C++ not java :p |
| 19:24 |
nerzhul |
if you overflow, 1111 + 0001 = 0000 |
| 19:25 |
nerzhul |
:) |
| 19:25 |
Krock |
ehm.. yes, I'm aware of that |
| 19:28 |
Krock |
boom found one missing occurence |
| 19:28 |
Krock |
connection.h L772 is still u32 |
| 19:29 |
Krock |
apparently I only ran my search expression for direct assignements |
| 19:29 |
nerzhul |
it's why IDE are better than grep |
| 19:29 |
nerzhul |
:) |
| 19:29 |
nerzhul |
in clion: find callees and that's all :p |
| 19:30 |
nerzhul |
and CTRL+MAJ+F if you are not sure it's only in code :p |
| 19:30 |
|
RobbieF joined #minetest-dev |
| 19:30 |
VanessaE |
nerzhul: confirmed. upstream HEAD with, reverted f727f541 -- works fine, 34 clients online. |
| 19:30 |
nerzhul |
nice |
| 19:30 |
Krock |
VanessaE, patch coming |
| 19:30 |
VanessaE |
Krock: ok |
| 19:30 |
nerzhul |
Krock, it's due to getTimeMs assigments ? |
| 19:31 |
nerzhul |
ouch we have same problem on client it seems |
| 19:31 |
Krock |
yes, m_last_timeout_check is still u32 |
| 19:31 |
nerzhul |
oh, exactly this bug |
| 19:31 |
nerzhul |
do you want me to fix all the occurences ? |
| 19:31 |
nerzhul |
there are many problems on client |
| 19:32 |
nerzhul |
s64 m_hovered_time = getTimeMs ... gtmlt |
| 19:32 |
Krock |
u32 BlinkStartTime intlGUIEditBox.h |
| 19:32 |
Krock |
should be u64 aswell |
| 19:32 |
nerzhul |
yes |
| 19:33 |
nerzhul |
and it's not the only case |
| 19:33 |
nerzhul |
there are many |
| 19:33 |
nerzhul |
u32 animation_timer = porting::getTimeMs() % 100000; in game.cpp |
| 19:33 |
nerzhul |
u32 timer = porting::getTimeMs() % 5000; |
| 19:33 |
nerzhul |
game.cpp too |
| 19:33 |
Krock |
err |
| 19:33 |
Krock |
are you serious? |
| 19:33 |
nerzhul |
delta = porting::getDeltaMs(m_hovered_time, porting::getTimeMs()); |
| 19:33 |
nerzhul |
in guiFormSpec menu |
| 19:33 |
Krock |
> % 5000 |
| 19:33 |
Krock |
u64 looks reasonable |
| 19:34 |
Krock |
/s |
| 19:34 |
nerzhul |
i hope compiler correctly strip it :) |
| 19:34 |
nerzhul |
u32 delta = porting::getDeltaMs(m_doubleclickdetect[0].time, porting::getTimeMs()); |
| 19:34 |
nerzhul |
in guiFormSpecMenu |
| 19:35 |
nerzhul |
m_internal_time = porting::getTimeMs() / 1000.f; in joystickcontroller |
| 19:35 |
nerzhul |
u32 delta = porting::getDeltaMs(m_key_events[0].down_time, porting::getTimeMs()); in touchscreengui |
| 19:35 |
Krock |
also unittests need some updating.. replacing the stuff |
| 19:36 |
nerzhul |
i don't find unittest problems where did you see them ? |
| 19:36 |
nerzhul |
test_connection.cpp is good |
| 19:36 |
nerzhul |
but many delta are not correct and can overflow |
| 19:36 |
Krock |
test.cpp: u32 t1 = porting::getTime(PRECISION_MILLI); |
| 19:37 |
nerzhul |
getTime has same problem... erf |
| 19:38 |
nerzhul |
u32 curr_time = porting::getTime(PRECISION_MILLI); |
| 19:38 |
nerzhul |
in map.cpp |
| 19:38 |
Krock |
only 49 hits here |
| 19:38 |
nerzhul |
10 getTime function usage |
| 19:38 |
nerzhul |
m_connection_time(porting::getTime(PRECISION_SECONDS)) |
| 19:38 |
nerzhul |
this is problematic too |
| 19:38 |
nerzhul |
m_connection_time is u32 |
| 19:38 |
nerzhul |
in remoteclient |
| 19:38 |
nerzhul |
u32 RemoteClient::uptime() const |
| 19:39 |
nerzhul |
u32 TimeTaker::getTimerTime() read getTime ... |
| 19:39 |
nerzhul |
sooooo many things to fix |
| 19:39 |
Krock |
well thanks. already found that 5 mins ago |
| 19:39 |
nerzhul |
:p |
| 19:39 |
nerzhul |
i'm waiting for your PR then and i will re-check after applying it |
| 19:40 |
VanessaE |
are we sure this'll work? ;) |
| 19:40 |
paramat |
awesome, great work |
| 19:40 |
nerzhul |
for connection pretty sure |
| 19:41 |
nerzhul |
overflowing the timeout timer seems a good reason for this bug |
| 19:41 |
nerzhul |
fixing it in all places permit to prevent other overflow with not undenstandable bugs;.. |
| 19:41 |
VanessaE |
yep, that sounds fair |
| 19:56 |
Krock |
nerzhul, #5818 |
| 19:56 |
ShadowBot |
https://github.com/minetest/minetest/issues/5818 -- Time: Change old `u32` timestamps to 64-bit by SmallJoker |
| 19:59 |
nerzhul |
there is one error to fix |
| 20:00 |
nerzhul |
the first fix precision second you changes it to ms |
| 20:01 |
Krock |
huh? |
| 20:01 |
Krock |
oh I see, thanks |
| 20:09 |
nerzhul |
VanessaE, https://github.com/minetest/minetest/pull/5818.patch |
| 20:09 |
nerzhul |
remove the revert on the problematic commit and apply this patch please |
| 20:10 |
nerzhul |
for me it's good, i will just verify if it's complete or not |
| 20:17 |
|
octacian_ joined #minetest-dev |
| 20:19 |
VanessaE |
I'll wait for you guys to test it a bit first :) |
| 20:21 |
nerzhul |
it's stable, it's just variable change |
| 20:21 |
nerzhul |
you can test it |
| 20:21 |
VanessaE |
ok |
| 20:24 |
nerzhul |
Krock, it's mergeable without problem i don't see any remaining entry |
| 20:25 |
VanessaE |
compile failure |
| 20:25 |
VanessaE |
https://pastebin.com/gdFPQ6Ue |
| 20:26 |
|
octacian__ joined #minetest-dev |
| 20:26 |
nerzhul |
looking, i missed it |
| 20:26 |
Krock |
totally forgot them. fixing |
| 20:26 |
VanessaE |
so yeah, I would suggest you guys test it ;) |
| 20:26 |
nerzhul |
travis told us |
| 20:27 |
nerzhul |
yeah pointers |
| 20:27 |
nerzhul |
debug stuff timers... |
| 20:29 |
nerzhul |
u32 updateareawaterpressure_time = 0; |
| 20:29 |
nerzhul |
u32 flowwater_pre_time = 0; |
| 20:29 |
nerzhul |
they can be removed, |
| 20:29 |
nerzhul |
they are not used |
| 20:29 |
nerzhul |
(you will modify the variables just upper :p |
| 20:31 |
Krock |
nerzhul, interestingly we thought about the same thing while looking at this problem |
| 20:32 |
Krock |
also noticed here that some variables are no longer used |
| 20:32 |
nerzhul |
oh i was just reviewing at the same time on clion :p |
| 20:32 |
nerzhul |
which IDE do you use ? VS ? |
| 20:32 |
Krock |
none :trollface: |
| 20:33 |
|
luk3yx left #minetest-dev |
| 20:37 |
nerzhul |
VanessaE, re-apply the patch please by re-downloading it |
| 20:37 |
VanessaE |
sure. |
| 20:39 |
paramat |
hehe waterpressure, finite water code? |
| 20:39 |
Fixer |
nah |
| 20:40 |
VanessaE |
that built ok |
| 20:49 |
VanessaE |
if anyone wants to help me load test that patch, |
| 20:49 |
VanessaE |
go to daconcepts.com 30001 |
| 20:49 |
VanessaE |
with as many client instances as you can safely run |
| 20:50 |
Krock |
is that the dreambuilder one? |
| 20:51 |
Krock |
yeah, just 207 (partwise heavy) mods :3 |
| 20:56 |
kilbith |
testing creative inventory' new design: https://lut.im/dsiMp4y5lY/oru0k0KRwaSHrHGu.gif |
| 20:59 |
VanessaE |
Krock: and yeah, it's dreambuilder |
| 21:04 |
VanessaE |
anyone else care to give it a try? |
| 21:14 |
rubenwardy |
oh nice, kilbith |
| 21:17 |
|
RobbieF left #minetest-dev |
| 21:18 |
nerzhul |
VanessaE, it works now ? |
| 21:18 |
VanessaE |
nerzhul: it seems to |
| 21:18 |
VanessaE |
but I'm having trouble getting you guys to actually sign on and try to push it to its limit :) |
| 21:18 |
VanessaE |
I've got 28 clients going now |
| 21:19 |
VanessaE |
need more :P |
| 21:20 |
nerzhul |
it's sufficient |
| 21:20 |
VanessaE |
ok |
| 21:21 |
Hijiri |
Does anyone see any potential problems implementing dimensions in the map db as separate tables rather than part of the mapblock hash? |
| 21:21 |
Hijiri |
"hash" |
| 21:21 |
Hijiri |
It would avoid increasing the map size too much |
| 21:30 |
VanessaE |
Kray: thanks again for the patch. Survival is happy now :) |
| 21:31 |
Kray |
wat |
| 21:31 |
nerzhul |
thanks Kray :D errr Krock xD |
| 21:31 |
nerzhul |
poor SmallJoker :p |
| 21:31 |
nerzhul |
Kray please change your nick it break autocompletion on core devs xD |
| 21:32 |
nerzhul |
erf MacOSX build is not happy |
| 21:34 |
VanessaE |
er Krock ^^^^ |
| 21:34 |
VanessaE |
sorry Kray. tab-complete error |
| 21:34 |
nerzhul |
i hope Krock can look at this tomorrow and fix it, but i don't know how to fix it, sfan5 and idea ? |
| 21:34 |
nerzhul |
https://travis-ci.org/minetest/minetest/jobs/236149634 |
| 21:34 |
nerzhul |
i don't know why there is a problem here |
| 21:52 |
Fixer |
had few crashes during closing of rtt=-1 clients, irrlicht.dll and ntdll.dll related |
| 21:54 |
Fixer |
will try with debug build to see anything interesting, i had 6 clients running from separate folders at the time of kick |
| 21:56 |
Fixer |
6 clients were kicked, 4 kicked to main menu, 2 had RTT=-1 and still game visible, esc-exit = crash on both |
| 22:01 |
Dumbeldor |
https://www.youtube.com/watch?v=sYsvgZyLUjY |
| 22:04 |
nerzhul |
nice video Dumbeldor , seems to improve the user experience on inventory |
| 22:05 |
Dumbeldor |
Thanks nerzhul :) |
| 22:24 |
|
Warr1024 joined #minetest-dev |
| 22:29 |
|
twoelk left #minetest-dev |
| 23:10 |
|
johnnyjoy joined #minetest-dev |
| 23:25 |
|
octacian joined #minetest-dev |
| 23:40 |
|
johnnyjoy joined #minetest-dev |
| 23:47 |
|
Tmanyo joined #minetest-dev |
