Minetest logo

IRC log for #minetest-dev, 2014-09-15

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

All times shown according to UTC.

Time Nick Message
00:00 SmugLeaf joined #minetest-dev
00:00 SmugLeaf joined #minetest-dev
00:01 SmugLeaf joined #minetest-dev
00:01 SmugLeaf joined #minetest-dev
00:01 khonkhortisan_ joined #minetest-dev
00:02 werwerwer joined #minetest-dev
00:04 Megaf_ joined #minetest-dev
00:09 Robby_ joined #minetest-dev
00:19 daswort joined #minetest-dev
00:32 kaeza joined #minetest-dev
00:33 hintss joined #minetest-dev
00:34 AnotherBrick joined #minetest-dev
01:19 ShadowNinja Comments? I've got 12 pulls, 5 of them have 0 comments:  https://github.com/minetest​/minetest/pulls/ShadowNinja
01:35 VanessaE_ I've been using pull #943 for several months now
01:36 VanessaE_ no problems with it except that it causes a mild FPS drop with the explosion of particles, but that's because minetest's particles method is stupid
01:36 VanessaE_ (see also, jin_xi's irrlicht particles work)
02:13 Miner_48er joined #minetest-dev
02:15 werwerwer joined #minetest-dev
03:25 OldCoder joined #minetest-dev
05:25 Miner_48er joined #minetest-dev
05:28 kahrl joined #minetest-dev
05:30 OldCoder joined #minetest-dev
05:35 sfan5 ,tell iqualfragile Linux numbers include desktop and android
05:35 sfan5 hm
05:35 sfan5 !tell iqualfragile Linux numbers include desktop and android
05:36 sfan5 .tell iqualfragile Linux numbers include desktop and android
05:36 sfan5 hm
05:36 sfan5 ¯\_(ツ)_/¯
05:36 sfan5 ShadowNinja: how do I use HLuaBot?
05:52 Hunterz joined #minetest-dev
06:19 Weedy joined #minetest-dev
06:19 Weedy joined #minetest-dev
06:21 Weedy joined #minetest-dev
07:18 Zeno` joined #minetest-dev
07:59 jin_xi joined #minetest-dev
07:59 sol_invictus joined #minetest-dev
08:13 CraigyDavi joined #minetest-dev
08:44 ImQ009 joined #minetest-dev
09:16 TriBlade9 joined #minetest-dev
09:16 TriBlade9 Is anyone on/active/awake at this hour?
09:18 TriBlade9 Just wondering where GameScripting->loadScript() is defined. scripting_game.cpp doesn't seem to have it as far as I can tell.
09:20 TriBlade9 Nvm, found it in s_base.cpp. Boy, now I remember why I hate C++.
09:20 kahrl TriBlade9: script/cpp_api/s_base.cpp
09:20 TriBlade9 ninja'd :P
09:20 TriBlade9 Thanks though
09:20 kahrl hehe :P
09:20 TriBlade9 Before you yell at me, I'm attempting to run some test mods with the duktape JS engine.
09:20 TriBlade9 After about 8 years of not writing a line of C or C++ x]
09:21 TriBlade9 I've seen the logs of people discussing V8, they pretty much ended along the lines of "If you want it, do it yourself, then we can talk."
09:22 kahrl sounds like a big project :)
09:22 kahrl assuming you want to be able to call the whole API
09:22 TriBlade9 Nah
09:22 TriBlade9 Just get a basic file or two to load
09:22 TriBlade9 My C++ skills are way too small
09:23 TriBlade9 I can't use V8 due to China's google-blocking ...
09:23 TriBlade9 But duktape is quite simple, with comparable performance to plain Lua (Without JIT)
09:23 TriBlade9 If only C++ wasn't so hard to read >:U
09:25 kahrl if you're looking for more functions, I used grep -R to find loadScript
09:26 TriBlade9 Orite
09:26 kahrl (or you can use ctags or some IDE solution)
09:26 TriBlade9 I always under-estimate the command line...
09:26 TriBlade9 I use Geany as an IDE, not much there for C/C++ support
09:27 TriBlade9 Uggh, C++ suddenly seems like a foreign language
09:43 casimir joined #minetest-dev
10:04 Megaf sfan5: this is epic \_(ツ)_/¯
10:09 TriBlade9 What is epic Megaf?
10:09 Megaf this guy -> ¯\_(ツ)_/¯
10:10 TriBlade9 Box-face?
10:30 PilzAdam joined #minetest-dev
10:35 Amaz joined #minetest-dev
10:39 ImQ009 joined #minetest-dev
10:54 proller joined #minetest-dev
10:56 AnotherBrick joined #minetest-dev
11:10 TriBlade-Away joined #minetest-dev
11:24 Hunterz joined #minetest-dev
11:44 mberends joined #minetest-dev
11:48 darkrose joined #minetest-dev
12:28 proller joined #minetest-dev
12:28 kaeza joined #minetest-dev
12:32 casimir joined #minetest-dev
13:00 VanessaE sfan5: it's ''tell <foo> <msg>
13:02 TriBlade9 VanessaE... What?
13:02 TriBlade9 Also, Hi
13:03 VanessaE hi
13:04 VanessaE he was looking for the offline-message command for HLuaBot.  For reasons that pass my understanding, the prefix is two apostrophes.
13:04 TriBlade9 Ah.
13:04 TriBlade9 You done anything with formspecs?
13:04 VanessaE I suck at using formspecs.
13:04 TriBlade9 Most people seem to -_-
13:05 TriBlade9 IMO the formspec spec is terrible
13:05 TriBlade9 Why is it a string?
13:06 kahrl TriBlade9: because a table would be even worse.
13:07 VanessaE kahrl: the irony of that is that many, if not most modders write their formspecs out across multiple lines.  Doing it as a table isn't too far off from that.
13:08 VanessaE the problem isn't the fact that it could be a table - it's the exact syntax and the naming of the keys that would be the issue I think.
13:09 TriBlade9 Why not make it more like the C++ API?
13:09 TriBlade9 (Assuming the c++ api is similar to Button button = new GUI::Button("TEXT");
13:09 TriBlade9 )
13:09 TriBlade9 More of a MVC-based system
13:22 VanessaE not sure, but I seem to remember there being a wrapper that's similar
13:22 VanessaE a toolkit that sapier is working on
13:24 TriBlade9 Hmm
13:24 TriBlade9 What on earth is light_decode_table supposed to mean in light.cpp?
13:25 TriBlade9 I have half a mind to tweak all the numbers and see what happens.
13:25 VanessaE https://github.com/minetest/minetest/issues/1362
13:26 VanessaE you mean this? :P
13:29 TriBlade9 Pretty much VanessaE
13:29 TriBlade9 Except light itself doesn't seem to travel far enough either
13:29 VanessaE RealBadAngel has some plans to fix that
13:29 VanessaE hardware-assisted lighting using irrlicht light sources I guess
13:31 TriBlade9 RBA has tons of plans x]
13:33 kahrl https://forum.minetest.net/vie​wtopic.php?t=6222&amp;p=96903 (never tried it myself)
13:42 Zeno` Why is light calculated using software anyway?
13:43 TriBlade9 Hmm
13:43 TriBlade9 What Irrlicht UI implementation does Minetest use? The stock one?
13:43 kahrl Zeno`: what else would you use?
13:43 VanessaE TriBlade9: stock as opposed to what?
13:43 Zeno` kahrl, ummm the hardware?
13:44 Zeno` like every other engine since about 1997?
13:44 kahrl the server doesn't necessarily have a GPU
13:44 TriBlade9 VanessaE, Irrlicht comes with a GUI library
13:44 Zeno` so why is it responsible for how much light things receive?
13:44 TriBlade9 I believe there are some plugins as well, and what MC uses doesn't seem to resemble the stock Irrlicht GUI
13:44 VanessaE kahrl: what about having clients send lighting calculations back to the server (on request)?
13:45 VanessaE spread the load around
13:45 Zeno` seriously, no software has done lighting since about 1997
13:45 kahrl VanessaE: there's not necessarily a client at the required place, and it could be hacked
13:45 kahrl Zeno`: say goodbye to minetest.get_node_light and mesecons solar panels then?
13:45 Zeno` I dunno why minetest does (apart from ABMs for growing stuff I guess)
13:46 kahrl and light-dependent mob spawning etc.
13:46 kahrl and that^
13:46 Zeno` no, not and that
13:46 VanessaE kahrl: so send a request to any available client, "hey, got time to light this?"  client answers yes or no.
13:46 Zeno` those could be inferred
13:46 TriBlade9 VanessaE: http://irrlicht.sourceforg​e.net/docu/example005.html
13:46 Zeno` have you profiled the server?
13:47 kahrl Zeno`: and how would "infer" that without doing all the light calculations anyway?
13:47 TriBlade9 Are we using that with a horribly ugly skin? Or something else
13:47 kahrl would you*
13:47 kahrl VanessaE: might be possible (again, it could be hacked), but seems very complicated
13:47 Zeno` kahrl, by approximating
13:47 Zeno` who CARES about hacking the client when the rendering is so slow?
13:47 kahrl Zeno`: btw, "no software', I think you forgot MC?
13:48 VanessaE kahrl: how would you "hack" it though?  we're talking about sending back at most 512K of lighting data (before some sort of compression).
13:48 Zeno` MC approximates light amounts hitting nodes
13:49 Zeno` the actual light values meant for rendering are left to the hardware
13:49 VanessaE Zeno`: RealBadAngel's plan is like that actually.
13:49 VanessaE the problem is the server-client split
13:49 kahrl Zeno`: what do you mean by approximate?
13:49 Zeno` kahrl, I mean it doesn't have to be perfect
13:50 kahrl if you mean the kind-of flood fill algorithm with linear decay, that's the same as minetest's
13:51 Zeno` the same algorithm does not mean the same results or complexity
13:51 kahrl well, yes, it does
13:52 Zeno` anyway MC's memory layout is much more efficient because "node ids" and "node meta" are not hidden away in classes
13:52 NakedFury joined #minetest-dev
13:52 Zeno` it does? How so?
13:52 kahrl how not?
13:52 Zeno` precision
13:53 TriBlade9 What would be preferable would be for the server to know the light levels of blocks, but leave the actual rendering and visualization of the light to the client
13:53 TriBlade9 Where the client only knows the node that is emitting
13:53 Zeno` you can use the same general algorithm but sacrifice precision to reduce complexity
13:54 kahrl well, then it's not actually the same algorithm
13:54 kahrl but anyway, where does mc sacrifice precision in that algorithm?
13:54 Zeno` TriBlade9, probably, but the main problem is that ALL light levels are calculated, in software, by the server
13:54 Zeno` kahrl, it's exactly the same algorithm
13:55 Zeno` it's like 7/2 though
13:55 Zeno` MC gives an answer of 3, MT gives an answer of 3.5
13:55 kahrl I wouldn't say introsort and quicksort are the same algorithm
13:55 kahrl even if they have the same general strategy
13:56 hmmmm joined #minetest-dev
13:56 Zeno` I didn't say strategy
13:57 Zeno` How to express this a different way. floats vs. double?
13:58 kahrl oh, you mean like that
13:58 Zeno` double vs. quad?
13:58 kahrl I still don't get the difference though
13:58 Zeno` yes, like that
13:58 kahrl both mc and mt use 4 bit integers for light levels
13:58 Zeno` do quad precision calcs and you'll see the difference if the hardware doesn't support it
13:59 Zeno` anyway, none of this answers my question
14:00 Zeno` Why is light calculated using software anyway?
14:00 kahrl didn't we go over that?
14:00 Zeno` And, please, don't tell me again that it's because all apps do it in software
14:01 Zeno` No, it was never answered
14:01 Zeno` Some tangent was embarked upon without addressing the question
14:01 kahrl <kahrl> the server doesn't necessarily have a GPU
14:02 Zeno` ok, let me change the question
14:02 Zeno` Why is the server responsible for calculating lighting values used for rendering?
14:07 kahrl as I said, you could move the computations to the clients perhaps
14:07 kahrl it would be a lot more complicated though
14:09 VargaD_ joined #minetest-dev
14:09 kahrl what are folks doing that makes the lighting system so slow for them? it works fast enough for me
14:10 Zeno` I dunno
14:10 kahrl (I don't build fields full of blinky plants, though)
14:10 Zeno` Everytime I profile the server though it seems that 90%+ of the CPU time is spent in calc_lighting() (or whatever the function is called)
14:11 kahrl well ok, I don't use lua mapgens either
14:12 kahrl or anything else that uses LuaVoxelManip:calc_lighting()
14:12 celeron55 is anyone working on something at the moment in such an extent that they would like some donations?
14:12 kahrl not me
14:12 celeron55 i feel like shamelessly plugging into the MC hatetrain money by opening donations, but i can't really use them myself
14:13 celeron55 because i don't work on MT
14:14 celeron55 anyone who i don't hate and who used >10 hours to work on something MT-related last month is okay to me
14:14 sfan5 not me
14:14 celeron55 you'll have to provide a paypal or bitcoin or some other address to be put on the website
14:16 Zeno` what about minegelds?
14:16 VanessaE send all donations to me.  I need to buy a new projector :P
14:17 Zeno` hang on VE.. I need minegelds to power my furnaces!
14:19 celeron55 well i guess i can take some myself to cover a year or two of hosting, at least that's useful even while not being very fair when so much stuff is hosted out of other people's pockets... at least it makes me a bit responsible for the website
14:20 sfan5 maybe xyz wants some?
14:20 sfan5 left #minetest-dev
14:20 sfan5 joined #minetest-dev
14:20 celeron55 i have asked, he has never wanted
14:21 celeron55 i kind of respect it that nobody wants any; it's handy because with no money there's less responsibility too
14:22 celeron55 but on the other hand money is handy sometimes
14:25 sfan5 "[...]| PayPal-Donations: fuslvz-donations@sigqu.it |[...]" if you have some spare drop some money there; it's the (free and only for trusted people) hosting service that currently runs servers.minetest.net
14:26 celeron55 how much do they need
14:26 sfan5 I dunno
14:27 proller joined #minetest-dev
14:31 Hunterz joined #minetest-dev
14:32 celeron55 i'll modify the donation page a lot and see what happens
14:43 sfan5 the new page seems to make it a bit clearer who to donate to
14:44 celeron55 i made the page a lot simpler and just state the minimal required information: http://minetest.net/donate
14:44 celeron55 also, i want to generally call them "tips" to make it clearer that we are not some kind of charity
14:45 celeron55 it kind of implies it's partly some kind of personal income
14:45 sfan5 the coffe jug is a bit displaced
14:45 sfan5 over the <hr/>
14:46 celeron55 maybe now
14:46 celeron55 it's still terrible, maybe i'll just remove it completely from the page
14:46 sfan5 did you change anything
14:46 sfan5 ?
14:46 sfan5 it's still over the <hr>
14:46 VanessaE put it back
14:47 celeron55 well now there is no coffee mug
14:47 VanessaE the page needs something there.
14:47 celeron55 welp
14:47 celeron55 done
14:47 celeron55 i declare it being fine now
14:48 celeron55 if someone wants money, /msg me and i will add you on the page if you don't seem completely malicious and selfish
14:48 celeron55 and incompetent
14:50 sfan5 top: -2em; left: 2.75em; position: relative; on the mug corrects it for me
14:50 celeron55 it's impossible to do that because it's dokuwiki and that would be a global style
14:50 sfan5 like the upper left edge of the mug is at the "." of "make sure to check those out."
14:50 sfan5 hm :/
14:51 VanessaE for me it sits below "those out."
14:51 VanessaE the bottom of the mug sits on the <hr> at the bottom, as if the <hr> forms a tabletop
14:51 VanessaE (chromium 37)
14:51 sfan5 on the <hr> ??
14:52 VanessaE mmhmm
14:52 VanessaE screenshot incoming
14:52 celeron55 i moved it to bottom center
14:52 VanessaE http://digitalaudioconcepts.com/vanessa/h​obbies/minetest/screenshots/Screenshot%20​-%2009152014%20-%2010%3a52%3a29%20AM.png
14:53 celeron55 it's weird but at least it isn't trying to fight with the text
14:54 sfan5 how it looked for me: https://cdn.mediacru.sh/IolakUDSS765.png
14:54 VanessaE celeron55: sometimes life was easier with plain old tables and CSS :P
14:54 celeron55 VanessaE: that's how it looked for me
15:02 Hunterz1 joined #minetest-dev
15:02 mberends joined #minetest-dev
15:03 Hunterz1 left #minetest-dev
15:06 Calinou joined #minetest-dev
15:18 kahrl sfan5: any preferences for httpd and database backend I should try to install weblate with? nginx/postgres?
15:19 kahrl seems to be going smoothly so far, btw.
15:19 sfan5 kahrl: nginx + postgres
15:19 sfan5 or any other db server that doesn't get in my way
15:19 kahrl alright
15:20 Billistic joined #minetest-dev
15:20 kahrl well, I only have admin experience with mysql, so I can't really say
15:32 Mikaela joined #minetest-dev
15:58 Megaf joined #minetest-dev
16:17 Hunterz joined #minetest-dev
16:23 Calinou joined #minetest-dev
16:35 celeron55 postgresql all the way
16:35 celeron55 mysql is kind of the php of databases
16:47 hmmmm what is your opinion on Cassandra and ElasticSearch
16:56 rubenwardy joined #minetest-dev
17:03 jin_xi soo, anyone with shaders experience? what kind of error does this look like? wrong transforms or plain wrong? https://www.youtube.com/watch?v=hB7NDuJ5iRI
17:06 Krock joined #minetest-dev
17:06 celeron55 hmmmm: haven't tried or thought about those, i'm not that much of a database guy
17:06 celeron55 do you have an opinion about their usefulness?
17:07 hmmmm not too much.  it's what the "industry" is using these days, however
17:09 hmmmm clearly not good choices to back minetest maps.  they're distributed and intended for large datasets
17:34 sfan5 hmmmm: the .mts format is inefficient for exporting a large number of nodes
17:36 khonkhortisan joined #minetest-dev
17:36 sapier joined #minetest-dev
17:41 sapier seems we've got a new enemy to deal with, now we're facing full microsoft market power :-/
17:42 proller they start kill all mt commiters
17:42 proller one by one
17:42 proller sapier, you first!
17:43 Amaz proller, I think they would want to kill off the commiters to forks first! :P
17:43 sapier well "free" is way more dangerous to microsoft then "test" so most likely you're gonna be first proĺler ;-P
18:00 salamanderrake joined #minetest-dev
18:04 Taoki joined #minetest-dev
18:34 * Krock wonders when or how cut-off borders of mapblocks could be solved
18:52 RealBadAngel ok guys, how about that? https://www.youtube.com/watch?v=​sX_rP-f6VCY&amp;feature=youtu.be
18:54 asdaaa joined #minetest-dev
18:58 sfan5 RealBadAngel: looks good
19:00 RealBadAngel this time insted of modyfing vertex light for selected tiles i made a separate mesh based on selection boxes
19:00 VanessaE RealBadAngel: now THAT looks good
19:01 VanessaE ah, multi-node too
19:01 VanessaE good edal
19:01 VanessaE deal
19:02 PenguinDad joined #minetest-dev
19:08 jin_xi RealBadAngel: +1 very nice
19:08 celeron55 by the way, the main site's traffic is looking like this: http://imgur.com/rapDmDa
19:08 celeron55 let's say there's plenty of interest at the moment 8)
19:08 sfan5 everyone please vote for this ;) https://github.com/minetest/minetest/pull/1635
19:09 sfan5 wow
19:09 sfan5 thats much traffic
19:09 RealBadAngel celeron55, and how about you? do you like such selection boxes?
19:11 celeron55 the box is too blue and i don't like it that when you point a thing, it's harder to see it than when you don't point at it
19:11 celeron55 contrast goes down
19:11 celeron55 can you make a selection thing that increases contrast instead of decreasing it?
19:12 celeron55 that would make the most sense really
19:16 RealBadAngel i can raise the contrast of the node inside halo mesh for that
19:17 RealBadAngel node atm is untouched and the mesh is blended with TRANSPARENT_ADD_COLOR
19:17 RealBadAngel btw, color is configureable, you can make it any color you want
19:18 RealBadAngel i just like blue ;)
19:27 casimir That selection looks very good.
19:27 casimir Does it use the same color setting as for selection boxes?
19:29 casimir And what about the amount of transparency? Is there an extra setting for that? Because ingame you probably would like it to be more discrete
19:29 casimir .
19:32 zat joined #minetest-dev
19:34 RealBadAngel casimir, this is mesh, it uses textures
19:35 RealBadAngel on video you can see it using all blue one
19:36 RealBadAngel so your game or texture pack could change it to whatever color you would like to
19:37 RealBadAngel also you can use animated textures with some sparkles if you wish to make it more "holographic"
19:37 RealBadAngel just an idea
19:37 PilzAdam joined #minetest-dev
19:39 RealBadAngel PilzAdam, https://www.youtube.com/watch?v=​sX_rP-f6VCY&amp;feature=youtu.be
19:39 RealBadAngel whats your opinion?
19:40 PilzAdam why is there water in the tree?
19:41 casimir Oh, that sounds useful.
19:45 salamanderrake probably should use a box out line instead of the entire area filled in with blue
19:46 RealBadAngel salamanderrake, lines you do have already
19:46 salamanderrake oh ok
19:46 salamanderrake have not played in a long time
19:47 RealBadAngel selection boxes are made with lines, this is meant to look like that
19:48 RealBadAngel and a option to choose from ofc, propably default on android devices
19:48 RealBadAngel PilzAdam, water? lol
19:54 kahrl heh, guess what error I got when installing weblate
19:55 jin_xi 404
19:55 PenguinDad kahrl: translation errors?
19:55 kahrl nope
19:55 kahrl sadly nope
19:56 kahrl 502 Bad Gateway
19:56 celeron55 does weblate with nginx use a reverse proxy setup?
19:57 celeron55 that's like the most common error when doing those 8)
19:57 kahrl yeah, something with uwsgi I think
20:09 Miner_48er joined #minetest-dev
20:21 iqualfragile joined #minetest-dev
20:25 kahrl ugh
20:25 kahrl whatever I try, upstart won't start uwsgi
20:26 kahrl and the weblate installation instructions silently assume I know where I have to put the files they spew at me
20:26 iqualfragile why would you use upstart?
20:26 kahrl iqualfragile: because servers.minetest.net runs ubuntu 14.04
20:26 iqualfragile uh
20:29 kahrl lmao, the uwsgi tutorial for nginx+django recommends to use /etc/rc.local
20:29 celeron55 upstart is pretty good in my opinion
20:30 celeron55 too bad it's a temporary fad in this madness of init systems
20:30 kahrl so much for the elaborate /etc/uwsgi/apps-{available,enabled} system that ubuntu install
20:30 kahrl installs*
20:33 kahrl I think the fact that ubuntu sometimes uses the old /etc/rc*.d and some other times /etc/init etc. is prettu confusing
20:33 kahrl but yeah, other than that it's pretty good
20:33 kahrl pretty*
20:38 iqualfragile joined #minetest-dev
20:38 kahrl oh
20:38 kahrl apparently files in /etc/uwsgi/apps-enabled are ignored unless they end in .ini
20:39 kahrl that's pretty dumb in opinion
20:39 kahrl but meh
20:39 kahrl in my*, I can't type today :P
20:54 kahrl great... > from django.db import models, migrations   > ImportError: cannot import name migrations
20:54 kahrl these are the times when I wish python was statically compiled
20:58 iqualfragile kahrl: what did you try?
20:59 kahrl iqualfragile: well
20:59 kahrl I noticed that uwsgi won't run because the tables for weblate are not in the database yet
20:59 iqualfragile i think migrations are a feature that was added in the latest django version
20:59 kahrl so I did a manage.py syncdb
21:00 kahrl which added some tables, but not all of them, and it told me the rest need to be migrated (for whatever reason)
21:00 kahrl so I ran manage.py migrate
21:00 sfan5 kahrl: I use /etc/init.d for upstart on servers.minetest.net
21:01 sfan5 kahrl: http://sprunge.us/OXad?sh that works good enough for me
21:03 kahrl sfan5: I got it to run now, but the script it's running is missing some tables so it bails out
21:05 iqualfragile kahrl: which django version are you using?
21:06 kahrl https://gist.github.com/kahrl/35ab89659c223d74b000
21:06 kahrl iqualfragile: 1.6.1
21:06 sfan5 It sounds like this is going to be a PITA
21:06 iqualfragile kahrl: was aded in 1.7 irrc
21:06 iqualfragile (migrations)
21:07 kahrl weblate supposedly works with 1.6 though...
21:08 iqualfragile kahrl: https://docs.djangoproject​.com/en/1.7/releases/1.7/
21:08 kahrl is it available in the ubuntu repos?
21:09 iqualfragile no idea
21:09 Megaf sapier: have you changed anything on the server side on the last week?
21:09 Megaf now Im getting up to 104 seconds of lag
21:09 Megaf it went from 3 seconds avarege to 40 seconds avarage
21:09 Megaf and Im not the only one with those kind of numbers
21:10 Megaf these*
21:11 sapier megaf no I didn't change anything for some weeks
21:11 Megaf I wonder whats going on then
21:12 sapier ok I did a change in connection but that's an output for a case where an assert is caused afterwards ... guess you'd have noticed that prior the lag ;-)
21:12 kahrl man, I hate frameworks so much
21:13 iqualfragile kahrl: django is actually quite a nice framework
21:13 iqualfragile saves you lots of work
21:13 kahrl iqualfragile: but most web applications just pile framework on framework on framework on django
21:14 kahrl (or another base, obviously)
21:14 iqualfragile to fix your problem i would recommend entering the depths of dll-hell and create a virtual environment
21:14 kahrl I think it might actually be fixable by installing an older version of social-auth
21:14 kahrl because that's where the exception is coming from
21:15 scourge joined #minetest-dev
21:15 sapier there's been basicaly no basic change for about 2 weeks now Ian I doubt the voxelmanip changes can cause lag too
21:15 kahrl the fun part is to find out which version ;)
21:15 iqualfragile kahrl: whats your endziel?
21:16 sapier so any issue causing lag was there before. Hmm maybe lag is map related or related to some mechanism mods started to use or use way more frequent? It'd not be first time for mods to cause lag
21:16 kahrl iqualfragile: install a weblate for minetest
21:16 sapier megaf can you try mod profiling on your laggy server?
21:16 kahrl iqualfragile: not the one that will end up in production, but make the installation reproducible enough so that I can tell sfan5 how to do it
21:17 Megaf sapier: I tried, but I dont know how to use that
21:17 iqualfragile kahrl: ah, ok
21:17 iqualfragile i read weblate and thought it would be another framework
21:18 kahrl python-social-auth doesn't seem to be an optional dependency for weblate either :(
21:19 kahrl who needs facebook authentication really?
21:19 kahrl although github integration would be kind of neat
21:24 kahrl pip uninstall is not a thing according to man pip? nice!
21:26 iqualfragile kahrl: its mainly used in virtenvs, you can just delete the virtenv
21:26 sfan5 wasn't it 'pip remove'?
21:27 kahrl what's a virtenv? http://virtenv.sourceforge.net/ doesn't seem to be it?
21:27 kahrl sfan5: pip uninstall works
21:28 iqualfragile http://virtualenv.readthedocs.org/en/latest/
21:29 kahrl seems even more complicated than this whole mess already is
21:29 kahrl I want my CGI scripts back!
21:36 proller joined #minetest-dev
21:43 Notgnoshi joined #minetest-dev
21:46 Notgnoshi left #minetest-dev
21:48 proller joined #minetest-dev
21:49 proller joined #minetest-dev
21:59 GeneralAltoids joined #minetest-dev
21:59 kahrl almost yay
21:59 ImQ009 joined #minetest-dev
22:00 kahrl I downgraded python-social-auth to 0.1.26 and now it migrated a lot more than before
22:00 kahrl https://gist.github.com/kahrl/330ebe3761f0d0b62c3c
22:00 kahrl now enough tables are migrated that uwsgi is no longer complaining and I can finally access weblate :D
22:01 GeneralAltoids left #minetest-dev
22:02 kahrl yay
22:02 kahrl did a syncdb and then the remaining migrations worked
22:02 VanessaE but can you actually *translate* yet? ;)
22:02 kahrl heh, nope
22:02 kahrl that's for another day
22:15 VanessaE [09-15 17:59] <friti> The crash upon checking online mods repo is a known bug, right?
22:15 VanessaE [09-15 17:59] <friti> Nope, Linux, Kubuntu to be exact.
22:16 VanessaE [09-15 18:07] <friti> john_minetest: http://pastebin.com/ejyXsSJq
22:16 VanessaE [09-15 18:10] <friti> I got it from the ubuntu repo, and it's version 4.10
22:22 kahrl VanessaE: yeah, trying -dev is a good idea there (like almost always)
22:23 kahrl at least "C++ exception" would be replaced by a more meaningful message
22:23 VanessaE [09-15 18:21] <friti> Well, the version from the PPA does NOT crash. YAY!
22:23 VanessaE figured I'd echo it over here anyway
22:23 mos_basik joined #minetest-dev
22:23 VanessaE less traffic, easier for the person who maintains that package to notice and perhaps update it
22:24 hax404_ joined #minetest-dev
22:24 kahrl I think we should do a 0.4.11 soon
22:24 kahrl that'd notify them :P
22:25 VanessaE yeah
22:25 T4im joined #minetest-dev
22:27 Megaf kahrl: make it stable and noty me, then I will switch from git to stable
22:53 proller joined #minetest-dev
22:55 Megaf notify*
23:23 OldCoder joined #minetest-dev
23:27 kahrl ehh, I give up
23:27 kahrl can't get into the admin interface.
23:27 kahrl went over the django FAQ and http://stackoverflow.com/questions/1189​4765/unable-log-in-to-the-django-admin-​page-with-a-valid-username-and-password
23:27 kahrl still no luck.
23:27 RealBadAngel kahrl, for 0.4.11 i want tile switching be finished, for that we did protocol bump
23:28 RealBadAngel releasing next stable without a way to use it is pointless
23:28 kahrl RealBadAngel: I wasn't involved in that at all so I have no clue
23:28 RealBadAngel its about special tiles count being raised to 6
23:29 RealBadAngel it forced protocol bump
23:29 RealBadAngel i made this to be able to easily switch between two full set of tiles for a node
23:29 kahrl is it still WIP?
23:30 VanessaE wat
23:30 VanessaE I thought that went in already?
23:30 RealBadAngel kind of, patch is hanging around
23:30 RealBadAngel it is being blocked by my current work
23:30 RealBadAngel im working on the very same files and functions
23:30 kahrl in any case, I didn't mean "feature freeze nao"
23:31 RealBadAngel patch was a few liner to make the feature complete, but things went as usual
23:31 VanessaE the next time we "feature freeze" it needs to last WAY longer than a week or even two
23:31 VanessaE it needs to be at least a month
23:31 RealBadAngel getting anything merged this days is a miracle
23:31 VanessaE and none of this "well, we want to get the release out by X date" stuff either
23:32 VanessaE every time a deadline gets set, it either gets delayed, or the release gets hurried out with some glitch that ends up pissing people off
23:32 VanessaE (and when it does get delayed, THAT always seems to piss someone else off :) )
23:32 RealBadAngel and waiting for 3 lines of code to be merged for a few weeks and blocking any further work is just insane
23:34 RealBadAngel and when i rebase the things into one pull i hear voices "hey it shouldnt be here, its another stuff"
23:34 RealBadAngel even when  code is in the very same function, line next to another
23:35 kahrl VanessaE: keep in mind that we'll have an influx of new people who have little clue about compiling (and might be scared of inofficial builds)
23:35 VanessaE kahrl: indeed, I've been warning of this surge for a little while now
23:35 VanessaE and we're starting to see it in #minetest as you've noticed.
23:36 kahrl well, yeah :P
23:40 kahrl RealBadAngel: well, it happens to every dev
23:40 kahrl RealBadAngel: any ideas how to fix it?
23:41 RealBadAngel im not sure
23:42 RealBadAngel but one thing i know is that mt lately is not moving forward
23:42 kahrl I agree
23:42 VanessaE certainly not in places where it's visible to end-users or modders, generally.
23:42 Weedy joined #minetest-dev
23:42 Weedy joined #minetest-dev
23:42 kahrl I think we should try iqual's suggestion of a scheduled weekly meeting
23:43 kahrl and if only one person shows up, that person has the right to merge things as he pleases mercilessly
23:43 werwerwer joined #minetest-dev
23:43 * VanessaE hands kahrl a stick and two blocks of wood (to craft a gavel and its sounding block)
23:44 VanessaE ^^^ meaning, who's gonna lead such a meeting?
23:45 kahrl I doesn't really matter, as long as everyone is focused on the goal of the meeting: merging as many patches as possible (or discussing issues if there's nothing to merge)
23:46 kahrl one day sapier and I did that and it was quite effective; none of us was really "leading"
23:46 kahrl well, to be more precise, the goal is merging/rejecting/improving patches
23:47 kahrl just getting the number of open pull requests down would help a lot to increase productivity, I assume
23:48 VanessaE oh you mean that big merge fest :)
23:48 kahrl (that doesn't mean that we should close pull requests just because they're not ready yet)
23:48 khonkhortisan_ joined #minetest-dev
23:48 VanessaE it's certainly an effective idea, but is it sustainable?
23:49 kahrl that's what I'm wondering about
23:49 kahrl I think it would at least be an immediate improvement
23:49 VanessaE possibly
23:52 kahrl this might be a bad idea, but in case the donations do come in as c55 predicted, perhaps a small amount of money could go to every dev that attends this meeting?
23:53 Taoki joined #minetest-dev
23:53 daswort joined #minetest-dev
23:55 zat joined #minetest-dev
23:57 Zeno` joined #minetest-dev
23:57 kahrl I guess the meeting could be integrated with some bot help
23:58 kahrl after discussing each issue, there could be a formal vote where each dev does /msg HLuaBot + or 0 or - or C
23:58 kahrl + meaning merge, 0 meaning abstention, - don't merge now, C close
23:58 VanessaE might not be a bad idea, per se.  certainly a good incentive, but the question then becomes "how do you divvy up the funds?"
23:58 kahrl then HLuaBot would print the results to the channel as well as the github issue
23:59 kahrl VanessaE: well, uniformly

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