Minetest logo

IRC log for #minetest-dev, 2013-11-04

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

All times shown according to UTC.

Time Nick Message
00:00 ritchie joined #minetest-dev
00:05 Taoki joined #minetest-dev
00:05 kahrl about the status of httpfetch:
00:05 kahrl OldCoder successfully tested the remote_media part of it
00:06 kahrl I'm also pretty confident that that part works
00:06 kahrl (can't rule out bugs of course, as always)
00:07 kahrl this also means the httpfetch core should work fine, and the lua api on top of it is thin enough that it should be OK as well
00:07 kahrl I tested the serverlist fetching several times and had no problems with it
00:08 kahrl now about the problems:
00:08 kahrl 1. it uses CURLOPT_PROTOCOLS which is a recent enough addition that the officially recommended binary curl build on windows does not contain it
00:09 kahrl (the offending lines could just be removed, though)
00:09 kahrl 2. not all uses of curl have been converted to httpfetch - but this could be done later (this means somewhat ugly code in the meantime)
00:11 kahrl 3. as sapier noted, it does not provide generic asynchronous calls, it just narrow-mindedly does http
00:12 kahrl (it also probably conflicts with his async events pull #977, at least the serverlist part)
00:14 kahrl 4. it adds a few more or less complex subsystems (httpfetch core, ClientMediaDownloader)
00:19 kahrl to conclude, it might actually be feasible to merge it with only minor modifications
00:19 ShadowNinja What do you think of changing README.txt to README.md? That way GitHub will parse it as Markdown, and it is already very close to markdown.
00:19 kahrl not sure if before or after 0.4.8
00:21 ShadowNinja kahrl: ^?
00:21 kahrl ShadowNinja: well .txt has the major advantage that windows users can simply click it
00:23 ShadowNinja Hmmm, but it's more readable on GitHub. I don't think most windows users are going to be reading it though. And thost that do will likely know enough to open it with notepad or whatever teir prefered editor is.
00:23 ShadowNinja their*
00:23 ShadowNinja Of course if we could tell GitHub to parse /README.txt as Markdown...
00:24 kahrl I don't really find the current version unreadable on github
00:24 ShadowNinja It isn't unreadable, but the Markdown version is easier.
00:26 Miner_48er joined #minetest-dev
00:30 Exio4 the console thingy, i fidn't keep it going because copypaste support
00:30 Exio4 didn't*
00:31 VanessaE the console has another problem
00:31 VanessaE under linux, it does not properly release mouse focus.
00:31 VanessaE e.g. the mouse still pans/pitches while the console is open.
00:31 Exio4 hmm?
00:31 VanessaE this is kinda cumbersom
00:32 VanessaE try it.  press F10 and move the mouse around :)
00:32 VanessaE in windows, it releases the mouse focus so you can highlight/copy/paste
00:32 VanessaE or so I was told.
00:32 kahrl I don't remember coding that...
00:32 Exio4 neither i saw code for thzt
00:33 Exio4 i thought it was a feature
00:33 kahrl it should be changed if copy/paste support is to be added
00:33 Exio4 true
00:33 VanessaE hm, it does the same in wine.  how is it that people are able to copy&paste from it then?
00:33 kahrl they're lying
00:34 VanessaE I'd call a claim a lie if it didn't also come with the actual paste :P
00:34 Exio4 they can't copy
00:34 VanessaE still, you can't shift-arrows to select either
00:34 VanessaE so F10 is impossible to copy out of.
00:36 Exio4 i was going to implement the ctrl-v support at first, bevause irrlicht sucks on linux, wasn't able to try if working :(
00:36 Exio4 because*
00:36 VanessaE and by extension you cannot copy&paste within the F10 drop-down either.  e.g. write part of a command, copy it to the clipboard, then finish writing it so you can paste it again later
00:36 Exio4 can't you?
00:37 VanessaE pilzadam will tell you copy&pase works on linux if you get KDE's Klipper involved, but that doesn't help the other 99% of the linux userbase :P
00:37 Exio4 99.9%
00:38 VanessaE nope, you can't highlight in the F10 console, by any means, either in Linux native client or under wine.
00:38 Exio4 i was able to copypaste using vbox+wxp as clipboard manager too
00:38 VanessaE it needs to support the X10 highlight-to-copy/middle-click-to-paste method also.
00:38 VanessaE er X11
00:42 ShadowNinja VanessaE: You were able to paste in the old/current chat box?
00:42 ShadowNinja I can't copy or paste to or from anything in Minetest on Linux.
00:43 Exio4 it works under/on windows
00:43 Exio4 irrlicht the problem
00:43 Exio4 is*
00:43 ShadowNinja Yes, I know that much.
00:44 VanessaE ShadowNinja: I can copy out and paste into it in wine, but I can only copy out in linux native.  and only with ctrl-c/ctrl-v
00:44 VanessaE sometimes if you try to paste into minetest, the program you copied from will crash.
00:44 Exio4 tf?
00:44 ShadowNinja Someone should write a patch for irrlicht ;-)
00:45 ShadowNinja Exio4: Yep, I heard of that too.
00:45 VanessaE tf?
00:45 Exio4 how is that even possible
00:45 Exio4 VanessaE: the fuck?
00:45 VanessaE I just now tested it with current git on both wine and linux native.
00:46 VanessaE I can hit 't' or '/', highlight with mouse or keyboard, copy, and then paste into e.g. here from either wine or linux native.
00:46 VanessaE but only wine allows me to paste into the T / dialog.
00:46 VanessaE other dialogs behaved similarly last time I tried.
00:47 ShadowNinja Oh, Minetest -> Minetest copy-paste works.
00:47 VanessaE yes, that always seems to work within a given session.
00:47 VanessaE if I copied from MT, I can paste it back.
00:47 VanessaE but not if I copied from e.g. IRC or my terminal
00:48 VanessaE U can even copy ans paste between the wine and the linux native client in either direction.
00:48 VanessaE s/U/I/
00:48 VanessaE s/ans/and/
00:49 troller ShadowNinja, https://github.com/minetes​t/minetest/pull/979/files
00:49 VanessaE keyboard or mouse highlight, as long as I use ctrl-C/ctrl-V to perform the actual copy&paste.
00:52 ShadowNinja troller: Looks good.
00:53 troller with super-auto load any
00:53 troller will commit
00:53 ShadowNinja troller: But squash it.
00:53 troller ofcource
00:53 ShadowNinja troller: Or better yet, just use --amend.
00:54 troller i have git shortcut
00:54 troller g upb
00:54 ShadowNinja I know, I have seen your .gitconfig.
01:03 troller https://github.com/minetes​t/minetest/pull/895/files - now with breaked abms processing (if took to much time)
01:18 ShadowNinja troller: Fix your git E-mail to match one of your GitHub E-mails.
01:19 ShadowNinja troller: And use "git commit --amend" in pull requests. It means no squashing later.
01:20 troller oops, mail fixed
01:20 troller I'm afraid any -- git params ;)
02:21 ecube joined #minetest-dev
04:24 Kamski joined #minetest-dev
05:14 ShadowNinja Does Settings support things like \# String with spaces, hashmarks, and \= equals signs = It's value?
05:32 ShadowNinja Warr1024 mentioned that leaves aren't growing on trees spawned via saplings anymore. Looking at the code nore appears to have used a seperate VoxelArea in a very incoreect way. He also has things like yy and y both being vars.
05:32 ShadowNinja I will try to rewrite this and fix it, but I think schematics are best.
05:33 VanessaE cy1 mentioned that he was seeing a similar artifact on my server also.
05:33 ShadowNinja They have a probability per-node, apparently PilzTest sets the probability to 100% for all nodes.
05:33 VanessaE ShadowNinja: take the "beech" trees model from moretrees, size it down a tad.  should be suitable for the default tree.
05:34 VanessaE schematics won't work because they have no randomness I guess.
05:34 VanessaE (no shape randomness I mean)
05:35 ShadowNinja VanessaE: There is the issue of speed. L-Systems are fast enough for saplings, but not for mapgen.
05:35 VanessaE true,.
05:35 VanessaE good point on that.
05:35 ShadowNinja Of course best to just fix L-Systems...
05:35 VanessaE the issue there apparently is the method used for lighting the trees.
05:37 ShadowNinja Perhaps hmmmm is experienced enough with lighting to fix that.
05:53 darkrose joined #minetest-dev
05:54 ShadowNinja Any agreement on SamII+Cape? A number of servers have implemented it, but nobody has merged it yet...
05:55 VanessaE I use SamII but not the model with the cape
05:55 VanessaE it's a 1000% improvement over Sam I, for sure.
05:56 VanessaE imho the cape, Sam-II skin, and Stu's improved articulation model should be made the defaults.
05:56 VanessaE (the one with knees/elbows, but not the wield3d version as that's a bit buggy)
05:59 ShadowNinja But I need approval from another dev and they are mostly away now I believe. And anyway I had better get going. Night.
06:00 VanessaE yep I know.  just pointing it out.
06:00 VanessaE night.
06:04 VanessaE I'm out too.
06:50 Ritchie joined #minetest-dev
08:09 Akien joined #minetest-dev
09:01 troller https://github.com/proller/minetest/commit​/f010e1fc0e6d25c834f7809212e6c4b31da68fdf
09:22 smoke_fumus joined #minetest-dev
09:35 celeron55 I'm strictly against README.md; it's very narrow-mindedly github-only
09:45 thexyz markdown is github only?
09:47 thexyz and that readme.md isn't even valid markdown, by the way
09:48 celeron55 it's github-only in practice
09:48 celeron55 i don't care about ideologies in here
09:49 thexyz oh, I see
09:49 thexyz and what's the problem with it?
10:19 celeron55 the problem is that most people read it in a cloned repo or a downloaded source
10:20 celeron55 and in there the markup means nothing and is just in the way
10:20 thexyz it's not "in the way", it reads exactly as your README.txt
10:20 celeron55 (or a final distribution)
10:20 celeron55 why can't github be set to show it with the markup even while it's .txt?
10:20 thexyz this is one of the points of Markdown
10:20 lanxu Well, just don't use the "special" markdown. Lists and such work nicely with and without github rendering
10:21 thexyz > Readability, however, is emphasized above all else. A Markdown-formatted document should be publishable as-is, as plain text, without looking like it’s been marked up with tags or formatting instructions.
10:21 celeron55 it just annoys people when they have to select a text editor from a list of 100 other programs
10:21 celeron55 for opening it
10:22 celeron55 on the other hand, it would annoy nobody if github had a checkbox for formatting it regardless of extension
10:22 thexyz on windows? perhaps; on any linux distribution it will open in some text editor
10:23 lanxu Atleast Bitbucket shows readme.txt or README. I thought github does that too...
10:27 celeron55 i guess it can be tried; but if we get complaints, it will be changed back
10:29 thexyz sure, complaints
10:29 celeron55 what?
10:30 thexyz why the hell would anyone complain about markdown? just for the sake of complaining
10:30 celeron55 not for markdown, but for .md
10:30 thexyz I'll fix that README.md if no one is against it https://gist.github.com/xyzz/e635c228e131b2afcbe4
10:33 thexyz I don't understand why would anyone prefer .txt over .md; on the other hand, the advantages of using .md can be clearly seen — i.e. text editors can highlight markdown syntax, various tools (github, yea) can support its rendering
12:52 Evergreen_ joined #minetest-dev
12:55 EvergreenTree joined #minetest-dev
13:18 hmmmm joined #minetest-dev
14:10 EvergreenTree I know I keep asking, but why can't VanessaE's lib_6d be merged?   Is it because it's in its own file?
14:11 ImQ009 joined #minetest-dev
14:15 ImQ009 joined #minetest-dev
14:57 zat joined #minetest-dev
15:23 iqualfragile joined #minetest-dev
15:33 troller who have access to http://minetest.net/servers  ? change <div id="servers_table"></div> to <div id="servers_list"></div>
15:35 troller or output:'#servers_table'
15:43 celeron55 umm... changed it to servers_list and nothing happened
15:54 troller oops, -> server_list
15:55 Jordach joined #minetest-dev
16:02 SmugLeaf joined #minetest-dev
16:04 celeron55 it still bugs the layout
16:05 troller yes, need to find correct css hiding lihe height:0
16:05 troller like
16:07 troller https://github.com/minetes​t/minetest/pull/895/files - best patch of year nomination
16:32 VanessaE EvergreenTree: I believe the intent is to merge it at some point, but it is not a separate file now, it's gonna be part of builtin/misc_helpers.lua .
16:44 Calinou joined #minetest-dev
16:55 sapier joined #minetest-dev
16:58 VanessaE Calinou: #959 - proposal to remove the "forcewall" flag (whose presence allows moreblocks-style sneak-to-force-wall).  comments?
17:00 Calinou https://github.com/minetest/minetest/pull/959 ?
17:00 Calinou oh
17:00 VanessaE yeah
17:00 Calinou I don't get what that thing does
17:00 Calinou is it purely a code cleanup or does it add a feature?
17:01 VanessaE it take the code I put into moreblocks, homedecor, and a couple others, polishes it up, and makes it available to everyone.
17:01 VanessaE so that mods have a single function they can call.  less duplicated code.
17:02 Gethiox2 joined #minetest-dev
17:05 Miner_48er joined #minetest-dev
17:10 VanessaE it also adds a couple of features
17:10 VanessaE in the old code, normal facedir was ignored since it was directed primarily at slabs.  this one allows normal facedir on the floor and an equivalent on the ceiling, and it has that invert feature we discussed that you didn't like :)
17:13 BlockMen joined #minetest-dev
17:14 sapier BlockMen could you try this fix: https://github.com/sapier/​minetest/tree/fix_json2lua
17:15 BlockMen sure, give me a second
17:20 BlockMen sapier, works fine for me
17:21 Akien joined #minetest-dev
17:21 sapier so you see mods to download now?
17:21 BlockMen no, wait. beginning at page 3 it says: "not a single version specified"
17:22 sapier that's ok
17:22 BlockMen ah, ok. and yes
17:23 sapier those mods don't have any downloadable file ... quite useless
17:25 sapier you could try 988 it's handled better there
17:26 sapier oops 977
17:27 BlockMen will do later
17:35 BlockMen joined #minetest-dev
17:43 rubenwardy joined #minetest-dev
17:44 deltib joined #minetest-dev
17:46 troller 30 clients on arm is too laggy 8(
17:47 thexyz unbelievable
17:48 troller they moving every second, but map not loading
17:48 troller 31/100 31
17:49 troller also 78.42% [idle]  from 4core i5 with 30 bots
17:49 troller * with my patch
17:50 thexyz what patch?
17:50 sapier sounds like great improvement
17:50 troller https://github.com/minetes​t/minetest/pull/980/files
17:51 troller now will try solve block loading..
17:52 thexyz not sure how's that related to server performance
17:53 sapier is client cpp even compiled into server?
17:54 troller this patch for bots
17:54 troller https://github.com/minetes​t/minetest/pull/895/files  - this for server
17:55 sapier what are bots meant for?
17:55 sapier ok 895 is a little bit tricky to judge
17:55 troller screen nice ./minetest --go --address $address --port $port --name $n
17:55 troller for testing server 8)
17:56 thexyz you mean, your i5 can handle 30 bots with 78.32% cpu load?
17:56 troller no, 320% load
17:56 troller 78 free
17:56 sapier i5 is dual core with hyperthreading right?
17:56 Calinou quad core without HT on desktop
17:56 Calinou on mobile, dual core with HT
17:57 PilzAdam joined #minetest-dev
17:57 troller its 2-core HT
17:57 troller notebook
17:57 Calinou then it's mobile i5
17:57 sapier ok guys it's great we're improving minetest that fast but did anyone look at pull request queue recently?
18:00 VanessaE Calinou: so, will you use #959 with the invert flag (you check for 'sneak' and set it when you call the code, places a wall variant on the floor, and a floor variant on the wall), or must you stick with the forcewall flag (what moreblocks basically does now, placing wall variants)?
18:01 sapier anyone to check this one so I can merge it? https://github.com/minetest/minetest/pull/982
18:01 troller maybe increase max players in #895 ? to 30 for example
18:01 troller sapier, 892 okay
18:01 Calinou whatever works, but I'll only change code when it's merged
18:04 EvergreenTree joined #minetest-dev
18:05 VanessaE Calinou: ok.  well it's up to the others here to decide if it's merged, but I wanted your comments on which method you prefer.  by default, both are available.
18:05 sapier any core dev to agree to 892? can't merge without ok ;-)
18:07 troller sapier, okay
18:07 sapier sorry I'm the new guy I need to stick to the rules ;-)
18:10 thexyz sapier: can you explain what exactly this fixes
18:11 sapier yeah actually this fixes a very silly bug, when you do a asString() a temporary string object is created
18:11 sapier doing a c_str() afterwards creates a pointer to a constant string within that temporary object
18:12 sapier so after that operation we have a pointer to a object that may not even exist anymore
18:12 werwerwer joined #minetest-dev
18:12 sapier most time it will still exist but depending on compiler optimization there's no guarantee
18:13 sapier on linux this typicaly seems to be fine while it almost everytime fails on windows
18:14 celeron55 that's common sense to any C++ developer
18:14 sapier yea silly bug I know ;-)
18:15 celeron55 (also one of the reasons why C++ is bad)
18:15 sapier developer beeing lazy is not a reason for something to be bad ;-)
18:15 sapier you need to know what you're doing in c as well as in c++
18:16 celeron55 it's bad because a well designed language keeps you from accidentally making that kind of bugs, but C++ is made that way because it's compatible with C and so on 8)
18:17 sapier yet json lib seems to beave different in win and linux ... at least it doesn't behave like I rememer it behaving in linux
18:17 thexyz then it's okay (despite really strange codestyle in convert_json.cpp)
18:17 thexyz but that's unrelated
18:18 sapier yes c++ having c compatibility is reason for many insane things ... but c was already insane so you couldn't expect anything else :-)
18:18 sapier ok I'm merging now
18:19 celeron55 well C is just a multiplatform macro assembler
18:19 celeron55 it just looks like something fancier while it isn't
18:22 sapier c is as mighty as any assembler language but as dangerous too ... as is c++
18:39 RealBadAngel joined #minetest-dev
18:51 sapier celeron did you have time to check https://github.com/minetest/minetest/pull/971 I added winapi lookup of locales
19:03 celeron55 now only if that was include inside windows or gettext!
19:03 celeron55 included*
19:04 sapier :-) no chance as gettext guys dropped msvc support for licensing issues
19:07 mrtux joined #minetest-dev
19:30 Miner_48er joined #minetest-dev
19:32 troller #max_simultaneous_block_sends_server_total=20    why???
19:33 SpeedProg joined #minetest-dev
19:39 celeron55 troller: i don't know; how does it behave if that is modified?
19:40 troller now when you have 10 sloow clients - to every queue = 4 (small too), = maybe 40 blocks in queue
19:40 troller but now it limited by 20 and you can wait minute to load block near you
19:41 troller it must be max_simultaneous_block_sends_per_client * max_users
19:41 troller or not limited
19:44 celeron55 eh, doing that is the same as removing the whole thing
19:44 celeron55 (which is a possibility)
19:45 troller if this limit enabled - block load flag enabled
19:45 troller lag
19:49 Zeitgeist_ joined #minetest-dev
19:53 khonkhortisan joined #minetest-dev
20:10 EvergreenTree joined #minetest-dev
20:12 VanessaE What's the word on using 'killall' to shut down a server?  is it safe?  is there a particular signal I should be using?
20:18 celeron55 kill and killall will send SIGINT which is the correct way
20:18 celeron55 if you send multiple SIGINTs, it will terminate immediately without cleaning up though
20:18 celeron55 uhm
20:19 celeron55 actually wait, that's not the default signal of killall and kill
20:19 VanessaE ok.  I've been using SIGQUIT out of habit.
20:19 VanessaE the default is TERM isn't it?
20:19 celeron55 SIGINT is what CTRL+C will send and that is handled as a "do a safe shutdown" signal
20:19 VanessaE ok
20:20 sapier wasn't there the rule you can't actually kill process just tell him to commit suicide ... yes I know semantic detail ;-)
20:20 VanessaE `sudo killall -WHATEVER minetestserver`  is 10x easier than signing into all five servers and shutting them down one at a time ;)
20:20 celeron55 yes, but the C library will default to proceeding with an immediate suicide 8)
20:20 sapier yes but that's not a clean shutdown vanessae
20:20 VanessaE sapier: hence my question.
20:20 celeron55 it is if WHATEVER is SIGINT
20:21 sapier we can't do anything about it sigquit can't be handled by user process as far as I know
20:22 troller btw client dows not send disconnect packet on kill
20:22 sapier you can't do anything on kill ;-)
20:24 VanessaE sapier: well on SIGKILL no, but he's right that it should try to to a proper disconnect if it can :)
20:24 sapier of course :-)
20:27 troller sapier, cant do on -9, on most other signals- can
20:27 sapier only -9 is kill
20:28 sapier yes I know the kill command is used for other signals too ... no idea who decided to call that programm "kill"
20:28 sapier I guess this is "start" button of unix os
20:31 VanessaE sapier: from the same mindset that put "Run" and "Stop" on the same key. :)
20:33 celeron55 sapier: the name probably comes from the fact that it probably has initially only been used for killing processes 8)
20:33 celeron55 (in early unix)
20:34 sapier maybe ... so someone forgot refactoring ;-)
20:47 sapier https://github.com/minetest/minetest/pull/884 any comment on this one?
21:28 BlockMen left #minetest-dev
21:40 troller https://github.com/minetes​t/minetest/pull/895/files need to test with 50+ players ;)
21:41 sapier you should try to find them in minetest ;-)
21:43 troller in weekend maybe
21:45 troller and with it every can write lot of slow mods
21:48 ShadowNinja Any objections to reverting making the creative hand dig instantly? It makes punching impossible in creative.
21:57 troller 20 clients on arm without lag
21:58 troller with liquid_finite
22:10 ShadowNinja Alright, can someone agree to ^? troller, sapier?
22:12 n4x joined #minetest-dev
22:13 sapier could you post a link so we can se what you're talking about?
22:17 ShadowNinja sapier: Basically reverting this, but leaving the longer range: https://github.com/minetest/minetest_game/com​mit/e547d279ab91be3249144fc2551fb73955ac9f25
22:17 ShadowNinja sapier: Punching now breaks the node instantl, making punches impossible.
22:17 ShadowNinja +y
22:17 Miner_48er Shadow any need for punching?
22:18 Miner_48er in creative?
22:18 Akien joined #minetest-dev
22:18 ShadowNinja Miner_48er: Yes, eg, reloading travelnets, setting WorldEdit coordinates, setting schematic probabilities.
22:19 sapier is it possible to reduce time to "next to 0" without breaking punch?
22:23 ShadowNinja sapier: About as low as you can put it is 0.3, and we want to avoid accidentaly breaking because you clicked too long.
22:24 sapier as it was that way in 0.4.7 I think it's better to preserve old behaviour than add buggy new one
22:38 VanessaE guys, I just decided to give 0.3.3 a try, compiled from sources.  there's something dreadfully wrong with the 0.4.x series.  stuff that would cause 0.4.x to choke (fps-wise) is super fast here (as in 45-40 fps with a view range of 300)
22:38 VanessaE I thought maybe I was imagining it or remembering it wrong.  I wasn't.
22:38 sapier did 0.3 have any lua support?
22:39 VanessaE nope, of course not.  but I'm talking fps, and thus client-side.
22:39 VanessaE (I decided to compile it to check out something entirely unrelated, and just happened to notice how FUCKING FAST it is)
22:39 sapier I guess client side with server on remote pc?
22:39 VanessaE client+server on the same box.
22:39 VanessaE same conditions under which I normally use 0.4.x
22:39 sapier shaders enabled or disabled?
22:40 VanessaE well it's 0.3.3 so no shaders.  0.4.x can't keep up with this even in its most minimal state.
22:41 sapier ok we should create a match as close as possible ... default 0.4 configuration isn't comparable to 0.3
22:41 VanessaE anyways /me wanders back to what she was doing previously.
22:45 sapier where did you get 0.3.3 from?
22:46 VanessaE direct from the minetest git repo, as usual.
22:46 celeron55 there are many things that can contribute to that
22:46 celeron55 for example the fact that 0.3.3 has efficient and thorough use of a texture atlas
22:47 VanessaE right
22:47 sapier yes but that's not possible with hd textures ... or at least way more difficult ;-)
22:48 VanessaE I'm not comparing with HD textures.
22:48 sapier doesn't make a difference
22:49 sapier if you put big fat wheels to a car to make it capable to run in desert you'll have the weels if you drive on roads too ;-)
22:49 celeron55 what
22:50 sapier removing texture atlas to improve hd textures does affect small textures too ;-)
22:50 sapier just tried to speak in pictures
22:50 celeron55 well what the fuck? that's the point
22:51 celeron55 that's why it can be one of the reasons for VanessaE's results, and that's why it matters
22:52 sapier I didn't want to tell it doesn't matter sorry if I didn't tell that clear enough
22:52 celeron55 VanessaE: just for kicks or maybe for testing, try copying the 0.3 world and loading it in 0.4 8)
22:52 VanessaE I can agree with that, but I have to wonder if it's just that, or something else?  I don't recall seeing much of a difference between enabling and disabling it, at the time.
22:52 celeron55 it's probably much more than that
22:52 VanessaE sure, I'll do that shortly.
22:52 celeron55 it's just what came to mind
22:53 kaeza I have such a screenshot
22:53 kaeza It's... interesting
22:53 VanessaE the difference betwene 0.3 and 0.4, for me, is easily around 6x or so.
22:54 sapier yes but personal feeling isn't quite reliable if environment has changed that much if we really want to know we need to create as similar conditions as possible
22:54 celeron55 kaeza: if it's old enough, there will be some pretty weird stuff going on :P
22:55 kaeza (it's actually from 0.3.1, but I don't think there were major changes in the mapgen)
22:56 Exio4 http://dpaste.com/1441996/ wut?
22:56 Exio4 did git checkout 0.3 and git clean -fd before compiling
22:57 kaeza celeron55, here's one https://dl.dropboxusercontent.com/u​/100008207/screenshot_444767519.jpg
22:57 kaeza and another one: https://dl.dropboxusercontent.com/u​/100008207/screenshot_690068789.jpg
22:59 sapier how do I start a 0.3 world with current minetest?
23:00 celeron55 Exio4: it won't compile with irrlicht 1.8
23:00 Exio4 ahm
23:00 kaeza sapier, copy ~/.minetest/world to ~/.minetest/worlds?
23:01 sapier doesn't work
23:01 sapier 00:04:51: ERROR[main]: Subgame [] could not be found.
23:01 celeron55 well someone has broken compatibility then
23:01 ShadowNinja nore has a horribly convoluted way of generating trees with the VM...
23:02 kaeza ah, yes... I had to add that line to world.mt
23:02 kaeza gameid = xxx
23:02 sapier ok forcing gameid works too
23:03 celeron55 (it used to work so that it would assume minetest_game if it wasn't defined; and really should still do that but apparently doesn't)
23:04 sapier not sure but could be I don't have minetest_game in that tree
23:10 sapier I don't understand what your're all talking about using same world same settings I have 5 FPS more in 0.4
23:10 sapier same view in world of course too
23:11 sapier ok not 5 more more exactly same
23:13 * ShadowNinja gets frustrated at WorldEdit's schematic creation functions and looks for a hex editor.
23:14 sapier at least for opengl 0.3.2 and 0.4.7-dev perform almost exactly same to me (phenomII x4/gtx260)
23:20 sapier http://imgur.com/bfMsr3r,1ProJQP#1 screenshots
23:20 sapier may be I haven't been in exact same position for those yet the screenshot with more fps seems to have slightly more complex scene
23:22 sapier does anyone have an idea why 0.4.7 replaces shore sand with dirt with grass? never realized that before :-)
23:23 sapier left #minetest-dev
23:28 ShadowNinja There's probably docs somewhere, but it's kind of fun to guess the MTSM format by looking at it. ;-)
23:35 zat ShadowNinja: I am breaking my head trying to finish the new rollback, for your info.
23:35 zat I just smoked though, and I feel unable to keep coding for today.
23:35 zat Smoking makes me all dizzy.
23:38 ShadowNinja Drat, zlib deflated. :-(
23:39 ShadowNinja zat: Ok, I hope you can finish it soon. :-)
23:40 zat kik

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