Minetest logo

IRC log for #minetest-dev, 2014-06-14

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

All times shown according to UTC.

Time Nick Message
00:16 sapier would it be difficult to enable tables in forum?
00:59 Piggybear87 joined #minetest-dev
01:36 w00tc0d3 joined #minetest-dev
01:41 sapier left #minetest-dev
01:48 kaeza joined #minetest-dev
04:06 KingsleyT joined #minetest-dev
04:48 us`0gb joined #minetest-dev
05:02 celeron55 joined #minetest-dev
05:15 Hunterz joined #minetest-dev
05:19 SEx2 joined #minetest-dev
06:06 Eater4 joined #minetest-dev
06:18 Eater4 celeron55: have you (are anyone who runs the serverlist) decide to extend the max servers to a higher number?
06:30 Eater4 joined #minetest-dev
06:45 celeron55 Eater4: i have nothing to do with the list
06:45 celeron55 xyz hosts it, proller has implemented the server side and sapier probably knows most about the client side
06:46 Eater4 oh ok
06:49 VanessaE kaeza and I and a couple others discussed the idea earlier of getting someone (read: not us because we have no access and no skill in this field) to implement a weighted sorting method into the list, both on the website and in the client, based on four factors:  uptime, lag, some sort of community upvoting, and some kind of alpha sorting (maybe by domain or so).
06:50 VanessaE as it stands now it is incredibly unfair to the less-popular servers to be left to flounder in the backwaters of the bottom of the list just because they have few users.
06:53 celeron55 lag is a bit troublesome
06:53 VanessaE yeah I know
06:53 VanessaE it's based out of Russia or something
06:54 celeron55 but voting and uptime would seem like something that people would find useful
06:54 celeron55 xyz hosts stuff in germany
06:54 VanessaE oh, ok
06:54 celeron55 (lag is going to affect votes anyway)
06:54 VanessaE not sure where I got the Russia bit from)
06:55 celeron55 well his main domain is minetest.ru
06:55 VanessaE the alpha order factor is there for the idea that if num_votes * uptime is roughly equal among two or more servers, then they'll end up listed in alpha order (in theory)
06:56 VanessaE right - I must still be stuck on the idea that he was hosting the list from some server over there then.  eh, no matter.
06:57 darkrose joined #minetest-dev
07:02 Garmine joined #minetest-dev
07:48 rdococ joined #minetest-dev
08:11 sapier joined #minetest-dev
08:33 LemonLake joined #minetest-dev
08:37 proller Eater4, what extend??
08:38 Eater4 expand the max servers on the public server list and website
08:39 Eater4 joined #minetest-dev
08:39 proller VanessaE, all games sorts server list by player number, or have adjustable srting
08:39 VanessaE proller: also, please read scrollback ... ok, you read it.
08:39 proller Eater4, how to expand?
08:40 sapier I don't know of any limitation to serverlist on client side
08:40 VanessaE proller: as we lack any kind of adjustable sorting, we need at least a more fair way of balancing out "best" versus "most busy".  "most busy" often is not the best but merely "has the most people signing on and off all day long".
08:42 Eater4 http://minetest.net/servers
08:42 Eater4 See the servers? 40/45?
08:42 VanessaE proller: that's why I suggested something that should roughly calculate out to a sort of a "best score"
08:42 VanessaE Eater4: 45 is how many servers the list has seen since its last restart, I think
08:42 proller Eater4, and what to expand?
08:42 VanessaE not the max possible number
08:43 proller 45 - was maximum online in one time
08:43 Eater4 Really?
08:43 VanessaE proller: I think he believes the list will hard-cut the number of entries at 45. :P
08:43 Eater4 I did not know, sorry
08:44 proller 703 - hacked player number, one bad man have hacked server with 500+x players online
08:44 VanessaE proller: yeah, that should probably be reset, imho :)
08:45 VanessaE and in fact that bad server is precisely why I left number-of-players out of my proposed "best score" idea :)
08:45 VanessaE you can't hack uptime, ping, or lag, and users' votes would be somewhat difficult to bias as well.
08:46 proller no, only ping fully masterserver-side
08:46 proller others recieved
08:46 VanessaE yep I know, but that's a variable I figure the master server could send to the clients as part of this proposed score
08:46 proller VanessaE, an you find good score formula?
08:46 VanessaE let the master server make these calculations server-side and just send the scores to the clients.
08:47 VanessaE proller: I have been thinking about it.  the terms would be pretty simple, you know..  just this times that times the other thing.  no complicated algebra or anything
08:48 VanessaE maybe votes*uptime/(lag*ping)
08:49 VanessaE then scale the result to some reasonable range, e.g. 8 or 16 bit int
08:49 proller most popular servers must be on first places
08:49 VanessaE well votes would be weighted
08:50 VanessaE so how about, ((votes*vote_weight*uptime)/(lag*ping))*scale_factor
08:50 sapier I wonder if we have any chance to not get a manipulated automatic weighting
08:50 proller you want implement vote system to pick up your average-0-players servers
08:50 VanessaE the idea being that lag, ping, and uptime could drag a server down to the ground if it has few votes
08:50 proller ?
08:51 VanessaE proller: almost but not entirely unlike that, yeah :)
08:51 VanessaE a good server has low ping times, low lag, a nice number of votes, and a reasonable uptime, yes?
08:51 VanessaE with number of votes and lag/ping probably being the most important factors
08:51 proller good server have most connected users.
08:52 VanessaE not really
08:52 proller ddo what you want.
08:52 VanessaE look at that bad one that had 500+
08:52 proller ask ShadowNinja to rewrite serve on php
08:52 proller and hardcode your servers on first place
08:52 proller bye
08:52 VanessaE um
08:53 VanessaE dude, that was uncalled for...
08:53 Eater4 VanessaE: take a look at isis >:(
08:53 VanessaE my servers will still end up somewhere in the middle of the list regardless of the formula that's used
08:55 sapier I'm not conviced trouble/use relation of this feature is worth the work, any algorithm we implement is easy to manipulate
08:56 PenguinDad joined #minetest-dev
08:56 asl and ping isn't really fair
08:56 VanessaE sapier: perhaps, but if the choice is between a formula that can be manipulated versus using a method that IS being manipulated?  I'd choose the formula.
08:57 VanessaE asl: by itself, you're right.
08:57 sapier ping is useless as having a good ping to server list doesn't tell anything
08:57 sapier and pings told by server aren't relieable
08:57 VanessaE ok so leave ping out and just use the lag value.
08:58 sapier which isn't relieable too as anything told by server ... that's the 500+x issue
08:58 VanessaE aren't those figures measured by the server list?
08:58 asl unless it's client side ping
08:58 sapier now how is server list supposed to do?
08:59 VanessaE sapier: the same way a client does?  connect to a server, get basic stats, measure ping/lag, disconnect?
08:59 VanessaE well I suppose not.
08:59 sapier no
08:59 sapier server tells serverlist
08:59 VanessaE the servers have to broadcast that data I guess
08:59 sapier and server can tell anything
08:59 VanessaE well going by just the number of players is a recipe for failure
08:59 VanessaE and it has been failing.  miserably.
09:00 sapier as anything else will fail
09:01 sapier thus we'd do work for what benefit?
09:01 VanessaE lemme put it this way:  if redcrab's server were in this list, it would be near the bottom because few people play on it now.  If it were properly weighted, it should be at the top.  It's one of the best servers in the community, you know that and I know that.
09:02 sapier someone could implement it's own server list and do user voting
09:02 VanessaE what good would that do?
09:02 VanessaE no one would see it.
09:03 sapier you can specify a different server list
09:03 VanessaE server owners and users won't do that and you know it.
09:03 VanessaE the point is to get a fair weighting on the servers in the list so that everyone gets a chance to be seen and get some usage out of their servers.
09:04 Eater4 I think it should mesure all of the componets that we talked about and avarage that out on a daily basis
09:05 VanessaE there are 40 servers in the list.  users only see the top 6.
09:05 VanessaE that means there are 34 they will never see.
09:05 VanessaE ever.
09:05 VanessaE because users don't like to scroll.
09:05 VanessaE for reasons I do not understand, users don't seem to figure out this concept of the scroll bar.
09:05 Eater4 Because, this is the thing that minetest is WAY diffrent than Minecraft
09:06 VanessaE that means we need to give servers a fair chance of getting into those critical 6 places in the list.
09:06 sapier I understand your intention VanessaE and I'm fully compliant, I just don't se a way to implement it in a way it's not easy to manipulate it
09:06 proller VanessaE, if he hardcode all your 7 servers to top - one always will be with 0 users
09:07 sapier proller I know it's not your way of doingt things but could you please start to be constructive ?
09:07 proller okay.
09:07 VanessaE sapier: then we just have to deal with having a formula that's easy to manipulate and do what's been done in the past:  blacklist servers that attempt to manipulate their numbers.
09:07 proller lets sell first 6 vip places
09:07 proller 1 place - 100$/month
09:07 kaeza ...
09:07 VanessaE *facepalm*
09:07 proller paypal: plroler@gmail.com
09:07 Eater4 WHAAT!
09:07 kaeza why are you even here?
09:08 sapier if you use values like ping and lag, it's way more difficult to find out someone is cheating
09:08 proller but better to implement auction system
09:08 sapier e.g. a server telling 20% less lag as really mesured with some minimum value
09:09 VanessaE sapier: this is true, which is why those numbers would have to be weighted as well.
09:09 VanessaE of course it becomes an arms race, but in the end ping and lag could be made to have only a small influence on the ranking.
09:09 Eater4 Wait
09:09 VanessaE maybe cap them above/below some arbitrary amount>
09:10 Eater4 Selling the places is really stupid
09:11 VanessaE Eater4: he's being sarcastic.
09:11 Eater4 Hm...ok looks like it is time for me to go to bed
09:13 VanessaE proller: I'm just trying to make it more fair for everyone.  I don't understand why you have to be so snotty about it.
09:15 jin_xi joined #minetest-dev
09:15 VanessaE one of the false positives in player counts, also, is "Guestxxxx".  If you *must* include the number of players in the ranking, the count should not include any names containing "admin", "test", "guest", etc.
09:18 sapier we don't know about this
09:19 VanessaE yes we do.
09:19 VanessaE look at the master server list.
09:19 sapier you wanna tell me servers send out all their playernames?
09:19 VanessaE most of them already do.
09:19 VanessaE the top server in the list has 11 guests and 23 total users.
09:20 VanessaE mouse over the "23/25"
09:20 VanessaE http://servers.minetest.net/
09:20 sapier ahh we're talking about the current players
09:20 VanessaE yes
09:20 VanessaE just whatever the servers are announcing with
09:20 sapier still I don't believe the benefit is in any relation to the work and trouble this will cause
09:22 jin_xi cant there be 'server list tabs' sorting by #players, lag or community ranking, ping, lexicographical whatever
09:23 jin_xi no function will ever find best servers reasonably
09:24 VanessaE jin_xi: no formula can ever find the absolute best, no, but I'm trying to propose something that would be used as the default ranking/sort method.
09:24 Eater4 joined #minetest-dev
09:24 VanessaE other ways to sort could still be offered.
09:25 VanessaE sapier: I guess it just depends on whether a person considers it important.
09:25 VanessaE but to the average user, minetest only has 6 public servers.
09:25 VanessaE because users for some reason, can't seem to figure out how to scroll.
09:27 jin_xi make scroll arrow big and red?
09:28 VanessaE won't help.
09:28 Eater4 Oh I know
09:29 Eater4 Make the list smaller
09:29 Eater4 So more servers can be seen
09:29 jin_xi or shuffle them everytime so you get a chance to see them all eventually
09:30 VanessaE Eater4: how would that help?
09:30 VanessaE jin_xi: well that would help to some degree.
09:31 Eater4 So you can see more servers, there fore there is a more likeyhood that you would see more than six
09:31 VanessaE if the list is "smaller" what happens to the servers that didn't make the list?
09:32 Eater4 I mean Not smaller
09:32 VanessaE well if you make the viewport bigger, you need screenspace to do it.
09:32 VanessaE it's already limited on PCs, and it's at a premium on a tablet
09:32 VanessaE so that's not happening.
09:33 Eater4 Hm
09:36 sapier while I'm generally positive about using "recent" science discoverys to improve our usability I wonder if this is a really important issue ;-) ... btw mainmenu is lua, any modder should be able to improve it
09:36 sfan5 >btw mainmenu is lua
09:37 sfan5 > any modder should be able to improve it
09:37 sfan5 the problem are formspecs
09:37 sfan5 they are awful
09:37 sapier I wont discuss that issue again ;-P
09:38 sfan5 there is no need for discussion
09:38 sapier at least not now ... maybe in about half a year ... or in unlikely case someone comes up with a fully implemented full featured replacemend ;-)
09:38 sfan5 formspecs ARE awful
09:38 sapier including compatibility code of course ;-)
09:40 sapier https://github.com/minetest/minetest/pull/1371 I don't know why there was used getBlockNoCreateNoEx(p) just to throw an assertion if this resulted in not getting that block ... that's pointless
09:41 sapier and of course noone uses the MapBlock passed back by that function so why is it passed?
09:55 jin_xi well, double click on scroll arrow starts connecting, which is not what i want.
09:56 jin_xi doubleclick outside of ingame volume control does not exit the menu if it matters
09:57 sapier first one has been this way before too
09:58 sapier later one is relic of volume control still being a non formspec menu
09:58 sapier guess that should be fixed by some time
09:58 jin_xi i just bring it up because of previous users cant seem to scroll comment
09:59 VanessaE jin_xi: you're not a user.  you're a geek :)
09:59 jin_xi idk, but i hardly ever connect to servers anyways
09:59 sapier the volume control menu is an open issue on android
10:12 restcoser joined #minetest-dev
10:21 ImQ009 joined #minetest-dev
10:47 proller joined #minetest-dev
11:25 PenguinDad joined #minetest-dev
11:30 vifino joined #minetest-dev
11:32 celeron55 joined #minetest-dev
12:36 nore joined #minetest-dev
12:49 CraigyDavi joined #minetest-dev
12:56 celeron55 joined #minetest-dev
13:08 diemartin joined #minetest-dev
13:10 PilzAdam joined #minetest-dev
13:27 NakedFury joined #minetest-dev
13:30 vifino joined #minetest-dev
13:42 Megaf joined #minetest-dev
14:27 rubenwardy joined #minetest-dev
14:48 nore "16:47:17: ERROR[main]: ServerError: error in error handling" <-- eh.. what?
14:51 sfan5 seems like this comes from Lua:
14:51 sfan5 ./lua/src/ldo.c:58:      setsvalue2s(L, oldtop, luaS_newliteral(L, "error in error handling"));
14:53 sfan5 can only happen in a few cases:
14:53 sfan5 ./lua/src/ldo.c:171:    luaD_throw(L, LUA_ERRERR);
14:53 sfan5 ./lua/src/ldo.c:374:      luaD_throw(L, LUA_ERRERR);  /* error while handing stack error */
14:53 sfan5 ./lua/src/ldebug.c:621:    if (!ttisfunction(errfunc)) luaD_throw(L, LUA_ERRERR);
14:55 nore I guess what happened was minetest.after(0, <not a function>)
14:55 nore so there is a problem there
14:55 sfan5 doesn't it check for that?
14:56 nore it does... perhaps it was the assert that crashed
14:57 Jordach joined #minetest-dev
14:57 sfan5 o.O
14:59 ShadowNinja I did rewrite the serverlist in PHP, but it's unfinished.
15:00 sfan5 >PHP
15:00 sfan5 pls
15:00 sfan5 use a decent programming language
15:00 sfan5 like Python or Ruby
15:00 ShadowNinja And that error should only happen if you didn't pass a proper error handle to lua_pcall(), xpcall(), or it otherwise broke internally.
15:00 ShadowNinja sfan5: Those are harder to set up for web usage.
15:01 Megaf or pure HTML 4
15:01 PenguinDad Megaf: why not HTML 5?
15:01 Megaf pure HTML 4 with a frame that shows a txt file
15:01 Megaf lightweight and works on everything
15:02 ShadowNinja Megaf: No, that won't work.  It has to receive JSON data, decode it, proccess it, and store it in a file on the server.
15:02 sfan5 ShadowNinja: be prepared to get hit by any of these http://me.veekun.com/blog/2012/04/09/php-a-fractal-of-bad-design/
15:02 Megaf PenguinDad; if you can write HTML5 without JavaScript...
15:02 LemonLake joined #minetest-dev
15:02 Megaf ShadowNinja; yep, there shold be another program or daemon getting the data and writing to a txt file,
15:02 Megaf and html 4 would show that file
15:03 Megaf IMHO the server list should be as light as possible
15:04 Megaf and simple as possible
15:04 Megaf then it would be easy to anything read from it
15:05 sfan5 it is already simple
15:06 rubenwardy left #minetest-dev
15:06 sfan5 dl http://servers.minetest.net/list and parse it as JSON
15:06 Megaf so, what change it?
15:06 Megaf why*
15:07 Megaf And can't we get rid of JSON?
15:07 sfan5 why?
15:07 hmmmm joined #minetest-dev
15:07 sfan5 almost nobody can read perl
15:07 Megaf because JavaScript
15:07 PenguinDad Megaf: what would you use instead?
15:07 sfan5 what does JSON have to do with JavaScript?
15:08 Megaf I don't know, I don't know how that works at the moment
15:08 Megaf I would use simple and standard xml, html 4 and all plain text
15:08 sfan5 XML
15:08 Megaf sfan5; The text inside :P
15:08 sfan5 fucking XML
15:08 Megaf why?
15:08 Megaf It works and it's simple
15:08 sfan5 XML is not readable
15:09 sfan5 Did you just really say XML is simple?
15:09 Megaf I depends on how you write it
15:09 sfan5 no it does not
15:09 Megaf well, I like the way it is now
15:09 sfan5 JSON is always simpler than XML
15:09 Megaf I'm not really familiar with JSON actually
15:10 sfan5 is the XML spec as short as the JSON one? http://json.org/
15:10 sfan5 the images with the green boxes _completly_ describe JSON
15:10 sfan5 this is what simplicity is
15:10 sfan5 XML is anything but simple
15:11 LemonLake +sfan5
15:12 Megaf Right, I'm wrong about XML
15:36 sapier wasn't javascript the reason why json was invented?
15:39 sapier any additional comments to https://github.com/minetest/minetest/pull/1309?
15:40 sapier if not I'm gonna merge it soon
15:41 sapier same for https://github.com/minetest/minetest/pull/1371
15:41 sapier and https://github.com/minetest/minetest/pull/1372
15:44 proller js stolen data format from perl {'a'=>1, 2=>[3,4,'five']}
15:47 proller today dice shows to use )\n{
15:50 PilzAdam joined #minetest-dev
15:54 hmmmm [11:08 AM] <Megaf> I would use simple and standard xml, html 4 and all plain text
15:54 hmmmm no.
15:54 hmmmm XML is *NOT* simple.
15:54 hmmmm the standard is actually very overcomplicated, making it just about impossible to parse or build without a third party library.
15:54 sapier hmmmm do you really believe that discussion will end anytime?
15:55 sapier :-)
15:56 hmmmm sapier:  probably not, but minetest SHALL NEVER need to use a third party library to achieve some task as mundane as data serialization
15:56 hmmmm I think we already have way too many dependencies the way it is
15:57 sapier you know we use json lib too? ;-) ... true it's quite lightweight
15:57 hmmmm I don't like it though
15:58 sapier I don't take part at that discussion, imho we have within minetest what we need if someone wants something new he's got to provide very very good reason why this can't be done with what we already have
15:59 hmmmm yes
15:59 hmmmm it's just that every time I see these sorts of discussion on #minetest-dev it scares me, because I'm not around hardly ever anymore and I don't want the project to become overrun with horrible design decisions
16:00 zat joined #minetest-dev
16:00 sapier I usually try to stop commits adding x-th way to do same thing but I don't always suceed ... see redis
16:01 sapier at least it's some completely different style of db ... and db interface is quite clean so adding and removing has almost no sideeffects
16:02 sapier talking about design decisions what about https://github.com/minetest/minetest/pull/1364 ?
16:03 sapier removes some reimplemented irrlicht gui element which doesn't compile on irrlicht 1.9
16:03 hmmmm i'm all for removing things that aren't used
16:04 sapier as of comment it's used for some hack ... I haven't found a way to trigger it
16:04 hmmmm however I think the reason that's there is due to Irrlicht brain damage and something wouldn't work unless it were basically reimplemented from scratch with the needed modifications
16:04 hmmmm absolutely ridiculous
16:04 sapier and the thing done by the hack is implemented in a way more clean way
16:05 sapier the only thing I can imagine is there might be quite old worlds out where that one was used
16:05 sapier but I don't know if it'S 2 or 4 years back
16:23 sfan5 sapier: "Shouldn't the dirt texture get moved to textures/base/pack/ to make it not depend on a minimal game being available?" you did not reply to that yet
16:25 sapier because I moved it
16:27 sfan5 I see
16:28 sapier actually I added same one again to base as I can't remove it from minimal
16:28 sapier as it's a 16x16 texture that's not a big deal
16:38 jin_xi so, i've been looking at the particle situation. irrlicht particles look nice for classic particle effects, but cannot do the physics which current node digging particles need
16:40 jin_xi so we could add/use a particle engine lib like sparkengine or combine a physics engine for node fragments and such and use irrlichts particles for smoke etc
16:41 jin_xi or just fix the wtf code for current particles and be done with it
16:51 Calinou joined #minetest-dev
17:11 casimir joined #minetest-dev
17:19 Eater4 joined #minetest-dev
17:44 sapier left #minetest-dev
17:51 Ritchie joined #minetest-dev
18:08 EvergreenTree joined #minetest-dev
18:10 EvergreenTree joined #minetest-dev
18:21 Megaf Ok, I'm downloading the Android SDK and will make builds of Minetest for Android
18:23 Megaf will we use the same name for the Android version?
18:23 sfan5 why would we not?
18:27 Megaf well, MC did the Pocket Edtion
18:27 Megaf Maybe Minetest Mobile Edition or Minetest Mobile
18:27 sfan5 it's not a pocket edition
18:27 sfan5 it is the full featured client
18:27 sfan5 it does not need a name change
18:27 Megaf indeed
18:29 Megaf well, my build will run on ARMv6, ARMv7, x86, x64 and maybe ARMv8 (64 bit)
18:29 sfan5 I don't think android supports arm64 yet
18:29 Jordach joined #minetest-dev
18:30 Megaf It does
18:30 Megaf the latest one does
18:30 sfan5 [citation needed]}
18:30 sfan5 s/}//
18:30 kahrl joined #minetest-dev
18:30 Megaf by Meaf
18:30 Megaf Megaf*
18:30 Megaf lol
18:30 Megaf let me see the official doc
18:32 Megaf sfan5; wow, there's no support from Android to ARMv8
18:33 sfan5 what did I say? >.>
18:33 Megaf ok
18:33 Megaf well, my build will run on ARMv6, ARMv7, x86, x64 and not ARMv8
18:33 Megaf >P
18:34 Megaf sfan5; I see my mistake, there's a number of Linux distros that supports ARMv8 and iOS 7 supports iOSv8 already too
18:35 Megaf citation = http://thread.gmane.org/gmane.linux.kernel/1367774/focus=1367868
18:36 sapier1 joined #minetest-dev
18:37 Megaf now, what Android versions should I support?
18:37 Megaf I think from 2.1+ ?
18:37 sapier1 2.3+
18:37 w00tc0d3 joined #minetest-dev
18:37 Megaf ok
18:37 sfan5 you can't support 2.1
18:37 sfan5 it does not have android_native_app_glue
18:38 sapier1 we could write it outselfs but as long as noone has a device fast enough to be capable of running mt I see no use in spending that time
18:49 werwerwer_ joined #minetest-dev
18:53 sapier1 merging rebased #1309 once travis build is complete
19:01 Megaf hm, eclipse will never load here
19:05 Eater4 Megaf?
19:10 Megaf I'm trying to run Eclipse from Android Developer Tools, it shows the splash screen, then asks about the workspace, then I select one and then it show the splash screen again "Loading orc.eclipse.help" and stays there.
19:12 sapier1 pushing now
19:12 Megaf gonna try with Java from Oracle
19:21 sapier1 I'm gonna merge #1372 soon
19:29 EvergreenTree joined #minetest-dev
19:30 EvergreenTree joined #minetest-dev
19:37 Megaf Ok, ADT doesnt work here sapier1 =/
19:37 sapier1 there are plenty of different reasons for this ... you know you need adt AND ndk?
19:38 Megaf Yep, I dont the bundle
19:38 Megaf It used to work on Debian Wheezy, Now I upgrade to Debian Jessey and it doesnt work anymore
19:39 sapier1 hmm I'm on wheezy
19:39 sapier1 pushing #1372 now
19:46 Megaf Wheres the bot that used to say github links to issues we wrote here?
19:46 sapier1 ShadowNinja just because you label a regression to "not a bug" it still is a bug
19:50 Megaf sapier1; Eclipse from Debian works fine
19:50 sapier1 maybe google did some strange patches
19:54 Megaf I wonder if I can use normal eclipse
19:54 sapier1 you don't need eclipse to build
19:55 sapier1 who feels responsible for redis?
19:58 Taoki So I'm experimenting with workbenches again. I must say, the node inventory lag even in singleplayer is really visible, for both a workbench and chest or furnace.
19:58 Taoki Can't anything be done to predict it locally?
19:59 sapier1 not without major changes
19:59 Taoki Like modify the menu / formspec code that handles inventory lists, to show an item as having been moved before th server reports that
19:59 Taoki Ok... that's strange as it would sound pretty easy
19:59 Taoki It's just a local visual effect to remove the feeling of lag
20:00 sapier1 only if you don't consider that a move may be prohibited
20:00 Taoki Can't the client know that when the formspec is initially notified to it?
20:00 sapier1 at this time it's far from sure that this move will be allowed
20:01 sapier1 if we display it we'll never get told that it didn't actually happen
20:01 Taoki Yes, but IMO that's far better than seeing this horrible lag in chests and all
20:01 sapier1 no that's not
20:01 Taoki There can be a client setting for it, if anyone's unsure
20:01 sapier1 one is lag other is inconsistent messed up inventory
20:02 sapier1 client will NEVER get told about the correct locations later
20:03 Taoki It can get told when the server tells it, like it happens currently. Just that the visual effect of item movement is also applied locally before that
20:03 Taoki And updated when it's told, in case something changes then
20:03 sapier1 server does only update inventorys if things changed
20:03 sapier1 if you do modify your local representation of the inventory you wont get back in sync to server
20:04 Taoki yeah, but if you do a change locally that's sent to the server automatically
20:04 sapier1 server is the one to decide not your local client
20:04 Taoki Also, if the local representation is also updated again when the server notifies it, alongside local effect, that can reduce any out of sync problem
20:04 Taoki True. But that's prediction
20:04 sapier1 server won't update it
20:04 Taoki Same with movement prediction in some games
20:05 sapier1 no not same
20:05 sapier1 it's scriptable if some move is allowed or not thus prediction isn't possible without client side lua
20:05 kahrl sapier1: how about: if the server doesn't update the inventory within X seconds after your request, revert to the last state sent by the server
20:06 kahrl see also: client.cpp:711
20:06 Taoki kahrl: X seconds means lag. The idea is for item changes to be represented instantly
20:06 sapier1 that'd ne the way to go for client side lua too but any at least somehow generic solution would be quite some work
20:06 kahrl Taoki: yeah. Change the local inventory instantly, but revert to the authoritative state if it turns out the server doesn't update
20:06 Taoki kahrl: Exactly
20:07 Taoki That's what I meant. Can't fully get why that's not possible
20:07 kahrl (either because of extreme lag or because the move was prohibited)
20:07 sapier1 you'd remove network lag but add permission lag
20:08 Taoki Permission lag ia s lot more rare however
20:08 sapier1 no matter how fast your connection is, if some move is prohibited client will realize at best after X seconds
20:08 sapier1 but it's a permanent fixed lag
20:08 kahrl sapier1: we could change the server to always send the inventory if a move was prohibited
20:08 kahrl so this lag only occurs with new client -> old server
20:08 Taoki sapier1: Yeah, but currently it does that for everything else
20:08 sapier1 hmm yes that'd be a way to go
20:09 Taoki IMO it's far better to have lag in a rare circumstance (like prohibited move) rather than the most frequent circumstance (every move)
20:09 sapier1 no it's not taoki
20:10 sapier1 if there's lag user expects it because it's always there, having some random huge lag on a few situations causes way more confusion
20:10 sapier1 yet kahrls suggestions make sense, at least if there's someone to implement it, I'm still busy with android port
20:11 Taoki sapier1: There are two solutions at least: 1 - Have the client know which inventories in a formspec hae permissions to put items in then, or 2 - Make this a setting turned off by default, so only weirdos like me can enable it :P
20:11 sapier1 and this is not a fix for crafting lag
20:11 Taoki Yeah, that's a fix too
20:11 sapier1 no that's not
20:11 sapier1 for crafting we'd need to tell clients about craft recieps too which isn't done by now
20:12 Taoki yes, seeing the craft result update would remain slow. But that's more acceptable
20:12 sapier1 write a feature request or implement it yourself ;-) or persuade someone to do it for you ;-)
20:12 * Taoki nods, will do
20:17 Krock joined #minetest-dev
20:19 snaaaaake left #minetest-dev
20:22 kaeza joined #minetest-dev
20:29 sapier1 anyone to comment #1371?
20:30 sapier1 https://github.com/minetest/minetest/pull/1371
20:37 sfan5 sapier1: looks good
20:39 VanessaE sapier1: *reading some of the backlog*  what exactly stops us from sending the crafting recipes to the clients?  aside from this needing a protocol bump I mean?
20:41 sapier1 I didn't say we can't do it it's just some work to be done
20:41 VanessaE ok.
20:42 VanessaE just that the previous conversation clearly shows a strong resistance to client-side prediction of crafting, and seeing as how that ^^^ is the only sane way to do it... :)
20:42 * VanessaE hopes taoki takes your second idea and just codes it :P
20:43 sapier1 ok that's not what I wanted to tell, I'm only reluctant to do this myself ... especially as last big thing is far from finished ;-)
20:43 VanessaE (as you and I both know, feature requests around here tend to be ignored, in favor of "code it yourself", and finding others to do the work tens to be...difficult)
20:43 Taoki VanessaE: Predicting crafting results would actually be hard, since they're a server function and the client couldn't know how it's used
20:43 Taoki Predicting movement of inventory items however would be easy to add
20:44 VanessaE Taoki: right, hence the client has to execute the same code the server would execute when conducting the craft.
20:44 sapier1 my guess it's gonna be done together with client side lua
20:44 VanessaE that means iterating through the cookbook.
20:44 * Taoki isn't fully sure about client-side Lua at this day, and how worth it it is
20:44 Taoki Though it could help with such stuff i guess
20:45 sapier1 for this special case it's not absolutely required
20:45 sapier1 maybe it's even of almost no use ;-) except of the permission things
20:53 Krock Just a little thing: I edited the font-image "fontlucida.png", so the special chars are better readable. Would this one ( http://imgur.com/xiCPGfT ) look better than the original?
20:55 VanessaE Krock: no one uses that font anymore.  use freetype.
20:55 Krock VanessaE, for the case _if_ freetype isn't used
20:55 sapier1 still if it is better I see no reason not to add it
20:55 VanessaE (besides, if freetype isn't used, use my better fonts)
20:56 kaeza Krock, how do they look in-game?
20:56 VanessaE https://forum.minetest.net/viewtopic.php?id=4436
20:56 sapier1 isn't your font bigger then original one vanessae?
20:56 kaeza as in, actual usage
20:56 VanessaE sapier1: yep.  because the default 2point font sucks :P
20:57 sapier1 well that's a point of discussion speaking for Krock's ;-)
20:57 VanessaE saqpso why not include mine as an option?
20:57 VanessaE sapier1*
20:57 Krock kaeza, it's just btter readable because it's whiter and some transpareny gotremoved
20:57 VanessaE as well as using Krock's
20:59 sapier1 well I prefere freetype too maybe we should make some example screenshots
21:12 sapier1 left #minetest-dev
21:20 sapier joined #minetest-dev
21:31 kaeza joined #minetest-dev
21:35 asl joined #minetest-dev
21:41 kaeza an user on ##minetest-es is having this issue: http://i.imgur.com/KvRui75.png
21:42 kaeza I recommended to update to latest git, but problem still persists
21:42 Taoki kaeza: Is that my workbench mod? I see a 2x2 grid so just asking.
21:42 Taoki Though it doesn't look related
21:43 ImQ009 joined #minetest-dev
21:45 Eater4 joined #minetest-dev
21:50 kaeza on a related note, this user seems to be trying to run it on a P4 with 448M of RAM (and from the screenshot, Win XP) :S
21:51 VanessaE that's winXP?
21:51 VanessaE looks more like 2K
21:51 kaeza could be *checks*
21:51 kaeza yep :S
21:51 VanessaE note the minetest version displayed - did they actually update?
21:53 kaeza that's an old screenshot; he already downloaded http://sfan5.pf-control.de/minetest-builds/minetest/minetest-0.4.9-9675d9e-win32.7z (which I don't think it's exactly latest git, but way newer than 0.4.8 or 0.4.9), but reports that the problem persists
21:53 VanessaE the displayed inventory MIGHT be realtest
21:53 sfan5 using direct3d9 might hel
21:53 VanessaE it has an inventory similar to that I think
21:53 sfan5 p
21:53 sfan5 but only my 64bit builds have that
22:08 smoke_fumus joined #minetest-dev
22:09 kaeza I guess he'll have to get a better comp :/
22:14 diemartin joined #minetest-dev
22:27 EvergreenTree joined #minetest-dev
22:36 EvergreenTree joined #minetest-dev
22:50 KingsleyT joined #minetest-dev
23:13 SEx2 joined #minetest-dev

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