Minetest logo

IRC log for #minetest-dev, 2014-03-28

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

All times shown according to UTC.

Time Nick Message
00:01 tomreyn joined #minetest-dev
00:18 BrandonReese joined #minetest-dev
00:44 VargaD joined #minetest-dev
00:48 salamanderrake joined #minetest-dev
01:05 MichaelRpdx joined #minetest-dev
01:09 celeron55 iqualfragile_: minetest operates like a retarded brick in that sense, but it does it on the server side which is usually not noticeable :-D
01:09 iqualfragile_ important point is: it operates.
01:14 VanessaE hah, quote of the day, right there.
01:18 domtron joined #minetest-dev
02:41 salamanderrake_ joined #minetest-dev
03:36 ShadowBot joined #minetest-dev
03:52 VanessaE [03-27 23:52] <VE-Minetest> <est31> so when I capture the hash I can gain access?
03:53 VanessaE re:  sniffing a user's password as he/she logs into a server
03:54 VanessaE maybe this is a good time to think about doing some basic ssl or some other encryption of the login part of the protocol
03:54 VanessaE I know sapier would love the idea, anyway :)
04:01 VanessaE [03-28 00:01] <xxxxxxxxx> Well, I managed to log into a password-protected server using a captured hash.
04:02 VanessaE [03-28 00:02] <xxxxxxxx> xxxxx: It's a local server (my survival world, to be exact).
04:02 VanessaE [03-28 00:02] <xxxxxxxxx> I used Wireshark to capture the hash (would work for any computer on the same LAN (basically, in the same house)).
04:03 VanessaE I think that's enough.
04:13 us`0gb Hmm. Assuming a compromised node somewhere along the route ... It wouldn't be limited to LANs.
04:17 est31 joined #minetest-dev
04:17 cheapie Hi, est31!
04:17 cheapie This is where the devs usually are (but they all vanished, for whatever reason).
04:18 cheapie ...sapier? PilzAdam? celeron55? Anybody home?
04:50 kaeza joined #minetest-dev
05:23 werwerwer joined #minetest-dev
05:34 smoke_fumus|2 joined #minetest-dev
05:50 salamanderrake joined #minetest-dev
06:07 kahrl joined #minetest-dev
06:12 alexxs joined #minetest-dev
06:52 grrk-bzzt joined #minetest-dev
07:20 darkrose joined #minetest-dev
07:20 darkrose joined #minetest-dev
07:27 rsiska joined #minetest-dev
07:43 Robby joined #minetest-dev
07:54 celeron55 VanessaE: that's well known; minetest doesn't even try to be secure; it just hides the password from plain sight
07:54 VanessaE just posting that as a reminder
07:54 VanessaE seems that folks are beginning to actually consider doing that sorta stuff now is all
08:08 celeron55 well, i guess that if someone comes up with a good way to do it, it might be done
08:23 est31 left #minetest-dev
08:25 est31_ joined #minetest-dev
08:29 est31_ Hi for login you could do SCRAM-SHA1
08:30 est31_ and on password reset the client should set the first half of the salt for the stored_key and the server the second half.
08:30 ImQ009 joined #minetest-dev
08:31 VanessaE better to bring that specific idea up next time sapier is around.
08:31 est31_ then you even can store the login safely, as it don't expose the password
08:32 est31_ when will sapier be online?
08:32 VanessaE I'd expect sometime in the next 6 hours, roughly.
09:01 kaeza joined #minetest-dev
09:17 est31 joined #minetest-dev
09:26 celeron55 est31: any suggestions for lightweight and portable and permissibly licensed (LGPLv2 minimum) libraries?
09:27 est31 2+3 easy
09:27 est31 1pls wait
09:39 est31 GNU SASL library.
09:40 est31 no experience with it
09:40 est31 but has 1+2+3
09:40 VanessaE celeron55: hah, SASL....   *gets bag of pork rinds*
09:41 est31 why that?
09:45 VanessaE because as I recall, c55 had a rather sour experience with SASL vs some IRC server(s) about a year or so ago :)
09:47 est31 :)
09:48 est31 ok don't need to be all sasl  algs just SCRAM-SHA1-PLUS
09:54 celeron55 SASL isn't wrong; but it's wrong that freenode banned my ISP and required me to use SASL
09:55 celeron55 that has nothing to do with the technology itself
09:55 est31 ok
09:56 VanessaE gah, way to ruin a joke :P
10:04 n joined #minetest-dev
11:04 kahrl joined #minetest-dev
11:48 PilzAdam joined #minetest-dev
11:52 Exio4 shouldn't be there a "dev" branch with all the work (the network protocol, and so on)?
11:54 PilzAdam Exio4, yes, its called "master"
11:56 celeron55 master is the unstable branch, stable is the stable branch; simple
11:57 celeron55 then if some very incompatible stuff is developed, a feature branch or a minor version branch can be used
11:57 celeron55 (or major)
11:58 celeron55 (but the major version has always been 0)
11:58 iqualfragile joined #minetest-dev
12:06 kaeza joined #minetest-dev
12:15 PilzAdam joined #minetest-dev
12:23 ImQ009 joined #minetest-dev
12:40 Shardvex joined #minetest-dev
12:50 PenguinDad joined #minetest-dev
12:55 hmmmm joined #minetest-dev
13:32 lanxu mitä itse käytät?
13:32 lanxu whoops. sorry :)
13:50 alexxs joined #minetest-dev
14:25 Shardvex joined #minetest-dev
14:30 tomreyn joined #minetest-dev
15:49 PenguinDad joined #minetest-dev
16:08 Jordach joined #minetest-dev
16:10 Zeitgeist_ joined #minetest-dev
16:16 rsiska joined #minetest-dev
16:30 spillz joined #minetest-dev
17:12 grrk-bzzt joined #minetest-dev
17:12 rubenwardy joined #minetest-dev
17:27 sapier joined #minetest-dev
17:54 proller joined #minetest-dev
18:07 est31 sapier can you read the logs here
18:09 PenguinDad I wonder why minetest doesn't do a "VACUUM" statement before closing the SQLite DB?
18:10 sapier done
18:10 sapier what's in there?
18:11 est31 i suggested how to improve the login to a MT server
18:11 est31 vanessaE said i should probably speak you
18:12 est31 the hash you use is static and can be subject to replay attacks
18:12 sapier yes as celeron55 said a well known issue ... if you have suggestions how to improve it you're welcome
18:13 est31 You may want SCRAM-SHA1 for login. It's most state of the art.
18:14 sapier do you know a free available implementation of it?
18:14 CiaranG It would still reveal the password when you did a 'change password'.
18:14 est31 GNU SASL library
18:15 est31 it depends
18:15 est31 SCRAM only specifies the login not the change password
18:15 sapier is GNU SASL available on windows too?
18:15 est31 its c89 no posix dependency
18:15 Calinou joined #minetest-dev
18:16 est31 ciaran you could make the client send the stored_key at 'change password'
18:16 celeron55 i glanced at it and it seemed fairly good
18:16 celeron55 so maybe someone could take an attempt at it
18:16 est31 the client only has to make sure that the salt is not used by some other server too
18:16 sapier as we're about to do major changes anyway we could do this too
18:17 est31 ... that can be achieved by the client randomly choosing a part of the salt
18:18 est31 (because if it is used by some other server that one can impersonate to the other)
18:18 sapier I didn't read the specs by now but whatever we do a password musn't be client specific ;-)
18:19 est31 how do you mean that?
18:19 sapier no data stored on client may be involved
18:20 sapier next question protect login only or do full encrypted communication?
18:21 celeron55 sasl doesn't care about encryption (except that it kind of recommends it on top)
18:21 sapier I'm for protecting password in first step only
18:21 celeron55 as far as i understand
18:21 est31 sasl doesnt care.
18:22 est31 but: scram-*-plus has channel binding
18:22 est31 that can be used to authenticate a servers certificate without CA.
18:23 est31 so that it is easy to set up encryption for people running servers
18:24 celeron55 do you happen to know a good library for encryption then? i guess gnutls is one
18:24 est31 sry idk
18:24 sapier I'm not thinking about servers and pc clients right now but about android clients
18:25 sapier without hardware encryption support I don't think android clients will be able to handle the additional encryption load
18:25 est31 how much traffic
18:25 celeron55 but encryption isn't needed for now
18:25 celeron55 also, the traffic is very small and android uses HTTPS connections all the time just fine
18:26 celeron55 it can't be a problem even if it's done some day
18:27 celeron55 the only request for encryption that i have seen has been by a chinese guy for the chat
18:27 CiaranG Important stuff. The last thing you'd want is a malicious network sniffer seeing that you'd dug the node at 132,56,216
18:27 sapier you greatly underestimate traffic and performance impact of encryption ... it doesn't matter if a webpage everyone expects to be shown after 20 seconds takes 21 seconds but if our packets take 100 instead of 10 ms we've got a problem
18:27 CiaranG Just encrypting chat might make sense
18:28 sapier all or nothing ciaranG
18:28 celeron55 well, in certain countries it certainly makes sense
18:28 proller minetest - game with <1000 active users, GAME, not bank
18:28 CiaranG sapier: all or nothing because?
18:29 sapier because it's crap to add encryption in 50 locations instead of 1
18:29 sapier if we do encryption it should be done at lower protocol level
18:29 VanessaE how difficult is it to detect what hardware support there is on a platform and, if suitable support is found, to use it (and if not, stick with the plain protocol)?
18:30 celeron55 well, encryption would just require adding a new packet in the protocol that encrypts the stuff it encrypts
18:30 CiaranG Seems more crap to encrypt 50 things that don't need encrypting instead of 1
18:30 proller all issues except chat security already fixed ?
18:30 celeron55 and then it's really quite trivial to just do that for chat packets
18:30 celeron55 but i agree with proller that it's unnecessary
18:30 sapier no it's not do you wanna authenticate any single packet?
18:31 CiaranG You would simply encrypt the payload of chat message packets
18:31 celeron55 no?
18:31 celeron55 it's certainly possible to only require authenticatiton for chat packets
18:31 sapier either you create a encrypted communication then you can send all data through it or not then encryption is somehow useless
18:31 celeron55 -t
18:32 sapier do you wanna ask user for his password on each chat message? ;-)
18:32 CiaranG !?
18:32 sapier encryption without authentication is useless as you're prone to man in middle attacks
18:33 celeron55 what the fuck now
18:33 celeron55 you can think of the chat as a separate encrypted connection if you want
18:33 sapier if we wanna do encryption we need to create a session on login and keep it alive for as long a s a user is logged in
18:33 * proller faceplam
18:33 celeron55 sapier: yes, so?
18:34 celeron55 how does doing it only for the chat make it impossible? in no way whatsoever
18:34 CiaranG You just need the client and server to negotiate a key on login, and use that for the rest of the session
18:34 celeron55 let's stop this useless discussion anyway
18:34 sapier so chat is encrypted but stealing inventory is not a issue?
18:34 proller dont forget check client signed certificates for chatting in game
18:34 celeron55 yes, that was the idea
18:34 celeron55 you can't really end up in jail for moving items in minetest
18:35 celeron55 but you can by chatting about politics
18:35 VanessaE proller: you jest, but as I understand a number of other games use exactly that method for logging in.  signed certs.
18:35 sapier well of course this can be done but I thought we're a game not a chat client for noth korea
18:35 proller VanessaE, this games use real money?
18:35 VanessaE proller: not all.
18:36 sapier Imho protecting the users password is best we can do right now encryption ... nice to have but for chat only isn't sane to me
18:37 VanessaE hell I use a signed cert to authenticate to IRC/services, but both c55's and sapier's points stand.
18:39 sapier crypto is always about making it more expensive to break data but it's useless to upgrade a stable wooden front door to a reinforced steel door if backdor is paper
18:40 celeron55 there's no backdoor to listening to minetest chat if the client and server refuse to transfer unencrypted unauthenticated chat messages
18:41 celeron55 well except if those endpoints run other software or addons that have weaknesses, but that always applies
18:42 cheapie joined #minetest-dev
18:43 celeron55 it can cause some weird situations for the user if someone does a man-in-the-middle attack for the unencrypted data and the chat stays encrypted and going to the right place though
18:43 celeron55 "i can't read the chat, write it on this sign" lol
18:43 CiaranG I know. Remove the chat packets and encode the messages as movements of different sized stacks of cobblestone in and out of the inventory. 41 cobblestone=A, 42 cobblestone=B, etc.
18:44 CiaranG desert stone for lower case.
18:44 sapier I'm sorry but I don't understand why chat is more important then our game data
18:46 celeron55 i'm all in for encrypting everything if encryption is used though
18:47 sapier me too but I fear we can burry android client prior even finishing it if we do so
18:48 celeron55 lol, this encryption is not going to happen in a long time
18:48 VanessaE sapier, honestly I think the renderer is gonna be FAR worse for the android client's CPU than the encryption would be
18:48 celeron55 that's why i said a few screenfulls of text back that this shouldn't be discussed further
18:48 sapier btw sasl seems to be a replacement for our current low level protocol if we choose to use it
18:51 sapier what do you think about a more simple way to prevent replay attacks, server sends random salt to client, client hashes it's hash with salt and sends back to server .... or did I miss something?
18:56 celeron55 http://tools.ietf.org/html/rfc5802#section-5
18:56 celeron55 SCRAM-SHA1 is quite like that
18:56 celeron55 somewhat more complicated for whatever reason
18:56 kahrl a MITM could obtain a one-time access code at least
18:56 kahrl (but since the protocol isn't encrypted anyway, it doesn't matter that much)
18:57 sapier yes this way doesn't prevent mitm attacks
18:57 sapier it's only safe against replay
19:02 sapier I guess I can add the simple variant I suggested within a couple of hours, but as it's just a first thought this might be completely useless if I missed some way to break it
19:03 kahrl well the first rule of crypto is don't roll your own
19:04 kahrl although in this case, any crypto would be strictly an improvement
19:07 kahrl another practical question, regardless of the new auth mechanism: how do you intend to send the initial password hash if the user is connecting to a server for the first time?
19:10 e1z0 joined #minetest-dev
19:11 est31 sapier kahrls comment is right: no experimenting
19:12 est31 for example, SCRAM also does PBKDF2 several (as in 4096) times
19:12 est31 hash functions were invented to be fast not to be slow. bad you need the opposite for passwords
19:14 BrandonReese joined #minetest-dev
19:15 est31 kahrl that password hash (storedKey) needs to be sent in plain text (or via encryption channel).
19:18 est31 important thing is that the server does never get the pw itself, and the salt is not equal to the one from another server
19:18 celeron55 it can be a simple hash to begin with so that it never goes in completely plain text
19:18 est31 heard of rainbow tables?
19:19 kahrl est31: a problem would be that any server could claim that the player is new and needs to send the storedKey, and be able to launch replay attacks afterwards
19:19 kahrl but this is MITM territory again
19:19 sapier it's either experiment or not do it anytime soon
19:19 celeron55 that's what they can do always anyway unless you authenticate them, and even then if you inadvertedly trust them
19:20 celeron55 :P
19:20 kahrl oh, I missed the thing about the salt
19:20 sapier if we wanna have a secure solution we need to use certificates and tls
19:21 sapier everything else is (more or less) unsafe
19:22 celeron55 does there exist tls over udp for unreliable data? 8)
19:22 sapier nope
19:22 kahrl one time pad is safe too :D
19:23 sapier kahrl I take this as volonteering to manually distribute millions of pads to each user ?
19:23 sapier ;-)
19:23 kahrl sure, we could sell OTP key books
19:23 kahrl every client-server pair needs one!
19:24 sapier :-) lets get serious again, what we can do quite quick is some custom thing. this isn't secure in more precise way but it'll make spoofing harder
19:24 est31 ... i think tls over udp can be reached with DTLS ...
19:24 celeron55 oh so it exists
19:25 celeron55 that's cool 8)
19:25 celeron55 it's even implemented by gnutls and others
19:25 cheapie This reminds me of something: <est31> ... i think tls over udp can be reached with DTLS ...
19:25 cheapie Why does MT even use UDP in the first place?
19:25 est31 what?
19:26 cheapie Did the authors of TCP kill celeron55's dad or something?
19:26 celeron55 it actually uses UDP because i wanted to try making a protocol based on UDP, and then you guys voluntarily started to use the end result
19:26 sapier because you don't need to retransmit a clients position of 10 seconds ago if you already sent the new one
19:26 celeron55 but it allows sending positions with not causing extra lag in packet loss
19:27 sapier same for entties and other moving things
19:27 cheapie OK, so what about things like inventory moves?
19:27 sapier we've got a reliability support uppon udp
19:28 sapier yes all of it already exists in other protocols
19:28 sapier no need to discuss this again ... we're already trying to find another protocol but by now everyone s quick in suggesting new ones but once evaluation starts same ppl beeing most loud ones demanding a change suddenly are very quiet
19:30 sapier obviously just now another requirement to our new protocol was added ... authentication and encryption support
19:31 tomasbrod joined #minetest-dev
19:32 sapier yet ... can someone please test the android port? I'd prefere to complete one task prior starting a new (huge) one
19:34 proller whats is new ?
19:34 tomasbrod If you dont mind me being noob, sapier, I would test it.
19:35 sapier noobs are welcome as they're not blind to the obvious things like developers usually are
19:35 e1z0 joined #minetest-dev
19:35 sapier http://animalsmod.comuf.com/​downloads/Minetest-debug.apk this is to test
19:36 tomasbrod Ok, see you later.
19:37 domtron joined #minetest-dev
20:00 iqualfragile_ joined #minetest-dev
20:31 Guest94087 joined #minetest-dev
20:37 spillz joined #minetest-dev
20:38 NakedFury joined #minetest-dev
20:39 spillz sapier: apk issues on galaxy note 2 (some of these will be familiar)... 1. very difficult to place blocks/items (threshold issue again?) 2. missing inventory textures for many blocks
20:39 sapier did you try to change the threshold in settings?
20:40 spillz digging works fine, only placement. is there a separate threshold for each?
20:41 sapier no it's just doubleclick for placing
20:41 sapier doubletab
20:41 sapier wait ...
20:41 spillz 3. Top row icons screen placement is not aligned with touch (have to touch below where they appear)
20:41 sapier maybe the doubletab detection still uses a hardcoded value
20:42 sapier ok how do you make them appear?
20:44 sapier doubletabdetection is missing the threshold, I'm adding it
20:44 spillz What appear?
20:44 celeron55 why are the items at the bottom so hard to tap
20:44 sapier the top row icons?
20:44 celeron55 i've tried multiple times and usually nothing happens
20:44 celeron55 more often it places a block or does something else
20:45 sapier I don't know
20:45 celeron55 oh wait what
20:45 celeron55 it picks the top ones when touching the bottom ones
20:45 celeron55 and the bottom ones cannot be selected
20:45 spillz They appear but to activate them i have to click the bottom row items. I can't activate bottom row at all because presumably the touch area is off the screen
20:46 spillz sapier: I didn't realize that it was double click to place. now that works but is super laggy
20:46 sapier ok please describe as precise as possible as on my tablet it's just single row
20:46 celeron55 it's on two rows
20:46 sapier yes lag is a big issue I don't know how to fix it by now
20:46 celeron55 four on the top, four on the bottotm
20:46 celeron55 like an inventory grid
20:47 celeron55 -t
20:47 spillz Same as celeron55.... I have two rows of 4 inventory items
20:47 sapier yes I added this dependent on screen dpi
20:47 celeron55 well what's unclear then?
20:47 sapier but I could only test it on pc client with window size and there it works :-(
20:47 celeron55 tapping the bottom ones selects the one at the top row
20:47 celeron55 tapping the top ones does not modify selection
20:48 sapier ok I'm gonna try to find out why this happens :-)
20:48 celeron55 is the walking movement going to be improved sometime?
20:48 celeron55 currently you have to raise your finger from the screen to press another button
20:48 celeron55 it should work so that you can slide the finger onto another buttotn
20:48 celeron55 -t (fucking t)
20:49 celeron55 (how is that t even possible, this keyboard must be broken)
20:49 sapier for movement buttons this should work (as long as you stay at buttons)
20:49 sapier need to fix the bug when leaving the movement button area
20:49 celeron55 oh, apparently it does
20:49 celeron55 but usually i want to move from right to backwards
20:50 celeron55 and then it doesn't
20:50 celeron55 or, well... occasionally does, occasionally doesn't
20:50 sapier yes found out some minutes ago too ... in this case you temporary slide above a not button
20:51 celeron55 and then! the inventory cubes are completely broken
20:51 celeron55 i think spillz mentioned that too
20:51 sapier ok :-) some work to do
20:51 celeron55 i guess they could be simply disabled, MT can just use the top face with... umm.. is there a setting for that?
20:51 spillz 4. I still have that  awful startup lag (this phone is snappy for everything else) and maybe that explains the block placement lag (bad write speed = SQL performance)
20:52 sapier sorry spillz but it's already leveldb on android
20:53 spillz sapier: either way it's storage access speed affecting with db access?
20:53 sapier possible
20:53 celeron55 apparently there isn't a setting for that
20:55 spillz btw,  double click to place/use kinda sucks usability wise... twice the arthritis...
20:55 sapier better ideas?
20:55 spillz none that will make your life easier
20:56 sapier if they are better that's not a problem ;-P
20:57 celeron55 i'd ask for a turn sensitivity slider
20:57 celeron55 or whatever way to configure it
20:57 celeron55 it's way too fast for this 4.5" screen
20:57 sapier it's configured with mouse sensitivity
20:58 spillz You really should try out some of the competition... bottom line finger!=mouse. should be able to touch anywhere on screen to use/place active item
20:58 sapier what about adding a dropdown similar to threshold
20:58 celeron55 oh, random stuff: you can't quit the "change keys" menu
20:58 celeron55 and if you press back there, minetest quits completely
20:58 celeron55 do you even test anything yourself 8)
20:58 sapier oops
20:59 sapier yes but not any single feature ;-)
20:59 sapier and I didn't even ever use key change menu on pc ;-)
21:00 celeron55 oh wtf, double tapping a main menu tab will make minetest quit
21:00 celeron55 how is this possible
21:00 sapier double tab (outside of menu) to close a formspec
21:00 celeron55 oh that explains
21:00 spillz Would be nice to have gui controls for draw distance (and tie chunk loading distance to that too)
21:01 sapier wait it's getting a little bit much :-) can we add it to the issues gist post
21:02 sapier https://gist.github.com/sapier/9745980
21:02 spillz personally would rather play at 25 fps with decent draw distance than the <30 nodes at 40 to 50 fps the game defaults to now
21:02 celeron55 the movement really should be a joystick-style thing where you can switch directly from left to right and forward to backward
21:02 celeron55 currently when you switch from right to left, you move forward in the middle which is just crap
21:03 VanessaE sapier: toldya. :P
21:03 celeron55 and you should be able to rest your finger there without the player moving
21:03 celeron55 then you might be able to keep the device in your hand a bit more comfortably
21:03 celeron55 oh well, not sure of course
21:04 spillz sounds like celeron55 wants survival craft's default controls
21:04 celeron55 this might be the worst ide aever
21:04 celeron55 idea ever*
21:04 celeron55 spillz: well, they work; what can i say
21:04 sapier well now with double line hotbar the joystick may work but for lots of screens it's a huge waste of screen
21:04 celeron55 does someone *like* the current movement buttons?
21:05 sapier well at least I prefere them to the previous cross wich did use 1/3 of my screen
21:06 sapier it's hard to design controls usable at different screensizes, if you make em to small they're unusable at high dpi+ small screens
21:07 sapier hmm well ... as I now have dpi support this may be solvable
21:07 ShadowNinja The current buttons are O.K.  But I'd prefer a joystick-like thing too.
21:07 celeron55 i'm pretty sure placing should just happen with single tap
21:07 sapier but I need ppl with different screens to test as I don't have different devices
21:07 sapier how do you wanna dig celeron?
21:07 sapier and view?
21:08 ShadowNinja Ditto on what celeron55 said.  Hold to dig/look.
21:08 sapier hmm ok dig place decision could be time of pressing down
21:09 sapier but you're gonna place a node anytime you accidently touch the screen
21:09 spillz hold to dig, slide around anywhere to view. The move buttons don't bother me all that much. torn whether jump should be on left or right thumb
21:09 spillz Press once to placr
21:09 spillz hold to use items like carts
21:09 spillz tap mob to attack
21:10 spillz Press once to placr..  meant tap once to place
21:10 spillz Doesn't happen that often in my experience
21:10 spillz and I am pretty clumsy
21:10 sapier I already understood this but I'm not sure if this will not cause a lot of accidentally placed nodes
21:10 VanessaE why can't there be a button to explicitly place the wielded object and another to dig, any other touch on the screen is move/look/use?
21:10 celeron55 what if placing worked so that it places if you hold the right thumb on the screen and tap with the left?
21:11 celeron55 of course double tapping could be enabled at the same time for single-finger operation
21:11 sapier some sort of modifyer key
21:11 ShadowNinja celeron55: That's even more clumsy and confusing than double-tapping.
21:11 VanessaE why overload the screen gestures?  use a button damn it!
21:12 VanessaE stop overcomplicating it
21:12 sapier buttons are ugly
21:12 VanessaE so?
21:12 VanessaE the whole GAME is ugly if you get right down to it :P
21:12 VanessaE since when did that stop the gameplay from being good?
21:12 VanessaE usability trumps appearance.
21:12 sapier the button can't be where you place the nodes so you always have to move away from where you're building
21:12 ShadowNinja VanessaE: A button means that you need to use the cursor thing at the center.  And it's hard to use.
21:13 celeron55 well actually, i would like to place nodes to where the cursor is
21:13 celeron55 it's hard to place nodes when your thumb is larger than the position where you want it on the screen
21:13 sapier you can already do this celeron
21:13 celeron55 and you need to tap over the position
21:13 VanessaE placing/digging where the cursor is makes more sense to me but on a tablet that IS kinda hard to do
21:13 celeron55 but of course people would usually use this on tablets and not 4.5" phones so that doesn't apply there
21:13 sapier hope it works (touch targeting)
21:14 ShadowNinja I'd like to be able to place and dig anywhere on screen, turning to point at something is a lot harder than just looking in that general direction.
21:14 spillz I think minecraft pe gives the player an option. split controls that use conventional mouse cursor and buttons, and touch controls that let the user touch on screen where to place
21:14 sapier me too it's quite handy to place a line of nodes just by moving on screen
21:14 VanessaE spillz: +100000
21:15 sapier guys I already have that setting ... could you please try it ;-P
21:16 sapier "Touch free target" ... reenables crosshair and makes it behave same way as on pc ;-P
21:17 sapier ok not "Touch free target" enables the crosshair ... does anyone have an idea for a better name of that setting?
21:17 VanessaE sapier: "Aim to Place" or "Aim-and-place"
21:18 sapier i wouldn't understand that way of calling it too
21:18 VanessaE well you use a crosshairs in a shooting weapon to aid in aiming that weapon.
21:18 VanessaE but in this case you are doing it to aid in placing a block.
21:19 sapier What about just "Enable crosshair" ?
21:19 VanessaE naw
21:19 VanessaE "What's a crosshair"?
21:19 VanessaE etc.
21:19 VanessaE remember, you're dealing with noobs who barely know how to turn their tablets on :P
21:20 sapier we've got only 2-3 words so we can't write the whole story from the begining some billion years ago there ;-P
21:23 VanessaE my point is,
21:24 VanessaE describe WHAT it does
21:24 VanessaE not what it IS. :)
21:24 domtron_ joined #minetest-dev
21:24 sapier well it disables the crosshair ;-P
21:24 VanessaE smartass :P
21:25 _BrandonReese joined #minetest-dev
21:25 VanessaE think of it this way...
21:25 VanessaE you know those cheapo cameras you can buy for like $USD 20 or so?
21:25 sapier no
21:25 VanessaE you just aim then and click the shutter and that's it?
21:26 VanessaE you know, those disposable ones you buy at the petrol station before you go on a long trip?
21:26 sapier actually I don't care how that thing is called as long as there are at least 2 people who think the new name is good ;-)
21:26 VanessaE they're among the class of devices we call "Point and Shoot".
21:28 VanessaE at the risk of two bad puns in a row, my point is that you need to aim for terms that the average idiot is going to understand, because people don't like to experiment.
21:29 VanessaE people think they'll break stuff when they do.
21:29 VanessaE hell my mom used to think her computer, I shit you not, would get "confused" if you clicked around too fast.  and she was far from stupid.
21:30 sapier but "point and shoot" wouldn't be reasonable too
21:30 sapier at least to me
21:30 VanessaE "point and place" would work better :)
21:31 VanessaE but I wasn't suggesting "point and shoot" anyway
21:31 VanessaE just giving you an example of how we use such terms
21:32 sapier what about "Touch to place" as name for the non crosshair mode?
21:32 VanessaE yes
21:32 VanessaE that would be good I think
21:32 ShadowNinja You still touch in crosshair mode.
21:33 VanessaE tap to place?
21:33 sapier same
21:33 ShadowNinja "Point to place" or "Crosshair mode".
21:33 sapier you don't place by pointing at it too ;-P
21:33 sapier you still have to tap
21:33 VanessaE ShadowNinja: that's why I used "and" instead of "to".  a conjunction is needed if you want to use "point" here
21:34 VanessaE oh G*d, I sound like my grade school english teacher. :P
21:35 sapier I'm gonna fix the other bugs now hope someone found a commonly accepted naming till I've finished
21:35 celeron55 your teacher would be proud of you!
21:35 VanessaE celeron55: if my teachers saw how I butcher the english language in some other venues, they'd shoot me, in turn. :P
21:39 tomasbrod left #minetest-dev
21:43 proller joined #minetest-dev
21:45 spillz Instead of a phrase, having a screen for controls with a picture for each control type. Also for the Crosshair controls, maybe you want to have a look around area above below some dig/use/place buttons on the right thumb side
21:45 spillz *How about having...
21:47 sapier good idea but I'd consider this as not crucial for first android version
22:03 ShadowNinja Pushing in a minute: http://ix.io/bjr
22:05 sapier are you sure this is equivalent?
22:07 sapier and why do you rename member variables to that crazy non m_ notation noone can see it's a class member without ide support?
22:08 sapier ShadowNinja: imho this change is way to big for a 1 minute announcement, please create a pull request for review
22:10 domtron_ joined #minetest-dev
22:15 spillz joined #minetest-dev
22:17 sapier ShadowNinja:  how did getBlocksOnZ exit before?
22:17 ShadowNinja sapier: Hmmm?  This is the mapper, not the engine, so we don't have to be so strict about commits.  And I removed hungarian notation because the coding style recommends not using it.  And what do you mean by equivalent?  The main change is that LevelDB is 20+x faster.
22:17 sapier ahh by break
22:18 celeron55 this is for the new https://github.com/minetest/minetestmapper
22:18 celeron55 sfan5 and ShadowNinja are pretty much free to do what they want with it
22:18 sapier oh :-) ok how am I supposed to see this is mapper only ;-)
22:19 celeron55 by following #minetest and every IRC channel remotely related to minetest, i guess
22:19 celeron55 and forum and github and...
22:19 celeron55 i.e. in no way
22:20 sapier well as it's mapper only a pull request is obviously nonsense ;-)
22:20 ShadowNinja Hmmm, I guess it wasn't obvious.
22:20 ShadowNinja Minetest's database stuff is different, but you wouldn't know unless you've looked.
22:20 sapier no, especially as database code isn't the part of minetest I know very well ;-)
22:21 celeron55 i immediately knew because the filename had "db-" instead of minetest's "database-" :P
22:21 ShadowNinja Last commit gives it away though.
22:21 sapier you've written the db code celeron ;-)
22:21 celeron55 i haven't
22:21 celeron55 i have written none of it
22:21 sapier thought you've been at least involved in sqlite code?
22:21 celeron55 no
22:21 celeron55 it was contributed initially completely
22:22 celeron55 by a friend of the admin of the largest server at the time
22:22 sapier ok ok :-) doesn't change the fact I don't have any idea how db code of minetest looks like ;-)
22:22 sapier that part did work quite well by now
22:23 sapier I don't have any idea why that damn multiline hotbar behaves that crazy
22:25 domtron_ joined #minetest-dev
22:27 sapier wait ...
22:33 domtron_ joined #minetest-dev
22:37 domtron__ joined #minetest-dev
22:43 sapier celeron55 I think I found the double line hotbar issue but I can't test it
22:43 sapier I already uploaded the new version
22:44 Sokomine it might be a good idea to mark "lightwight" minetest servers in the public list somehow. the android client will be more successful connecting to one of these
22:47 sapier there's no way to decide this aitomaticaly
22:49 Sokomine hm. it might be possible to check how large the amount of media data the server offers is. might still be too troublesome. yet...with the normal server list, connecting to a public server will seldom work on the first attempt
22:49 spillz joined #minetest-dev
22:49 spillz sapier: bottom row now works correctly, top row not at all.
22:49 sapier hmm
22:49 sapier very strange
22:50 sapier so there are two bugs
22:51 sapier argh ... a very silly one
22:56 sapier spillz: uploaded new version
23:02 spillz working. now missing textures
23:02 sapier a new bug?
23:02 spillz Also all of the interactive elements are kind of big
23:03 sapier welll size of elements is quite a matter of personal preference
23:03 sapier and dpi and device
23:03 spillz Not new. I mentioned earlier that many inventory items don't have textures
23:03 sapier yes you did so the old bug ... I believe this is a ogles or irrlicht issue so I can't do very much about it atm
23:04 sapier but any suggestion is welcome
23:05 celeron55 just comment out itemdef.cppp:436-437
23:05 spillz Re size - maybe just because of my phablet. 5.5 inch screen 1280x720 or something. All buttons seem too big.
23:05 celeron55 or make irrlicht not report that it supports render target textures
23:06 celeron55 because they obviously are broken
23:06 sapier could this be device dependent like the npot2 issue?
23:06 celeron55 or add return NULL; to tile.cpp:830
23:07 celeron55 no, it's creating 64^2 textures
23:08 celeron55 or well maybe it is... but survivalcraft and whatever had it disabled too
23:08 celeron55 so they didn't find it working either
23:12 sapier hmm is there something simple to test?
23:13 Sokomine sapier: the 8 inventory slots at the bottom are now too big on my device (rather small phone). also the inventory cubes seem to stand on their head (compared to the normal client)
23:14 iqualfragile_ hmmmm: any new thoughts on mapgen?
23:14 sapier sokomine can you create a screenshot? best case add your screen size (inches) as text
23:15 sapier the size is finetuning the more examples I can see the better I can tune ... it's useless to make it good for one device if this is even worse for another
23:16 VanessaE I keep telling you sapier, you need to provide options to scale the UI
23:16 VanessaE this isn't 1985 anymore, we have the technology :P
23:16 sapier we CANT scale the UI when will people finally realize this
23:16 VanessaE um...
23:17 sapier we can scale some elements but not all
23:17 VanessaE ok, ninja'd
23:17 VanessaE I was about to say that formspecs DO scale with screen size
23:17 Sokomine *nod* perhaps you might offer a setting somewhere
23:17 sapier no they don't
23:17 VanessaE yes, they do
23:17 sapier images do scale but buttons don't completely scale
23:18 VanessaE is this a deficiency in minetest or in irrlicht?
23:18 sapier buttons have minimum size, texts as well as textboxes have fixed size
23:18 Sokomine hm, yes. thumbs don't scale well. a tablet may have the same resolution as a phone. afaik the physical screen size is available somewhere on android as well. at least there are apps who can see it
23:19 sapier btw andoid doesn't do ui scaling too as almost no other os does ... because scalable button and icon graphics are quite costly (svg)
23:19 Sokomine stupid question, but....how do i do a screnshot on android?
23:20 proller render menu to image and scale image ;_
23:20 sapier no it isn't Sokomine but dpi is (somehow) available it's 0.75 1 2 3 4 ... so it's a rough guess only
23:20 VanessaE proller: cheater.
23:20 VanessaE (that would work, however hacky it is)
23:20 sapier and would look like crap
23:21 VanessaE sapier: not if you did it right.
23:21 sapier no
23:21 VanessaE again, this isn't 1985 anymore.  we can scale images very nicely up or down
23:21 sapier you can't scale a 32 px rexture to a 256 button no matter what you do
23:21 VanessaE then scale from a larger size.
23:21 VanessaE but that's hacky
23:21 VanessaE surely there's a better way
23:22 sapier svg is the only way but it's slow
23:22 sapier and I don't know if irrlicht supports svg
23:22 spillz I think the standard way is to provide multiple sizes and pick accordingly
23:22 sapier exactly spillz
23:22 VanessaE sapier: it doesn't appear to support svg, no.
23:23 sapier then we have to find a way to decide what size to use on what screen ... wich is what I need screenshots for
23:23 Sokomine sapier: you solved the issue i had with text input lately very well. that works now convincingly
23:24 sapier what issue?
23:24 Sokomine sapier: only issue is when opening the chat. it would be really great to make that equivalent to f10 and show at least the last chat messages :-)
23:24 sapier hmm you should still see the last messages? the chat overlay should be transparent
23:26 Sokomine no, it isn't. it's completely white, showing me nothing of the game anymore. the one stu used was transparent afaik
23:26 sapier it's transparent for me too ... what android version/device do you have andhow is this displayed for others?
23:29 sapier is there any item that causes the broken textures?
23:30 sapier to me this issue seems to be quite similat to what we've seen on pc client about a year ago
23:30 Sokomine the textures work fine so far. the inventory cubes are inverted (they're seen from below and not from above as with the desktop version) - but that's a very minor issue
23:30 Sokomine hmm
23:31 sapier I see disorted inventory item textures every now and then but I haven't found a schme by now
23:32 sapier hmm I believe xyz had a fix for the inverted inventorycubes but I'm not sure about it

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