Minetest logo

IRC log for #minetest-dev, 2014-01-24

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

All times shown according to UTC.

Time Nick Message
00:05 Miner_48er joined #minetest-dev
00:18 ShadowNinja ~tell sapier Can you check https://plus.google.com/u/0/b/1168307643035179​37720/113166754097327759829/posts/KuYZoBrim82 ?
00:18 ShadowBot ShadowNinja: O.K.
00:19 cisoun joined #minetest-dev
00:19 cisoun Hey.
00:22 cisoun celeron55: I'm managing the Google+ page of Minetest (since 2011) and ShadowNinja told me I should transfer you its ownership. Interested ?
00:30 ShadowNinja This should fix ^: http://ix.io/a5H
00:40 proller joined #minetest-dev
01:06 VanessaE ok, progress report on the current state of the network code, without the pull request sapier mentioned wanting shadow and I to try:  with round about 30 users online, the map loader/emerge thread starts to lag rather badly - BUT things like chat, building/digging etc are working just fine.
01:06 VanessaE and the total CPU, disk and network bandwidth have plenty of headroom left, so it's not any kind of hardware bottleneck, at least not at the server end.
01:07 RealBadAngel so what about #1112 ? can we merge it?
01:07 ShadowBot https://github.com/minetest/minetest/issues/1112 -- New HUD element - waypoint. by RealBadAngel
01:07 VanessaE (and these are 30-odd active, real players, not a bunch of bots)
01:11 VanessaE inb4 pr0ller mentions that fm can do 10x as many players before emergethread starts to lag.
01:12 VanessaE it seems to be directly related to the number of players who are downloading new media, as before.
01:12 kaeza FM can do 20x with 1% CPU usage!
01:12 kaeza anyway, any more comments on crouching?
01:13 VanessaE but it's hard to be sure, since I can't directly monitor downloads vs. regular usage.  i can only go by what my network graph showed for the periods where the lag got bad.
01:13 VanessaE http://digitalaudioconcepts.com/va​nessa/hobbies/minetest/stats.html
01:16 diemartin joined #minetest-dev
01:24 salamanderrake joined #minetest-dev
01:44 kaeza I'm trying to implement a `kick_player' method, and have a question about style: is it better to have a std::vector<u16> listing the peer IDs to disconnect each step, or add a bool field to RemoteClient and check that?
01:44 kaeza (vector in Server)
01:46 cisoun left #minetest-dev
01:48 kaeza I guess calling DenyAccess from a Lua C function is not really safe...
02:11 kahrl kaeza: I think sapier just restructured the way players and clients are stored, but that isn't merged yet
02:11 kahrl so it might be better to wait on that
02:11 kahrl (or start from his branch)
02:30 kaeza ah I'll wait for him to finish his thing; it's not much work anyway :P
02:30 kaeza thanks kahrl
02:56 zat joined #minetest-dev
03:01 ShadowNinja Or rather, this should fix it: ,,(tell sapier I made this patch for it http://ix.io/a5V does it look good?)
03:01 ShadowBot ShadowNinja: O.K.
03:11 AllegedlyDead joined #minetest-dev
03:33 zat joined #minetest-dev
03:47 salamanderrake joined #minetest-dev
05:24 khonkhortisan joined #minetest-dev
05:27 Neological joined #minetest-dev
07:21 celeron55 cisoun (if you read the logs): I don't use google+ and i am definitely not interested
07:44 RealBadAngel celeron55, hi. so can we merge #1112 ?
07:44 ShadowBot https://github.com/minetest/minetest/issues/1112 -- New HUD element - waypoint. by RealBadAngel
08:13 werwerwer_ joined #minetest-dev
08:48 us_0gb joined #minetest-dev
09:15 us_0gb joined #minetest-dev
09:46 darkrose joined #minetest-dev
09:47 proller joined #minetest-dev
10:18 rsiska joined #minetest-dev
10:39 ImQ009 joined #minetest-dev
11:14 ImQ009_ joined #minetest-dev
11:19 ImQ009 joined #minetest-dev
11:54 Weedy_lappy joined #minetest-dev
12:05 ImQ009 joined #minetest-dev
12:21 ImQ009 joined #minetest-dev
12:26 Weedy_lappy joined #minetest-dev
12:26 Weedy_lappy joined #minetest-dev
12:27 Weedy_lappy joined #minetest-dev
12:39 mrtux joined #minetest-dev
12:45 EvergreenTree joined #minetest-dev
12:45 EvergreenTree joined #minetest-dev
12:53 mrtux joined #minetest-dev
13:07 mrtux joined #minetest-dev
13:17 mrtux joined #minetest-dev
13:27 proller joined #minetest-dev
13:27 troller joined #minetest-dev
13:34 hmmmm joined #minetest-dev
13:39 OldCoder joined #minetest-dev
13:39 VanessaE ha!  someone tried the signs "${text}" exploit on my server.  FAIL
13:45 dzho joined #minetest-dev
13:48 ImQ009 joined #minetest-dev
13:51 OldCoder joined #minetest-dev
13:55 rsiska joined #minetest-dev
13:58 bas080 joined #minetest-dev
14:08 PilzAdam joined #minetest-dev
14:20 rsiska joined #minetest-dev
14:32 RealBadAngel joined #minetest-dev
14:41 RealBadAngel celeron55, are you here?
14:47 zat joined #minetest-dev
14:55 xyz VanessaE: fail? why?
14:55 xyz it only depends on client version
15:00 VanessaE xyz: because if you're smart, you keep on top of bug fixes? ergo, the exploit should fail.
15:00 VanessaE oh right.
15:01 VanessaE > users
15:01 VanessaE > smart.
15:01 VanessaE what WAS I thinking.
15:06 sfan5 VanessaE: ${text} exploit? I'd like to know more
15:07 VanessaE sfan5: I don't know much about it.  some recursion bug, crashes some clients when they try to parse the text.
15:07 VanessaE (I thought it affected servers, originally)
15:10 sfan5 ah
15:17 Yepoleb joined #minetest-dev
15:18 xyz lolno
15:19 xyz if you're smart you're using stable version of software
15:19 VanessaE stable == buggy....
15:19 VanessaE if minetest's history is anything to go by.
15:19 VanessaE every one of us here knows that.
15:20 xyz well
15:20 xyz if you're smart and don't know much about minetest
15:20 xyz i mean, everywhere else you're supposed to use stable versions of software
15:21 VanessaE yes, perhaps.
15:21 xyz i'd call it confirmation bias
15:21 xyz i'm sure most of people use stable builds
15:21 xyz (it's similar to saying everybody who plays minetest uses linux)
15:21 VanessaE they don't? ;)
15:22 xyz well if you consider android
15:22 xyz although the app got pulled
15:22 xyz or something like that
15:22 VanessaE I dunno, somehow or another I seem to get more tablent users every day
15:22 VanessaE they're coming out of the fucking woodwork.
15:22 VanessaE tablet*
15:23 VanessaE I guess because the iOS client is still out there
15:27 xyz ask them
15:28 VanessaE waste of time :-/
15:28 VanessaE what I DO know is most of them are *paying* for it
15:28 VanessaE which pisses me off.
15:29 VanessaE <insert generic anti-capitalist/pro-communist rant here>
15:30 xyz why?
15:31 VanessaE because the game was meant to be free.
15:31 xyz define free
15:31 VanessaE I refuse to argue semantics, you know very well which definition I am referring to.
15:32 VanessaE if it were meant to be commercial, it already would be by now.
15:32 xyz no I don't
15:32 xyz it's perfectly fine to sell GPL'd LGPL'd software and you know it
15:32 VanessaE besides which, did we already forget that this for-pay version is an illegal fork?  under US copyright law, as I understand how it works, that makes it 10x as illegal as it otherwise would be.
15:33 VanessaE no, it isn't, if it's in violation of the license to begin with
15:33 xyz i bet it'd be hard to prove it's in violation
15:34 VanessaE probably hard to prove, sure.  But then again, it didn't seem to actually take that much to get it pulled from Google Play.
15:34 VanessaE I guess no one has approached Apple yet.
15:34 xyz we still don't know why was it pulled
15:34 VanessaE yeah I know
15:35 NakedFury joined #minetest-dev
15:35 VanessaE I assumed the initial response was automatic/boilerplate and that someone had finally reviewed the DMCA claim for real.
15:35 xyz they reported to me that they've taken care of it
15:35 xyz but then the app reappered
15:36 xyz +a
15:36 xyz not just the app, but three of them or so
15:36 xyz and then after a week or so everything got down
15:36 xyz well, I honestly don't care about apple
15:37 xyz and if they just pull the apps instead of releasing the source (which I kind of hoped they'd do) then I see no point in doing that
15:37 xyz it'll only piss users off
15:38 VanessaE maybe they were re-pulled automatically.  you know, duplicate/re-violation detection?
15:38 VanessaE but you're ultimately right, if they'd just release up-to-date, buildable source, no one would complain
15:38 VanessaE (at least about the legality of it)
15:39 VanessaE I'll still bitch about the interface, but that's only because my users do
15:39 xyz when I contacted google and pointed that the dev reposted those apps they said to open another claim
15:39 xyz when I did they once again asked to prove that this indeed is an illegal use
15:39 xyz then I thought "fuck it" and didn't reply
15:42 VanessaE that's what I meant by boilerplate, I think the "prove it" part is automatic isn't it?
15:42 VanessaE they must get thousands, perhaps tens of thousands of DMCA complaints per day
15:42 VanessaE surely they've automated that
15:43 xyz no, only the first reply is automatic
15:43 xyz which says something like "we're processing your request wait a week or so"
15:44 xyz the second time they also asked for a screenshot
15:44 xyz "of the allegedly infringing content as it appears in the application in question"
15:44 xyz well then it got boring
15:44 VanessaE hm
15:44 VanessaE well this kinda stuff can't be expected to be interesting
15:46 VanessaE actually, that's not entirely true
15:46 VanessaE I suppose some copyright cases could get interesting
15:47 VanessaE like that whole apple/samsung patent case.  or SCO vs...  IBM was it?  for a while there anyway.
15:55 book` joined #minetest-dev
16:18 Jordach joined #minetest-dev
16:33 john_minetest joined #minetest-dev
16:38 iqualfragile joined #minetest-dev
16:40 Weedy_lappy joined #minetest-dev
16:40 Weedy_lappy joined #minetest-dev
16:43 Jordach joined #minetest-dev
16:44 EvergreenTree joined #minetest-dev
16:44 EvergreenTree joined #minetest-dev
16:55 sapier joined #minetest-dev
16:57 sapier ShadowNinja http://ix.io/a5V seems to be correct
17:00 sapier kaeza if you use RemoteClient you have to lock client list while you use it
17:01 sapier I suggest looking at 1116 it changes quite a lot about client handling
17:02 sapier I wonder if vanessae's 30 player lag is result of number of blocks emerged
17:03 VanessaE I believe it is.
17:03 VanessaE because it's *only* the emergethread that lags, and even ordinary block loading (not mapgen).
17:04 VanessaE server commands (e.g. /status) and incoming/outgoign chat don't lag during that time
17:04 sapier node digging?
17:04 VanessaE s/emergethread/loading of the map data/
17:04 VanessaE even digging/placing nodes seems to behave nodmallt.
17:04 VanessaE normally*
17:05 xyz what map backend do you use?
17:05 VanessaE sqlite.
17:05 sapier it'd be interesting to know if there are other packets on lower prio channels as blocks
17:06 sapier obviously node digging is at higher priority
17:07 sapier if digging nodes is fast, it's not exactly lag but a send priority issue
17:07 sapier but I don't know how you could verify digging is really fast as prediction hides it
17:08 sapier you could try increasing "max_packets_per_iteration" but this increases jitter at benefit of per peer throughput
17:09 VanessaE sapier: I can only assume the placement is fast based on how much was still there when I signed off and back on again to try to combat the map-not-wanting-to-load issue
17:09 sapier ok
17:09 VanessaE that is to say, it was all still there save for the last couple of nodes I placed, and my sign off/on was only a few seconds.
17:10 VanessaE meanwhile, the map was failing to load 30 seconds on.
17:11 sapier a client side wireshark dump of this situation would be interesting
17:11 VanessaE I might do later if it gets into this situation again
17:12 sapier if I'm here I can do it myself but you might have to trigger me
17:12 VanessaE I expect a similar situation in about 8 hours
17:12 VanessaE http://digitalaudioconcepts.com/va​nessa/hobbies/minetest/stats.html
17:12 VanessaE you can see yesterday's peaks were between ~1800 and 2000 or so, EST
17:12 VanessaE it's now about 1212 EST
17:13 VanessaE (the big spike at 0800 was the daily backup cycle)
17:14 sapier hmm you could add a ram usage statistic ;-)
17:14 VanessaE I thought about it. :)
17:15 sapier this would show packet queuing ... and therefore you could see texture downloads quite easy
17:17 VanessaE I guess just a generic "total used", "buffers", and "cached" would be sufficient.
17:18 sapier used and swap usage might be usefull ... cached ... as fat as I know linuy uses any free memory for disc cache so this number isn't very usefull
17:20 sapier a friend of mine once implemented a read ahead feature for processing large data ... just to see a speedup of exactly 0 ... because of linux builtin disc prefetch mechanisms already did this before
17:20 VanessaE there's 512MB of swap in the machine, but with 8GB of RAM, of which more than half is usually free, only 29MB of that swap is used :)
17:20 sapier yes I didn't expect swap to be anything else then store for rarely used memory pages
17:27 rubenwardy joined #minetest-dev
17:27 grrk-bzzt joined #minetest-dev
17:31 sapier VanessaE if you want to know which threads are using cpu you could apply this patch https://gist.github.com/sapier/8602068
17:31 sapier it's linux only
17:32 sapier after this you will see thread names as well as per thread cpu usage if you enter "top -H -p <pid>"
17:34 VanessaE per-thread, as in per-mod?  or just per-some-generic-internal-minetest-thread?
17:34 sapier no
17:35 sapier per thread as in emerge thread server thread receive thread send thread
17:35 VanessaE right, so the latter :)
17:35 VanessaE still, that could be useful
17:35 VanessaE why not just merge that? (with appropriate ifdef's)
17:36 sapier I don't know how to do this on windows so I can't do a official pull request right now ... and I'd have to think about names in this case too as they're cut of right now
17:36 sapier e.g. async threads are all named identical ... they should have their number at end
17:37 sapier but of course those are minor things
17:37 sapier I could add a setThreadName to porting ... but someone would complain why this fct is empty for windows ;-)
17:37 sapier os specific ifdefs in other places then porting are crap
17:38 sapier at least for a feature that unimportant
17:42 Jordach someone might wanna deal with THIS again; https://itunes.apple.com/us/a​pp/minetest./id717384883?mt=8
17:45 Jordach you might also want to poke Gambit from the forums, his pack has been stolen
17:45 xyz wtf
17:45 xyz i don't think it's even minetest
17:46 xyz did someone test it?
17:46 Jordach afaik, Gambit's stuff is CC-BY-ND-NC
17:46 Jordach and that mapgen is minetest's i know that
17:46 sapier that "sheep" seems to be a simplemob
17:47 rubenwardy Doesn't matter. It uses our name.
17:47 rubenwardy Has anyone legally registered "Minetest" as a trademark?
17:47 Jordach btw, MC textures are stolen
17:47 Jordach i'll poke Mojang too
17:47 xyz rubenwardy: lol
17:48 rubenwardy Yeah, I wouldn't expect anyone too
17:48 sapier you don't need to officially register a trademark in a lot of countrys
17:48 xyz the 3rd pic is definitely not minetest
17:48 Jordach yeah
17:48 xyz http://a1.mzstatic.com/us/r30/Purpl​e/v4/7e/e8/30/7ee830f7-eb46-effa-ad​bb-ccd0bb746569/screen568x568.jpeg
17:48 Jordach xyz, i'll poke Mojang about it
17:48 xyz why?
17:48 sapier true the sky doesn't match
17:49 Jordach xyz, it's their IP
17:49 Jordach which the MC textures happen to BE
17:49 xyz no, I mean, why do you care?
17:49 xyz also did you dl it and cra^W check that they really use them?
17:50 Jordach xyz, their presence will have BC developers killed once and for all
17:50 xyz I don't understand
17:50 xyz why do you want to contact mojang
17:50 xyz what will you say
17:50 xyz hi i found this game which has one screenshot which looks a bit like your game
17:50 rubenwardy I think you should, as long as they don't sue US by mistake.
17:50 Jordach xyz, copyright infringement
17:51 xyz inb4 celeron55 gets sued
17:51 xyz well that'll sure be fun
17:51 Jordach xyz, the engine part
17:51 xyz hm?
17:51 Jordach cannot*
17:51 Jordach we know for a fact that this isn't legal via the LGPL
17:51 xyz did you download it?
17:51 Jordach xyz, no
17:51 xyz then you should stop being so gigantic faggot
17:51 Jordach why would i pay those tards who made BC
17:52 xyz how can you know without even downloading it?
17:52 sapier you don't have a proof so be very very carefull
17:52 Jordach xyz, there's a image with the MC farming texturews
17:52 xyz and this proves that it's based on Minetest how exactly?
17:52 Jordach with G4JC's almost MC pack
17:53 rubenwardy 1) Check that they do infringe the copyright (buy and check). 2) If so, send email to Mojang, stating the problem, and making sure the Minetest community is not affected.
17:53 xyz haha
17:53 rubenwardy Or just file a complaint/ DMCA with Apple
17:53 rubenwardy easier
17:53 xyz you guys sure have nothing better to do
17:53 xyz well then have fun, I guess?
17:54 rubenwardy !title https://forum.minetest.net/vie​wtopic.php?pid=127211#p127211
17:54 ShadowBot rubenwardy: Forum message — Minetest Forums
17:54 rubenwardy no, it's not
17:54 rubenwardy no, it's not
17:54 rubenwardy no, it's not
17:54 rubenwardy !title https://forum.minetest.net/vie​wtopic.php?pid=127211#p127211
17:54 ShadowBot rubenwardy: Forum message — Minetest Forums
17:54 rubenwardy ^ ShadowNinja
17:55 PilzAdam rubenwardy, this is not #minetest
17:55 rubenwardy Yep. That was to xyz.
17:55 rubenwardy And it is on topic, ish
17:57 xyz what?
17:57 xyz also I don't think you can just file DMCAs left and right
17:59 rubenwardy or do a cease and desist notice.
18:00 PilzAdam we should have a seperate channel for this, since its not really core dev related; maybe #minetest-ripoffs
18:00 rubenwardy lol
18:03 rubenwardy http://translate.minetest.​ru/projects/minetest/core/ is down
18:03 john_minetest It's great to be at the D - M - C - A!
18:03 john_minetest :P
18:04 rubenwardy lol
18:10 * sfan5 throws the GEMA at john_minetest
18:15 john_minetest GEMA doesn't work. First because the spoken letters haven't got the right sound and second because YMCA is surely blocked by GEMA
18:20 sapier VanessaE I know what happens in your case
18:21 sapier You've got >30 clients, this reduces max packets per step to 1000/30 -->34 packets. therefore if you receive just a single position update from each client per step no other unreliable data can be sent to this peer within this step
18:23 sapier set it to 2000 to handle about 60 clients
18:25 sapier hmm not exactly 2000 is only enough for ~45
18:26 sapier you need almost 4000 to handle 60 clients
18:26 Calinou joined #minetest-dev
18:26 sapier network load increases n²
18:31 Jordach joined #minetest-dev
18:57 NakedFury joined #minetest-dev
19:04 smoke_fumus joined #minetest-dev
19:46 EvergreenTree joined #minetest-dev
19:47 EvergreenTree joined #minetest-dev
20:06 daswort is there a DB where i can look which crafts are already in use by mods?
20:12 sapier no
20:12 anadon joined #minetest-dev
20:12 anadon Hey, I've gotten deep into re-factoring the code, and am running into some trouble. How is src/activeobject.h generated?
20:12 sapier I don't think we have any generated code
20:13 sapier "deep into re-factoring" sounds like a lot of changes?
20:13 anadon could you run find for that file on your repo then?  It doesn't seem to be in mine.
20:13 anadon The build system is touchy on my machine so I'm working on improving it.
20:13 sapier src/activeobject.h
20:13 anadon which is a whole lot of hurt.
20:14 anadon mhm
20:14 anadon weird
20:14 anadon ok, must have made a rm mistake.
20:14 sapier you are aware of refactoring huge parts of minetest doesn't improve chances for merge of a change?
20:14 troller joined #minetest-dev
20:14 proller joined #minetest-dev
20:16 sapier It's even hard for core devs getting support for refacoring small parts ;-)
20:18 Calinou joined #minetest-dev
20:35 salamanderrake joined #minetest-dev
20:42 xyz anadon: can your changes be seen somewhere?
20:45 sapier1 joined #minetest-dev
20:55 anadon xyz:  they are too mangled for me to let on someone else.
20:55 anadon I'll be back on tonight if you want to talk about it.
20:55 anadon (classes)
20:55 sapier1 what do you intend to do?
21:03 anadon joined #minetest-dev
21:11 xyz anadon: alright, although I'm even more curious now!
21:11 xyz meanwhile, a patch https://github.com/freeminer/freeminer/comm​it/edd354b83be0fe72f7da85cdaca8e47301cf64dd
21:11 OldCoder joined #minetest-dev
21:12 ShadowNinja anadon: Could you provide more details?  We wouldn't want you to waste your time if we can't accept it...
21:13 sapier xyz is 5.0 the trigger time?
21:13 xyz yes
21:13 xyz it's slightly-bigger-than-default (2.0)
21:14 xyz before if you connect to server too fast (in less than 2 seconds, singleplayer game) you won't receive any blocks for those 2 seconds
21:14 sapier what do you think about adding a define this way we wouldn't have to change two locations. 5.0 <-> 2.0 is a factor of 2.5? slightly? :-)
21:15 sapier I did a lot of changes in this area lets see if this still does what it does before
21:15 sapier -does +did
21:15 xyz yup, y'know float comparisons are inaccurate
21:15 xyz well you can change it to 1337.0
21:16 sapier yes but not as inaccurate as 2.5 ;-)
21:16 xyz sfan5: that's not how you fork Irrlicht
21:16 xyz sfan5: what I propose is:
21:16 sfan5 xyz: svn2git crashed
21:16 xyz really?
21:16 xyz well, anyway
21:16 sfan5 yes
21:16 sapier and we compare for >= (at least I hope so)
21:16 xyz you clone it with svn2git to separate branch
21:16 xyz then merge to master
21:16 xyz and then keep separate branch in sync and merge it to master when needed
21:17 xyz sapier: are you seriously going to argue on such simple small thing?
21:17 sapier no no don't get me wrong I'm not arguing ;-)
21:17 xyz sfan5: well if you're fine with what I suggest I can try it
21:18 sfan5 yes, i'm fine with it
21:18 xyz although I'm not sure if I want to contribute to that your repo :P
21:18 Miner_48er joined #minetest-dev
21:20 sapier why the hell does git patch think this change is in SetBlocksNotSent?
21:21 xyz you could've changed it by hand D:
21:22 sapier confusing patch ... not it's content put the patch surroundings ;-)
21:23 xyz sfan5: well, I guess that'll take a bit of time
21:23 sapier hmm maybe we should fix the issue mentioned here. there's no sane reason why this thread should be restarted
21:23 * sfan5 has enough time
21:24 xyz maybe
21:25 hmmmm hmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm xyz, that's not a very good idea
21:25 hmmmm that might cause a race condition
21:25 hmmmm that is, starting emerge threads immediately
21:27 Gethiox4 joined #minetest-dev
21:27 sapier we shouldn't  rely on some sleep delays to avoid race conditions
21:28 hmmmm wait, what..?  doesn't that delay starting emerge threads?
21:28 hmmmm and the delay seems to be 0 by default, not 5
21:28 hmmmm er not 2
21:29 hmmmm see a lot of freeminer changes seem to be tweaks to settings that modify the behavior of already established interfaces that aren't really solutions to problems
21:29 sapier if I interpret code correct and haven't missed another starting location emerge thread is started 2 seconds after server thread in current version
21:29 * hmmmm looks
21:29 hmmmm lol i get home from a week of code to do more code
21:29 sapier which is quite arbitrary ... why 2 not 5 or 10?
21:30 hmmmm i don't know.  a lot of this stuff is arbitrary to begin with.
21:30 sapier but I guess it's a startup delay in worst case. If there's a race condition it should be handled by a sync point
21:31 hmmmm ahhh i see what it does
21:31 hmmmm 5.0 is not how long it takes for the emergethread to trigger
21:31 hmmmm it's the initialization of the counter that's greater than the interval
21:31 hmmmm really all it needs to be is 2.0
21:32 hmmmm but this is a meaningless patch
21:32 hmmmm the emergethreads are started immediately
21:32 sapier yes but I'd prefere to use a synchornisation point if there's a race condition and start this thread only once
21:32 hmmmm in fact that block of code in asyncrunstep is dead
21:32 hmmmm as in it doesn't do anything
21:33 hmmmm you modified it sometime ago it seems, but it was useless before then too
21:33 hmmmm ever since I redid the emergethreads
21:33 hmmmm it was a hacky "fix" to a bug that was gone for quite some time, and now people misconstrued it as the starting point for the emerge thread
21:33 sapier :-) ok lets remove it ... removing bugs by removing code is best way to do it ... as you don't have to look for the bugs
21:33 xyz they don't start immediately
21:34 xyz since this patch fixes it
21:34 sapier maybe we should start them immediatly?
21:34 sapier could be a bug introduced by switch from simple to jthread
21:34 hmmmm let's see
21:35 sapier I guess I have interpreted this to be starting location too
21:36 hmmmm emergemanager is created right after the serverthread is created, mapgens are initialized, environment is initialized
21:36 * hmmmm blinks
21:36 hmmmm initMapgens() doesn't start the associated emergethread?
21:37 hmmmm wtf
21:37 hmmmm in any case the environment needs to be initialized before the emergethreads are started
21:39 sapier good to know. I intend to cleanup server stap by step as it seems to have more dead code in there
21:39 hmmmm the serverthread isn't started in the Server ctor
21:40 hmmmm oh the server initialization is pretty order-critical
21:40 hmmmm we need to collaborate together on that one
21:40 hmmmm also it's a huge mess
21:40 hmmmm i'd like to do some cleanup of my own
21:40 hmmmm except with environment initialization and how it binds together with the emergethread
21:41 sapier yes I'm thankfull for any help in there as it will save me some days of trial and error
21:42 sapier I've added proper client state handling in #1116 but it's still quite experimental if you have time to review the changes you're welcome
21:42 ShadowBot https://github.com/minetest/minetest/issues/1116 -- Add proper client initialization by sapier
21:42 xyz oh i also wanted to propose another patch
21:42 sapier just propose it ;-)
21:44 sapier does anyone know how to give a name to a windows thread?
21:45 hmmmm lol.  http://msdn.microsoft.com/e​n-us/library/xcb2z8hs.aspx
21:45 hmmmm undocumented yet documented
21:46 hmmmm that is incredibly hacky looking
21:46 sapier true
21:46 hmmmm (is that what you meant by name?  or did you mean like a named pipe or something)
21:46 sapier but lots of win32 code looks hacky to me :)
21:47 sapier no the thread itself like used in linux to show threadnames in top -H -p <pid>
21:47 hmmmm oh.. this is for debugging purposes
21:47 hmmmm I don't think you could have thread names in windows...
21:47 sapier :-/ so we can add a empty dummy for windows only :-/
21:48 xyz https://github.com/freeminer/freeminer/comm​it/bdf9ae3739c0a42fa62851f72a256cb065215229​#diff-ad60d65b34e16a3319296bb5d683acd6R983 https://github.com/freeminer/freeminer/comm​it/ca356c53f38640eae7f2d58f9f1c5f887a4acbf4
21:50 hmmmm hmm don't see the heap-use-after-free in the firsto ne
21:51 hmmmm also I think that's a bad idea to delete the environment before clients are deleted
21:52 xyz it's a bad idea why?
21:52 sapier because player is read from env
21:53 sapier if anyone calls a function triggering player usage in between it's gonna crash
21:54 sapier I've reduced player usage but I didn't make sure it's never used ... and I don't know where all those functions are called from .... almost everyone has a copy of server class pointer
21:54 sapier same for env pointers
21:54 hmmmm that's because everybody needs one
21:54 xyz ah now i'm not sure if my changes don't introduce another possible heap-use-after-free
21:55 hmmmm what's the first heap-use-after-free...?
21:55 xyz anyway, ~ServerEnvironment calls ~ServerMap which calls ServerMap::save which calls ServerMap::saveMapMeta which goes to EmergeManager::setParamsToSettings
21:55 sapier shouldn't we cleanup this mess by adding adding proper encapsulation? not at once but step by step
21:56 hmmmm ahhh.
21:56 hmmmm that's pretty hidden
21:56 sapier most time "player" is only used to fetch it's name from ... I had to move the name to RemoteClient for delaying player appearance maybe we can use this name at some locations where player is used by now
21:57 xyz yea so hidden it happens every time so..
22:00 john_minetest left #minetest-dev
22:00 xyz although map probably should be deleted after both environment and emergemanager
22:01 salamanderrake joined #minetest-dev
22:02 hmmmm hmm
22:03 hmmmm I think a better way of fixing this would be to remove setParamsToSettings from saveMapMeta
22:03 hmmmm or is it
22:04 hmmmm too tired to think straight :(
22:06 hmmmm honestly the mapgen params should be in Map
22:06 hmmmm but they aren't because that's not what gets passed along to the emergethreads
22:06 hmmmm but anyway yes, i agree, the emergemanager should be deleted a lot later than it is currently as lots of things depend on it
22:07 hmmmm emergethreads should be stopped, clients deleted, then the environment gets deleted
22:07 hmmmm and then after all that emergemanager gets deleted
22:07 hmmmm the whole point of deleting the emeregmanager early in the shutdown process like that is to stop the threads anyway
22:13 john_minetest joined #minetest-dev
22:13 john_minetest left #minetest-dev
22:27 sapier does anyone know why there are reliable as well as unreliable active object messages?
23:03 anadon joined #minetest-dev
23:13 EvergreenTree joined #minetest-dev
23:13 EvergreenTree joined #minetest-dev
23:31 ShadowNinja xyz: Can you add a favicon to the wikis?
23:31 xyz later
23:33 xyz and send me the favicon too
23:36 ShadowNinja xyz: The Minetest logo, same as the forum.
23:39 kahrl sapier: I guess unreliable is meant for movement and reliable is for long-term state changes?
23:56 sapier hmm maybe can't see it from data beeing sent ... the message itself isn't dissected

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