Time Nick Message 02:33 * blaise hides quietly in the corner to observe 02:37 iqualfragile xyz, sapier, ShadowNinja: we could use socialist milionaire protocoll for password authentification 02:39 iqualfragile both client and server have the clients passwordhash, so they could just test if its the same 02:39 iqualfragile an attacker would then have to intercept the message the client uses to create a new account/ change his password to get it, which is a lot less likely 02:45 iqualfragile (it allso makes shure the server is not fake, but i do not think that actually matters) 02:45 iqualfragile *sure 09:15 sapier iqualfragile according to wikipedia socialist milionaire requires an additional "key" p. If we make that one to be hardcoded in minetest, it's useless. If it's generated on first connect (what I understand is what you suggested), you can't ever connect using a different client. 11:07 sapier john_minetest: did you do additional tests to 1090 yet? 11:11 VanessaE sapier: for my servers, your 1090 pull, as it existed about 8 hours ago, just went online. Do I need to update? 11:12 VanessaE that is to say, I'm at commit 9a12056dacafb8be67e95cbfcad203d234b89ae5 right now. 11:13 VanessaE meh, I hate merge commits too. make that 1f1a4c9 11:15 VanessaE celeron55: about http://paste.ubuntu.com/6698208/ ... is there any point to actually applying those last couple of lines? I'm a little confused as to whether you'd decided it helped or not. 11:31 sapier tha last commit only affects low bandwidth connections 11:32 sapier I recommend not to add celerons changes as they cause a lot of cpu load and will render test of networking useless 11:33 VanessaE ok 11:33 VanessaE I will leave my servers' configs at their normal state and just work off of what I've got from #1090 11:33 ShadowBot https://github.com/minetest/minetest/issues/1090 -- Network fixes udp 1 by sapier 11:33 VanessaE shaddup, bot :) 11:33 VanessaE (this is 1090 applied on top of a clean clone, btw) 11:34 sapier but of course, once we know networking works it's worth a try according to celeron those settings will make minetest faster (if your server is fast enough) 11:34 VanessaE john_minetest: rrdtool 11:35 VanessaE http://digitalaudioconcepts.com/vanessa/hobbies/minetest/stats.html 11:35 VanessaE in fact, I *JUST* put that online an hour ago, as you can see 11:35 VanessaE (those big spikes are from the nightly backup script) 11:35 sapier we tried 1090 yesterday on celerons adsl connection, with the additional fix applied it was slightly faster on downloading then master, but master didn't perform as bad as it did for you either 11:36 sapier without it id did overload the adsl connection ;-) 11:36 VanessaE heh 11:36 VanessaE well you *do* want to be able to flood it out 11:36 VanessaE but only if it's legit traffic and not tons and tons of resends/bad packets 11:37 sapier yes but you have to adjust that setting if your connection is extremely slow 11:37 VanessaE john_minetest: --> #minetest 11:38 sapier meaning if there's no chance to send 1024*512 bytes in a single shot 11:39 sapier 512*554 ... I always forget overhead 11:39 sapier and change wrong number 1024*554 11:44 VanessaE sapier: well in some near future, maybe you'll figure out some kind of auto-ranging feature. 11:59 VanessaE not right now, busy with other stuff 12:04 VanessaE turn off noclip. 12:04 VanessaE they need time to generate, but that ^^^ will stop that behavior. 12:04 VanessaE (they'll still need time to generate though) 12:06 VanessaE sapier: potential problem... and this may have already been there and I didn't see it before: my servers are randomly disappearing from the master server list. 12:07 VanessaE they'll be there on one refresh cycle, then on the next cycle one of them will be gone from the list, then it'll be a different one that's gone, etc. 12:07 VanessaE john_minetest: because of things like floatlands 12:07 VanessaE and there are mapgen mods that might put stuff way up high 12:07 proller john_minetest, indev have floating islands 12:07 VanessaE otherwise what's the use of a map that goes to +31k? 12:08 proller and FM have no unloaded problem ;) 12:08 proller VanessaE, Road server have ~20000 mountain 12:09 VanessaE proller: case in point. 12:14 VanessaE Is there some limit to how many servers can show up on the public list? 12:26 VanessaE ah, figured out the server list issue. two servers fighting for the same port (a screwup when I migrated to the new host). 12:26 VanessaE so I *still* don't have the announce problem that ShadowNinja was having. 12:27 VanessaE for once I DON'T have a bug! :D 12:27 VanessaE john_minetest: what stopped? 12:28 proller VanessaE, announce problem already solved 12:29 VanessaE proller: oh was it? what was the problem? 12:29 proller too long get ?query= 12:29 proller now it via post 12:30 VanessaE ah 12:30 VanessaE odd though that my servers probably issue the longest such GET commands and they worked. 12:31 VanessaE oh ok. 12:31 proller it was strange floating problem only for ShadowNinja 12:38 sapier VanessaE I don't se a way to determin maximum server bandwidth automaticaly, I can and do this for each individual peer but that's something completely different 12:39 VanessaE sapier: well I meant for some time down the road. 12:40 sapier maybe but as this is supposed to be a interim solution ... 12:40 VanessaE right 12:40 VanessaE meanwhile: 12:41 VanessaE I am connected to one of my servers, with your 1090 patch, using a vanilla mt client (at current HEAD) and it's working perfectly fine so far 12:41 VanessaE just routine dig/build stuff, nothing wrong at present but theres no other-user activity right now 12:42 sapier that's good, if no critical errors occur until tonight (utc) I'm gonna merge it 12:42 VanessaE I would give it at least a couple of days before merging, if it can wait that long 12:43 VanessaE (owing to the need to rebase, etc) 12:45 sapier didn't you already try it for days? 12:46 VanessaE sort of yes 12:46 sapier 1090 is just some refactoring 12:46 VanessaE I mean, in the form of your separate fork 12:46 VanessaE but not in the forum of this pull + merged against master 12:46 VanessaE I'm paranoid like that :) 12:47 sapier of course it's not 100% there wasn't a error on refactoring but it's dev not stable branch ;-) 12:47 VanessaE did you fix that one exception that kept popping up? 12:47 sapier yes 12:48 VanessaE ok good 12:49 sapier maybe other core devs could write their opinion too, I'd not have a problem waiting some days too 12:54 VanessaE getting some reports now 12:54 VanessaE three users online plus myself, seem to have no trouble playing. 12:54 sapier what trouble? 12:55 VanessaE no, no trouble. 12:55 VanessaE one user reports losing some mese, but that was yesterday and may be a player error, not a server error. 12:55 sapier oh :-= NO :-) 12:56 sapier guess the word "trouble" is conotated that bad I don't see the "no" on first reading it :-) 12:57 VanessaE reading comprehension FAIL :) 13:02 VanessaE One user reports that they're using "4.9" (presumably 0.4.9-release) and it works fine 13:20 VanessaE Still seeing a fair amount of sign-OFF lag though 13:20 VanessaE (but I always exit-to-OS, so maybe it's the client crash issue and I just don't notice, as such) 13:50 sapier VanessaE that might be result of fixing singleplayer shutdown hang, if client doesn't manage to get disco out within half a second it's not sent at all ... causing peer to timeout at server 14:08 sapier enet jitter calculation seems to be totally broken or is rounded to an extent beeing useless 14:09 sapier to be correct not jitter calculation but rtt calculation 14:31 Exio4 the feature i would love is the "send-view-distance-to-server", it would be good for people with not-that-good networks 14:34 ShadowNinja proller: I'm about 99% sure that your semi-POST commit doesn''t fix my issue. But I'll test it. 15:36 iqualfragile sapier: the key is the hashed password 15:39 iqualfragile btw: my server is still running on sapiers branch 16:45 sapier iqualfragile I did understand this different for what I read it you need a key p AND hashes to verify 16:46 sapier but I could be wrong about it 16:46 iqualfragile you are :D 16:46 iqualfragile while there is another key in use that just gets generated 16:46 sapier so I'm right 16:47 sapier that one generated is the one I'm talking about 18:19 iqualfragile sapier: nah, thats allright, will still work 18:19 sapier I don't understand how is this supposed to work? 18:20 iqualfragile the socialist milionarie protocoll checks two values for equality 18:20 iqualfragile without giving any information about their actual value 18:20 sapier yes I assume this is the hash 18:20 sapier but to be able to check it it requires that key "p" 18:20 iqualfragile so if both the client and server have the same value (pwhash in this case) they will successfully compare it 18:20 sapier that both of them need to know 18:22 iqualfragile sapier: https://otr.cypherpunks.ca/Protocol-v3-4.0.0.html better description 18:23 sapier that's otr wich requires to accept a common secret first 18:23 iqualfragile search for socialist milin 18:23 iqualfragile *milion 18:26 xyz please 18:26 xyz if you're going to harden security or use crypto in any way 18:26 xyz do use libraries, don't write your own crypto 18:26 proller lets use openssl with clients certs 18:27 proller and usb tokens 18:28 sapier ok iqualfragile p doesn't have to be secret so I misinterpreted this point, but I agree with xyz imho we don't have any chance to implement crypto in a save way on our own 18:28 xyz correct 18:29 sapier but I disagree that using libraries in general will help, security is only as good as library itself, and there are plenty of broken libs out there but just a few proofen ones 18:29 sapier so decision of what library to use is as important as using a library at all 18:30 xyz just use openssl 18:31 sapier openssl is one of the most proven libs for sure, gnupg (used by curl) might be another option 18:32 sapier yet adding encrypted communication will increase requirements for local password storing too ;-) 18:33 sapier I'm not gonna implement secure communications but I won't stop anyone from doing it, and I will assist where possible of course 18:36 iqualfragile libotr contains smp irrc 18:36 iqualfragile iirc 18:38 sapier I'm not sure if otr is right thing to use, does anyone how it performs for large data transfers? 18:39 xyz sapier: so what do you suggest for local password storage? 18:40 sapier right now I don't suggest any change but we have to talk about it once secure communication is most likely to be merged 18:42 celeron55 this is the go-to solution for cross-platform up-to-date "don't write your own crypto" things: https://github.com/jedisct1/libsodium 18:42 xyz other apps just store password in plaintext and no one ever had problems with that 18:44 celeron55 (using openssl is quite DIY these days) 18:44 sapier "no one intends to build a wall" ... "no one is watching american citizens" ... "no one has to fear anything" ... 18:44 celeron55 (and if someone wants to use openssl, use gnupg instead because of licenses) 18:46 sapier and storing passwords in cleartext wasn't even state of art yesterday ... of course that doesn't stop people from using stone age softwar 18:47 xyz yeah well, your point? 18:48 xyz storing them in a keychain? 18:48 xyz and ask for a password to this keychain which kinda defeats the purpose 18:49 sapier that's the problem that solution would only help if someone regularly connects to different servers 18:50 sapier but why do people compain about having to enter a password for minetest they have to do this for any web service 18:51 xyz no 18:51 xyz for web people can just use browser's password save 18:52 xyz games typically save passwords, don't they? 18:52 xyz people are used to this behavior 18:52 sapier I don't know as all online games I know require some kind of "I herby sell my sole to (google/valve/apple/microsoft/....) 18:54 celeron55 well just generate a keypair on the client and use public key crypto to get in like ssh 8) 18:54 sapier not storing passwords in browser internal password storage is one of number one suggerstions to improve security ... everyone recommends to use a password safe which is locked by a master password 18:54 xyz yet people still do that 18:55 sapier yes just using ssl to encrypt connection would probably most easy and reliable way to do 18:55 sapier because most ppl just can't understand what they're doing, that's not a excuse to those knowing 18:57 xyz hmm... 18:58 xyz sapier: your irc client stores passwords in plain text, also nice read https://developer.pidgin.im/wiki/PlainTextPasswords 18:58 sapier that's why I use a nonsense password for irc 18:59 sapier irc isn't encrypted anyway 19:00 xyz that's not true 19:00 ShadowNinja It has the option via SSL/TLS. 19:00 sapier the only reason why I'm registred at all is because I lot of people kept asking me to do so, I still se no reason for doing it except of avoiding to be asked to do so twice a day 19:01 xyz my point: no need to implement any security around storing passwords/hashes (like keychains, etc); if you're willing to do so then I won't object anyway since it's your time, just don't forget to add an option to turn password-protected storage off 19:02 sapier basicaly this irc thing is same as minetests current security state, with the small difference that you gamers usualle know even less about security then irc users 19:03 sapier as long as we don't have encrypted communication you're absolutely right about no need to encrypt hashes 19:04 ShadowNinja Actually, IRC is fairly secure, in fact I know of a network that requires TLS with SHA-256, and a number of other things to keep everything securely encrypted. 19:06 ShadowNinja Minetest needs some work. In fact I beleive SHA-1 was depreciated a few years ago... 19:06 sapier yes 19:07 sapier because it was supposed to be broken in a few years by that time, as far as I know this hasn't happend by now ... but who knows what nsa knows 19:08 ShadowNinja Well, better to be overly secure than inadequately so. 19:09 sapier the only reason why I demand non cleartext passwords beeing saved right now is to protect users from their own ignorance ... I assume they will use passwords the better not should use for minetest 19:09 xyz sapier: nah, my point should be applied to both encrypted and unencrypted communications 19:11 ShadowNinja We should also be using a long, random hash that isn't the username. 19:11 sapier sorry don't understand, "your point" is disabling it? 19:11 celeron55 ShadowNinja: it's not deprecated; it's deprecated for certain uses (like password hashing) 19:12 celeron55 sha1 and even md5 and even md4 are fine for eg. checksumming files 19:12 sfan5 how about we use crc32 for passwords? /s 19:12 celeron55 sfan5: lol 8) 19:12 sfan5 It is especially hard to brute force crc32 19:13 ShadowNinja celeron55: Well, what a coincidence, that's what we're using it for. ;-) 19:14 sfan5 john_minetest: want a script that turns auth.txt into something understood by jtr 19:15 sfan5 there is a salt too.. 19:15 Exio4 is it $username-?? or $username?? 19:16 Exio4 with ?? being the pwd 19:16 xyz sapier: no, my point is that "no need to implement any security around storing passwords/hashes (like keychains, etc); if you're willing to do so then I won't object anyway since it's your time, just don't forget to add an option to turn password-protected storage off" 19:16 xyz i didn't say anything about encrypted/unencrypted network 19:17 sapier I consider storing passwords something completely different to storing hashes, maybe that's why I don't understand 19:18 sfan5 Exio4: sha1(${username}${password}) IIRC 19:18 ShadowNinja Exio4: https://github.com/minetest/minetest/blob/master/src/util/string.cpp#L61 19:18 xyz sapier: well, i think both are fine 20:42 xiong What code base controls the public server list? 20:43 xiong Do I raise an issue against 'minetest' itself? 20:43 proller wat? 20:44 xiong A public server list is displayed in the client main menu. This is also available, in more detail, on a web page. Where should I raise an issue against the list itself -- not the rendering of it? 20:46 celeron55 it's in the main minetest repo, util/masterserver 20:47 celeron55 with BlockMen about 3rd person view: http://paste.ubuntu.com/6717038/ 20:47 xiong Excellent. How do, celeron55? 20:47 celeron55 we probably need to whip up a preliminary lua api for setting those 20:47 celeron55 but not implement it yet because lazy (and lua and networking stuff is too hard to implement anyway) 20:48 celeron55 xiong: so the issue belongs in minetest's issue tracker 20:48 xiong Yes, thanks. How are you doing, celeron55? 20:48 celeron55 the thing that provides the list is called the masterserver (doh) and it's hosted by thexyz (so hosting issues go to him) 20:49 celeron55 xiong: i'm finnish so i don't like that question! but according to your culture, i should say "fine, thanks" 20:49 xiong What do Finns say? 20:50 celeron55 they grumble something funny 20:50 xiong Well we do that too. Some people, if you say "Good Morning!" they grumble, "What's good about it!?" 20:51 xiong I meant to ask what sort of greeting you do like. 20:52 celeron55 that was the answer to that question 20:53 xiong Well, in any case, Greetings. 20:54 xiong https://github.com/minetest/minetest/issues/1092 20:56 celeron55 oh and, proller, who said "wat?", is the one who develops the masterserver 20:56 proller we cant add all random flags to list 20:56 proller "angry admin" 20:56 proller "griefers disalloewed" 20:57 celeron55 i'd guess just prefixing the server name with "TEST: " would be enough 20:57 celeron55 ...as proller already commented on the issue (slow me) 20:59 proller now you can see at "uptime" and "age" of server map - it it low (less than hour) - server 99% temporary 20:59 celeron55 anyway about this: http://paste.ubuntu.com/6717038/ 21:00 celeron55 or, really, nothing in particular about that; but there's that and it seems to largely work while not messing with a terrible amount of stuff 21:00 celeron55 and it seems to be wanted by a number of people 21:00 celeron55 so i will merge some variant of it unless someone objects or wants to develop it more 21:01 celeron55 (and that is 3rd person view, for the logs) 21:01 xiong Testing may go on a long time; live servers may go down for backup every day. The concept of a testing server is a personal view that must be expressed; it cannot be inferred. 21:02 celeron55 yes; name and description are for those personal views that cannot be inferred 21:03 xiong There may be a practical limit to the number of flags displayed in the list; however an identical argument can be made against any such flag: C, D, P, etc. 21:03 celeron55 in fact, most of the flags could be inferred by joining the server 21:03 xiong The point of a flag is to indicate to a player whether he might want to attempt to join at all. 21:04 xiong Whether the server is testing or live is a critical point. I mandate no metadata in identifiers; that's the Hungarian bump route. 21:04 celeron55 but the meaning of "testing" is very fuzzy too 21:04 celeron55 for someone it might mean something completely different than for someone else 21:04 xiong Everything is fuzzy. 21:04 celeron55 flags are gotten from actual server configuration 21:04 celeron55 so they aren't fuzzy 21:05 xiong Why debate this so hard? It's an obviously good idea. If you don't like it, I can't force it. 21:05 xiong You don't gain anything additional by "proving" me wrong. 21:05 proller maybe i want "dev" "RC" "stable" flags to my server.. 21:05 BlockMen celeron55, some variant of it? 21:06 celeron55 yes i don't; but as it seems proller doesn't see the need for it and i don't particularly care, i'm trying to see if you have some argument for it that would make sense for everyone 21:06 xiong Yes, proller; that's would be okay too -- except the additional complexity now needs to be explained. 'T' is well understood. 21:06 celeron55 BlockMen: assuming someone develops it more 21:06 xiong celeron55, it's like a lot of things: blatantly obvious. 21:07 celeron55 let's just wait for someone else to come up and comment 21:07 xiong If you think you can set the flag programmatically, great. Accumulate uptime data and set a bar. Therefore all servers join the list with T set; and that flag is only cleared after crossing that performance standard. 21:08 BlockMen ic. in general i will continue developing it, but ofc other can take care of it too 21:08 xiong I've come to notice that the majority of servers in the public list are unplayable for one reason or another. I've also come to believe that many are not intended to be played. I know I run a testing server and I tell players upfront to expect nothing. 21:09 proller maybe UNSTABLE flag if uptime less ~30 minutes and age less ~2 days 21:09 celeron55 why is it in the list then? 21:09 celeron55 maybe you have misunderstood the purpose of the list and maybe it shouldn't even contain those testing servers 21:09 celeron55 (i don't know the purpose of the list) 21:10 sapier celeron that mesh fix floods console 21:11 celeron55 sapier: what is "that mesh fix" 21:11 sapier now on activation of each entity a "Loaded mesh: .... " is show 21:12 celeron55 i know 21:12 celeron55 also it re-loads them every time without caching a single one; there's a comment in the code about that with a suggestion of improvement 21:12 xiong proller, servers do go down for daily backup -- just restarting the server regularly seems to improve matters, deal with leaked memory and so forth. So it's asking too much to say that the server should be running continuously for X hours. 21:12 sapier I haven't explicitly checked but I guess this is it https://github.com/minetest/minetest/commit/d76957ee22c27adab89cee551e3ab1c85d8717cc 21:14 xiong However, a given address *and* port -- since multiple servers run from one address -- will accumulate some % of uptime. I'd say, roughly, 95% uptime with restart intervals averaging 6 hours or more constitutes stability. 21:15 xiong Still, please provide a setting that operators can use to override and declare a testing server regardless of uptime. I ran a testing server for a week without a restart and wiped it without regret or notice. 21:16 xiong Note that the client does not display enough characters in each item to permit verbose descriptions, which are only available after user has selected. 21:16 proller why tou cant place TESTING: to name? 21:17 xiong Because then that's the entire displayed name. 21:17 celeron55 do enough users understand the flags in any case? 21:17 celeron55 a flag is not some magic wand of understanding 21:17 xiong C clearly is creative; P is, I imagine, PvP. I don't know what D means. 21:17 xiong No; nothing is magic. :( 21:18 xiong Might be nice to explain the flags in the client, too. 21:19 celeron55 ...well, they're probably going to be replaced by images once someone manages to get that functionality in the gui system 21:19 Calinou D = Dedicated 21:19 Calinou err no 21:19 Calinou it's Damage enabled 21:19 celeron55 even i don't know what those flags mean, lol 21:19 Calinou * is password required 21:20 Calinou "Ded" stands for dedicated, it's a flag that doesn't appear in-game, only on the web list 21:20 Calinou it appears if the server is not run from a client 21:20 Calinou "liq" is finite liquid, also doesn't appear in-game 21:20 Calinou "rol" is rollback, I guess 21:21 kaeza "password required" flag is kinda useless, as you can check that by simply trying to join 21:21 celeron55 kaeza: you can check everything by trying to join 21:21 Exio4 you can know if it is creative by joining too 21:21 Exio4 damn 21:21 celeron55 we don't need even names, just join and see 8D 21:21 Exio4 ^ 21:21 kaeza celeron55, you can't know if it's survival or creative before actually joining 21:21 kaeza but meh 21:22 celeron55 anyway, i'm not against the flag; if more people would like it, tell that to proller 21:23 celeron55 (this you should be doing without me requesting it) 21:23 kaeza I'm not against it, just giving an opinion 21:24 kaeza (I almost never play online anyway because crappy 3G) 21:24 xiong It's possible to argue that it's pointless to describe any server in the public list because players don't care and join randomly, wreak havoc, and move on. 21:24 proller [far future] in game flags must be show as images with hover tooltip with long description 21:25 xyz I think it's already possible with kahrl's formspec table 21:25 xyz at least images are 21:26 sapier guess we should merge the formspec table soon 21:32 celeron55 what's stopping it from being merged? 21:33 sapier as far as I know the only reason is no one did it by now 21:42 ShadowNinja Are nodeboxes still "EXPERIMENTAL"? I think not... 21:44 ShadowNinja I'll fix this in a minute: https://github.com/minetest/minetest/blob/master/doc/lua_api.txt#L2013 <--- missing "wielditem" 21:46 ShadowNinja The only other thing marked experimental is the HUD API, I guess that's unlikely to drasticaly change at this point too, although maybe not. 22:19 ShadowNinja I'll push #1084 in a while too. 22:19 ShadowBot https://github.com/minetest/minetest/issues/1084 -- Add InvRef::get/set_lists() by ShadowNinja 22:21 ShadowNinja hmmmm: Can you look at #1061? 22:21 ShadowBot ShadowNinja: Error: ProcessTimeoutError: Process #230 (for String.re) aborted due to timeout. 22:22 ShadowNinja https://github.com/minetest/minetest/issues/1061 -- Crash with no error message with `minetest.place_schematic` and table schematic specifier. 22:23 ShadowNinja kahrl: Is this ready? https://github.com/minetest/minetest/pull/920 22:27 ShadowNinja I don't like how this means that players have to be kept loaded: https://github.com/minetest/minetest/pull/804 22:33 ShadowNinja This seems good, any comments? https://github.com/minetest/minetest/pull/706 22:39 sapier for most other formspec elements scale factors are multiples of inventory item size is this different in hud? 22:43 ShadowNinja GCC warnings: http://pastebin.ubuntu.com/6717607/ clang has a whole bunch of them: https://travis-ci.org/minetest/minetest/jobs/16617091 22:43 ShadowNinja sapier: The HUD uses percentage values and pixels. 22:44 sapier hmm someone musthave been sleeping when this was added :-( 22:44 ShadowNinja sapier: The formspec or the HUD? :-) 22:46 sapier as formspec was there before hud should've been implemented consistent ... but it's to late so we need to ensure consistency within each somehow separate thing ... lets hope noone ever wants to use formspecs as hud elements 22:47 ShadowNinja sapier: No, the formspec system doesn't scale to the HUD at all, it's essentially pixel values multiplied by some unknown number. 22:47 sapier the "unknown" number is calculated as fraction of current window size ... same as inventory item sizes 22:48 sapier that's why formspecs grow if you increase window 22:48 sapier except of mainmenu I added a fixed size mod for this 22:50 ShadowNinja That's hacky. 22:50 sapier yes but pilzadam insisted on mainmenu having fixed size 22:51 ShadowNinja Just use percentage values. It would be nice if the HUD was just a 100% width formspec with a clear background that didn't prevent playing. 22:51 sapier you could use this from in game too if you had a parameter in size field 22:51 sapier sorry but I think we're talking about different things 22:52 ShadowNinja +pixel values for things that need to be absolute. 22:52 sapier as you tell hud already does use pixel values so it has to use them in future too 22:53 sapier while formspec doesn't so it can't switch to pixel anytime soon too 22:54 emptty Hello there 22:54 ShadowNinja Well, formspecs badly need a redesign. The inventory item scale thing can be changed then. 22:55 emptty Can the recent versions of minetest be built without luajit ? 22:55 sapier you know my opinion about formspecs, bad but not as bad as requireing immediate action 22:56 emptty I have a bug against the debian package because luajit is not available on some of our architectures: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=734461 22:56 sapier as far as I know lua is still default 23:16 ShadowNinja emptty: Yes, but only by removing the system lib or modifying CMakeLists.txt. 23:18 emptty ShadowNinja: I'm currently trying to compile on an host where lua-dev is installed but not luajit(-dev), and cmake didn't complain 23:18 emptty I'll soon tell you whether it compiles till the end, and whether it works afterward 23:19 ShadowNinja emptty: Lua 5.1 is bundled. 23:19 emptty not in the debian package. It's prunned 23:19 ShadowNinja Well, that's good. 23:20 emptty we use the system libs as much as possible. We are a distro ;) 23:20 ShadowNinja But apparently the lib location varies by distro, so we can't do that. 23:20 emptty sure 23:22 ShadowNinja What happened to the travis integration in pull requests? It still builds pulls, but the results don't show up. 23:22 ShadowNinja xyz: Do you know anything about ^? 23:29 emptty of course I had to patch a bit the CMakeList.txt to build with the system lua when luajit is not available, but it seems under control