Minetest logo

IRC log for #minetest-dev, 2014-02-15

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

All times shown according to UTC.

Time Nick Message
00:00 sapier no they don't
00:00 VanessaE well for example:
00:00 VanessaE if we weren't using http to send media, the server could be guaranteed to know when the last possible file has been sent to a client.
00:00 sapier that's why people are afraid of complexity because they believe to have to fix/change all at once
00:00 VanessaE now, we don't know that.
00:00 VanessaE that's what I mean.
00:00 sapier that's wrong
00:01 VanessaE why?
00:01 sapier it's not related to http or not, it's only about decision to make media sending by a external source possible
00:01 VanessaE ok so maybe I get a few things mixed up
00:01 VanessaE but you get my point
00:01 sapier that's a feature, and noone did realize the consequences on adding it
00:03 sapier no I don't ... if we start mixing all together we wont progress at all, that's what minetest did by now ... but it has grown far to big to mux up everything. the only way to handle it is separating different functions
00:03 sapier this way you can replace the low level protocol without thinking about the application handling
00:03 VanessaE but the thing is, well then what do we do?
00:03 sapier and other way round
00:04 VanessaE well no you can't, if you replace the low-level protocol, you have to guarantee that all your clients can speak that protocol :P
00:04 sapier my opinion and vote is n-1 compatibility but that's not what poll shows
00:04 sapier yes but I don't have to touch the applevel protocol at all
00:05 VanessaE the problem with "n-1" compat is that if we go with that, we are stuck with the situation we have now where pretty much everything older than 0.4.10/0.5.0/whatever-the-hell-the-number-will-be is broken, even *with* 1143.
00:05 sapier therefore I can make both protocol version run in parallel for at least a version and let old one die in follow up version
00:06 sapier no it isn't for what I tested all our stable version work with minor glitches
00:06 VanessaE well remember, I've got 0.4.9-release clients on PCs (not tablets) who still get invisible players even after multiple connects.
00:06 sapier n-1 doesn't account any interim versions between stable ones
00:07 VanessaE we can't do anything about the tablet users, but if PCs are broken do, that's a problem.
00:07 VanessaE to*
00:07 VanessaE too**
00:07 VanessaE I know, I'm talking about stable releases also.
00:07 sapier well if you really have those clients this may be but I'm not sure this is true
00:08 VanessaE I'm certain it is true.  I trust my users enough to at least be able to tell me they can't see anyone when they could before :P
00:08 sapier I know from experience user bug reports are often not really "exact"
00:09 sapier maybe that user has a 0.4.9 build from forum? or his client did crash and has a broken cache ... or he did manually modify files ... yes I know thats unlikely ... but things like that happen
00:09 VanessaE I'm checking.
00:09 VanessaE stand by
00:10 sapier or uses a proxy filtering some sort of files
00:10 sapier there are various more or less likely options
00:11 VanessaE says the man who won't disable his proxy so that he can see what everyone else sees ;)
00:11 sapier yes because I know proxys can fix or break things
00:12 VanessaE yes, and if yours is "fixing" things, how are we supposed to be able to have the same fix?
00:13 sapier without this proxy I couldn't fix minetests original protocol because someone forgot to add a setting "disable curl for media download"
00:13 VanessaE christ.  the user whose laptop shows invisible players now claims it is 0.4.8
00:14 sapier and it's up to kahrl to fix this http header issue
00:14 VanessaE header issue?
00:14 sapier that's what I meant with user error reports are sometimes unprecise
00:14 VanessaE well she claims she uses 0.4.8 because she "likes it"
00:14 VanessaE one sec.
00:15 sapier http doesn't work because of minetest uses some non standard headers ... not exactly sure if those headers are correct or not
00:16 VanessaE ah, right
00:16 VanessaE forgot about that
00:16 VanessaE perhaps the same issue that causes certain servers *cough*lighttpd*cough* to return code 417 when index.mth is not found instead of 404 ;)
00:17 sapier maybe
00:17 sapier imho its absolutely useless to use http if you break it by using non standard headers
00:17 VanessaE ok I've got her to try 0.4.9 again and asked her to run it for a while.
00:18 sapier either I want to take benefit of stable out of box http server solutions or not
00:18 VanessaE of course, I get you
00:19 VanessaE taking Selah through a set of tests with 0.4.9 now.
00:20 sapier ok I'm recompiling 0.4.8 to see if this is really happening there
00:22 sapier1 joined #minetest-dev
00:26 VanessaE ok
00:27 VanessaE what a pain in the ass this is
00:27 VanessaE not because of the code but because of the chat
00:27 VanessaE the damn chat is moving so fast and Selah doesn't know how to use IRC, that I am having a hard time explaining to her how to do these tests
00:29 VanessaE Ok, with 0.4.9 and a populated cache, she CAN see players.  I am now directing her through a cleared-cache startup.
00:31 VanessaE ok, using 0.4.9-release with an empty cache, Selah cannot see any players on the first connect.
00:31 VanessaE With a populated cache on the second connect, she sees everyone
01:03 rsiska joined #minetest-dev
01:04 VanessaE so the problem is limited to 0.4.8 and prior and from my talking with her and leading her through the process of verifying this, it may have been a broken copy.  Nevertheless, she now knows not to bother with 0.4.8
01:04 ImQ009 joined #minetest-dev
01:10 OldCoder joined #minetest-dev
01:16 kahrl joined #minetest-dev
01:18 OldCoder joined #minetest-dev
01:28 lordawe joined #minetest-dev
01:28 lordawe left #minetest-dev
01:54 ImQ009 joined #minetest-dev
02:38 Miner_48er joined #minetest-dev
05:43 proller joined #minetest-dev
06:09 proller joined #minetest-dev
06:43 Gethiox3 joined #minetest-dev
07:00 nore joined #minetest-dev
07:04 RealBadAngel joined #minetest-dev
07:05 rambomedic joined #minetest-dev
07:06 rambomedic who maintains the website?
07:08 rambomedic nvm it's c55
07:10 rambomedic celeron55_ ?
07:13 proller joined #minetest-dev
07:15 nore Is there anything against merging #1119 ?
07:15 ShadowBot https://github.com/minetest/minetest/issues/1119 -- Fix rendering glitches when far from the center of the map by Novatux
07:20 troller joined #minetest-dev
07:47 grrk-bzzt joined #minetest-dev
07:51 rambomedic joined #minetest-dev
07:53 rambomedic ShadowNinja?
07:53 ShadowNinja rambomedic: Hmmm?
07:54 ShadowNinja nore: It adds jumps, ad seems a bit hacky.
07:54 rambomedic do you have ftp access to the website? I'm creating an appdata for minetest, and they want screenshots that are 16:9, and you have two screenshots there that are very near 16:9 so I cropped them and need to add them to the appdata file
07:54 nore what do you mean, jumps?
07:54 rambomedic so would uploading them to the mt server be possible
07:55 rambomedic it's http://minetest.net/_media​/screenshot_4032289578.png and http://minetest.net/_media/screen2.png but with 41 pixels cut off
07:55 rambomedic the top
07:55 rambomedic to make them 1440 * 1810 rather than 1440 * 1851
07:56 rambomedic this also serves (thankfully) to remove the message in the top left signifying the version number
07:58 ShadowNinja rambomedic: I don't have FTP access (And FTP is insecure), but I can edit the website.
08:00 rambomedic I used FTP as an umbrella term for remote file management
08:00 rambomedic I just need two screenshots uploaded
08:00 rambomedic to the mt server
08:00 rambomedic they don't necessarily need to be displayed
08:01 ShadowNinja rambomedic: Um, minetest.net isn't a image hosting site...
08:02 ShadowNinja Oh, wait...
08:02 rambomedic not personal images
08:02 ShadowNinja So you made those screenshots 16:9?  Why?
08:02 rambomedic you're hosting the two screenshots anyways
08:03 rambomedic the spec *prefers* they be 16:9
08:03 rambomedic and the two existing screenshots are just shy of it
08:03 ShadowNinja Yes,  NVM, got confused.
08:03 ShadowNinja What specification?
08:03 rambomedic appdata
08:03 rambomedic Fedora is moving to it being a requirement in Fedora 22
08:04 rambomedic it's what they use to show metadata for gnome-software
08:04 ShadowNinja What?  Screenshots have to be 16:9?
08:04 rambomedic the new package management GUI that fedora will be using
08:04 rambomedic they don't have to be
08:04 rambomedic but gnome-software prefers they be that way for formatting
08:05 ShadowNinja So... This would be for displaying in a package manager?
08:05 rambomedic but thinking about it now makes me realize 40 pixels won't make that much of a difference
08:05 rambomedic yes
08:05 rambomedic for now I'll link to the existing images
08:05 rambomedic if there are any serious issues it can be solved from there
08:06 rambomedic and who should I put as the maintainer for the appdata file?
08:06 ShadowNinja Alright.  Do you have cropped images?  And I (probably) won't be able to finish this tonight.
08:06 rambomedic there's a tag to specify the email address of the maintainer of the appdata file so that they can email updates to the spec
08:06 ShadowNinja Yourself?  ;-)
08:11 rambomedic :/
08:15 rambomedic #1149
08:15 ShadowBot https://github.com/minetest/minetest/issues/1149 -- Add AppData file by davidgumberg
08:19 VanessaE 16:9 for a computer display/screenshot is sacrilege. :P
08:19 rambomedic based on what?
08:19 rambomedic I don't think they expect your screen to be 16:9
08:20 VanessaE this ain't a television, that's what. :P
08:20 VanessaE content is usually tall and scrollable, not wide.
08:20 rambomedic they just did some maths to realize that 16:9 fits best within the UI
08:21 rambomedic the image is "natively displayed as  624x351"
09:00 celeron55_ http://minetest.net/screenshots
09:00 celeron55_ wtf has someone done to that page
09:00 celeron55_ it has an image fetched from elsewhere and the elsewhere gives 404
09:03 celeron55_ it was there to begin with? well shit
09:06 darkrose joined #minetest-dev
09:06 darkrose joined #minetest-dev
09:10 celeron55_ (i'll find some other one in place of it)
09:13 nore http://nore.mesecons.net/screenshot_885264396.png <-- thoughts on this?
09:15 nore celeron55_, I don't get 404... have you already changed the screenshot?
09:53 tomreyn joined #minetest-dev
09:57 Calinou joined #minetest-dev
11:29 Jordach joined #minetest-dev
11:49 PilzAdam joined #minetest-dev
12:02 grrk-bzzt joined #minetest-dev
12:48 Gethiox joined #minetest-dev
12:49 Yepoleb joined #minetest-dev
12:55 Gethiox joined #minetest-dev
13:02 Gethiox joined #minetest-dev
13:06 Yepoleb joined #minetest-dev
13:28 Yepoleb joined #minetest-dev
13:32 grrk-bzzt_ joined #minetest-dev
13:43 Yepoleb joined #minetest-dev
13:50 werwerwer_ joined #minetest-dev
14:06 Yepoleb joined #minetest-dev
14:23 Yepoleb joined #minetest-dev
14:36 Yepoleb joined #minetest-dev
14:48 hmmmm joined #minetest-dev
14:48 EvergreenTree joined #minetest-dev
14:48 EvergreenTree joined #minetest-dev
14:54 Yepoleb joined #minetest-dev
14:59 Yepoleb joined #minetest-dev
15:01 rambomedic joined #minetest-dev
15:08 OldCoder joined #minetest-dev
15:10 Yepoleb joined #minetest-dev
15:43 Zeitgeist_ joined #minetest-dev
15:48 joannis joined #minetest-dev
16:22 NakedFury joined #minetest-dev
16:24 OldCoder joined #minetest-dev
16:52 Yepoleb joined #minetest-dev
16:54 Yepoleb joined #minetest-dev
16:58 BrandonReese joined #minetest-dev
17:01 Yepoleb joined #minetest-dev
17:03 Yepoleb joined #minetest-dev
17:36 Sokomine hmmmm: the mod ethreal uses schematics to place decoratins. many of them are trees or tree-like structures. players tend to harvest them. now the problem is: how to regrow those schematic-based trees? placing the schematic would of course be easy, but - and that's the trouble - there may be player-built structures in the area already. normal trees do not replace existing nodes when growing
17:36 Sokomine is there a way to reach the same with schematics?
17:43 Calinou joined #minetest-dev
17:53 hmmmm sokomine, that's a problem the mod has to deal with
17:53 hmmmm decorations and their capabilities are already well defined
18:05 celeron55_ enabling that behavior sounds useful and not hard at all
18:24 VanessaE oh G*d this mapgen lag is horrible :(
18:45 hmmmm just freaking sick of people (well mostly sokomine) making constant feature requests to my lua api
18:45 hmmmm i'm not doing it.  if you want it done, do it yourself.
18:49 hmmmm no doubt he wants to use place_schematic, except place_schematic is being abused to all hell... it was never meant to arbitrarily place schematics, just merely to test what you captured from the map in create_schematic
18:50 hmmmm schematics are meant to be map-generation-time only things
18:50 celeron55_ but if people find a feature handy, why not extend it rather than force some other unhandy api to them?
18:51 hmmmm if they want to do that then it's fine
18:51 celeron55_ as you i would be glad that i would have made something that people like
18:51 hmmmm nagging me constantly to add it on the other hand TO MAKE THINGS EASIER FOR A SPECIFIC MOD is just something that makes me not want to do it
18:54 celeron55_ well that's kind of dumb, assuming they've already seen you specifically won't make the stuff
19:10 NakedFury joined #minetest-dev
19:20 Sokomine hmmmm: sorry. but it's a really useful feature, that schematic one. i like it a lot. things that are useful will get more feature requests than those nobody ever uses
19:21 Sokomine hmmmm: and i was mostly asking if there is a way i didn't know about. you're the expert there and certainly know more than i do. please regard it as holding what you created in high value and nothing else
19:22 Sokomine hmmmm: and i'm afraid it's similar on other levels: to a modder, those constant feature requests from people who use the mods are sometimes annoying, and i bet those irrlicht people arn't terribly enthusiastic either if minetest people find a bug or want a feature :-)
19:25 Sokomine hmmmm: in this particular case, the creator of the mod did use your decoration-feature in the way it was intended. it's just that there has to be a way to place the thing again if players harvested it. an l-system tree might have been easier in that case. if you want to see how well it works, take a look at that xanandu server. it makes heavy use of what you wrote
19:28 hmmmm sokomine, the right way to do regrowing trees is to put an ABM on the bottom of trees and if it's cut down except for that last log node it'll regrow
19:28 hmmmm (cut down, i.e. in the space above it is air)
19:30 Sokomine hmmmm: that's an option as well. at least then people would know where to expect plants regrowing. most of them would cut down the last log as well. and if it regrows from a new trunk with air above placed somewhere, then it's like a normal sapling
19:30 rambomedic joined #minetest-dev
19:31 rambomedic sfan5 PilzAdam #1149
19:31 ShadowBot https://github.com/minetest/minetest/issues/1149 -- Add AppData file by davidgumberg
19:31 sfan5 well
19:32 rambomedic maybe put you as the maintainer of the appdata file?
19:32 sfan5 hm
19:32 sfan5 what if I'm on vacation
19:33 rambomedic I can maintain it
19:33 sfan5 it would be perfect if we had a mail addr for ever dev
19:33 sfan5 every*
19:33 rambomedic maybe have a mailing list?
19:33 sfan5 I'm pretty sure we don't have mailing list
19:34 rambomedic my point was to *make* a mailing list
19:37 rambomedic I doubt there will be little changes to the spec
19:37 rambomedic the email address only serves to email maintainer's of appdata with updates on spec.
19:37 rambomedic just put your's or PIlzAdam's
19:37 rambomedic I'm keeping track of the spec anyways
19:44 sfan5 hm, ok
20:04 rambomedic joined #minetest-dev
20:05 rambomedic sorry sfan5 lsot connectio ndid you see anything
20:05 sfan5 what is 'anything'?
20:05 sfan5 oh
20:05 sfan5 no I did not say anything
20:05 rambomedic *did you say anything
20:33 nore thoughts on #1119?
20:33 ShadowBot https://github.com/minetest/minetest/issues/1119 -- Fix rendering glitches when far from the center of the map by Novatux
20:33 nore the flickering thing has been fixed now
20:36 celeron55_ nore: don't store the offset in every object in existence
20:37 celeron55_ just make them get it from the environment into where they have access to anyway
20:37 nore ok, I will try that
20:37 celeron55_ when they need it
20:38 nore do the particulespawners have access too?
20:38 celeron55_ dunno
20:39 celeron55_ but they're even more important because there can be a huge amount of particles
20:41 nore what is the problem anyway? even with 10k particles, that will only make 60k more memory used, i.e., almost nothing
20:42 celeron55_ it increases random memory access which is slow
20:42 rambomedic joined #minetest-dev
20:42 celeron55_ or, is that done like once in never?
20:43 celeron55_ it looks like it's done on every update whatsoever
20:43 nore well, the code is executed once you get more than 200 blocks from the origin
20:43 nore ah, I see what you mean
20:43 nore yes, I should only make the call if it has changed
20:45 nore but usually, the offset will only change once every 10-20 seconds when in fast+fly mode
20:45 celeron55_ but still it adds unnecessary boilerplate to many things
20:46 nore but yes, I will try to make that change (at least for objects, and I will see if particles have env access)
21:27 hmmmm i don't get it
21:28 hmmmm why do people do unnecessarily inferior ways of accomplishing things
21:28 hmmmm and then argue not to do it a better way because "it doesn't matter"
21:28 hmmmm "you'll only be saving 500k of memory"
21:29 hmmmm i've encountered this personally with 3 other people and it makes me want to punch something
21:31 VanessaE hmmmm: maybe because they don't want to have to rip apart everything they've done already just to see if the perhaps-better idea is indeed better for their application?
21:32 hmmmm the change is usually trivial and it just makes so much more sense so you wonder why they didn't do it right the first time
21:47 sapier joined #minetest-dev
22:08 Taoki_1 joined #minetest-dev
22:09 Taoki joined #minetest-dev
22:14 hmmmm hmm
22:15 hmmmm alright, I want to make a specific ore type that's a blob (deformed sphere)
22:17 hmmmm I'm thinking I could probably accomplish this by drawing a circle by plotting points at (cos(deg) * radius, sin(deg) * radius) where radius is specified by the ore type and then has (rand() % 3 - 1) added to it
22:17 hmmmm and then I do a fast flood fill of the shape
22:17 hmmmm can anybody come up with a better way to do this?
22:18 sapier sorry I don't have a better idea ... do you have an idea how to get rid about wstrings?
22:18 hmmmm rather, the radius is varied by + or - 1 for each unique pixel plotted
22:18 hmmmm no clue
22:19 sapier I'm trying to get formspecs work correct on android but for a (yet) unknown reason field names are messed up after first event ... seems like some memory corruption
22:19 sapier -seems + looks
22:20 hmmmm i hate formspecs
22:20 sapier well that's not a formspec issue
22:20 hmmmm yea.. still though
22:21 hmmmm soo, did you try setting a memory write breakpoint on the field names? :/
22:22 sapier considering how simple they are it's amazing what you can do with em ... haven't got debugger running on remote android emulator target yet
22:23 sapier did you know bionic semaphores return with error code busy if you try to delete them without having used them? ;-)
22:24 hmmmm did not know that
22:24 hmmmm well that doesn't sound right... just on android, correct?
22:24 sapier >>V/Minetest( 1813): Buttonhandler: tab: adv fields: {["main_tabMain,Advance"] = "2"}<< that's what I see on console .. that field obviously should be named "main_tab" only :-)
22:24 sapier It's a bug in android known as od 2.3 ... but seems to be still in there in 4.2
22:25 sapier and it's not messed up on translation to lua it happens before
22:28 VanessaE ?
22:28 VanessaE what's the net effect?
22:32 sapier well you can click exactly one button as after this data is messed up completely
22:45 VanessaE I wonder how many more obscure bugs like that are affecting minetest :P
22:47 sapier that memory issue is strange because there's no sign of it on x86 with gcc nor clang
22:51 OldCoder joined #minetest-dev
22:56 hmmmm agh
22:56 hmmmm now i'm having second thoughts about my 'great idea' to set default settings through Lua
22:57 troller 02:57:09: ACTION[ServerThread]: proller places node default:cobble at (467,10,0) 02:57:10: ACTION[ServerThread]: proller times out. List of players: Admin Guest1830
22:57 hmmmm well then again... alright here:  global settings are global.  the way you set mapgen noise parameters through mods right now is to simply set the global config setting string for it
22:58 hmmmm so even if that mod is unloaded afterward the effect will still stay
22:58 hmmmm so setting the defaults for noise params has the same effect; it's persistent throughout the lifetime of the process
22:59 hmmmm nobody was worrying about this before, so i doubt anybody will worry about it now
23:00 celeron55_ troller: lol, is that some bug made or exposed by sapier once again
23:00 celeron55_ how is that even possible
23:03 hmmmm you rag on sapier a lot
23:04 hmmmm i personally think he does great work.  do you see the sheer volume of stuff he does for minetest?
23:04 hmmmm bugs are inevitable
23:04 VanessaE nevermind that troller is probably using freeminer.....
23:04 VanessaE :P
23:05 VanessaE I cannot reproduce.
23:05 VanessaE (not with mainline + #1143 anyway)
23:05 ShadowBot https://github.com/minetest/minetest/issues/1143 -- Protocol version <= 22 compatibility mode + clean new client init by sapier
23:07 rambomedic joined #minetest-dev
23:08 hmmmm at this point freeminer diverged from minetest a substantial amount
23:08 sapier prollers log is possible e.g. if server is busy not beeing able to handle the events, timeout is done at network level, who knows for how long that action was queued
23:08 hmmmm almost no freeminer patches apply cleanly to minetest
23:08 MichaelRpdx joined #minetest-dev
23:09 sapier case server was hanging before it's quite possible that timeout of a client is recognized once server can handle packets again ... the way more interesting issue would be, 1) was there a hang and 2) what did cause the hang
23:11 VanessaE perhaps but it would take, what, 2 or 3 minutes of straight lag to do that wouldn't it?
23:12 VanessaE not even my Survival server has ever lagged THAT bad.  in fact the only time I've EVER seen a minetest server lag badly enough to actually induce timeouts was due to /clearobjects or large worldedit operations.
23:12 VanessaE ("has ever lagged" as in lag due to normal day-to-day player actions)
23:13 sapier yes a lag that big should be recognizable
23:13 VanessaE yes
23:13 sapier oh ... and there's another case where this happens
23:13 sapier if server is shutdown as timeout is set to almost zero in this case (I'd have to look for the exact value)
23:14 VanessaE yeah well I don't think that's the case here :)
23:14 VanessaE unless he triggered some obscure mod crashj.
23:14 VanessaE -j
23:15 sapier well it's just guessing without more details
23:18 hmmmm usually our guesses as to why a bug happens turns out to be correct
23:18 hmmmm s/turns/turn/
23:20 deltib joined #minetest-dev
23:20 hmmmm https://github.com/kwolekr/minetest/commit​/c603a47f7c0b8806ca4fd12523fdbdc965001fff
23:20 hmmmm comments complaints?
23:21 MichaelRpdx joined #minetest-dev
23:22 sapier may g_settings.get throw an exception?
23:23 hmmmm it may, why?
23:23 sapier in this case the empty check dosn't have any effect?
23:24 sapier in settings.h imho the porting.h should be removed not commented
23:24 hmmmm oh whoops
23:24 hmmmm I forgot to delete that totally
23:24 hmmmm i just wanted to make sure there were no more dependencies on that
23:24 sapier I thought so :-)
23:24 hmmmm oh no i have to modify settings.h again
23:25 hmmmm there's another minute and a half of my life wasted again
23:25 * hmmmm sits and watches shit scroll on the screen
23:28 sapier I don't see any obvious errors in there guess it's fine
23:29 hmmmm basically that whole thing is so that you can do this:  http://pastebin.com/qZku8igS
23:36 sapier that's way more readable
23:36 sapier I'm giving up on this damn memory issue for today I guess I need to get some sort of valgrind running on target
23:37 hmmmm btw those mapgen params I pasted in there are REALLY nice looking
23:39 MichaelRpdx joined #minetest-dev
23:55 rsiska joined #minetest-dev

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