Minetest logo

IRC log for #minetest-dev, 2014-08-06

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

All times shown according to UTC.

Time Nick Message
00:03 ^v joined #minetest-dev
00:28 domtron_ joined #minetest-dev
00:47 RealBadAngel cg72, and squash it please
00:48 cg72 wtf does that mean???
00:50 cg72 RealBadAngel what do you mean???
00:59 cg72 RealBadAngel if you mean the img i got a smaller one up now! i hope thats what the hell it meant lol
01:01 RealBadAngel i mean squashing commits into one
01:01 RealBadAngel multiple commits per pull are not allowed
01:01 cg72 how do i do that?
01:02 RealBadAngel go to your git rep directory
01:02 cg72 on my pc or online?
01:02 RealBadAngel and do: git rebase -i HEAD~2
01:03 RealBadAngel then: git push -f
01:03 VanessaE on your PC.
01:03 RealBadAngel -i means interactive, follow on screen messages
01:04 RealBadAngel you have to pick name for your commit out of two
01:04 RealBadAngel save
01:04 RealBadAngel and do the same with comments
01:05 cg72 how do i do it with the comments?
01:06 RealBadAngel insert #
01:06 RealBadAngel for unwanted lines
01:06 cg72 ok
01:07 RealBadAngel process is pretty self explanatory
01:07 RealBadAngel just read the screen
01:07 cg72 ok now it wont push
01:07 cg72 i hate github
01:08 RealBadAngel hehe
01:08 RealBadAngel after git push -f
01:08 RealBadAngel you have to login
01:08 RealBadAngel it should ask you for user name and pass
01:09 cg72 i got that part and the  -f did it
01:09 RealBadAngel yes
01:09 cg72 it told me before i was behind on here lol
01:09 RealBadAngel now its just one commit
01:10 VanessaE looks right to me
01:10 cg72 thanks rba :D and sorry im still very new to github
01:10 VanessaE it's not just github, it's any git repo that'll do this
01:10 RealBadAngel cg72, np
01:10 cg72 this is the first git repo thing i ever got (github 5 months ago)
01:13 RealBadAngel cg72, now you need to have two votes of devs for the change to be merged
01:13 RealBadAngel you have mine
01:13 Exio4 two votes?
01:14 cg72 awwe thanks rba i really want to help out but im new to c++ on pcs, i did all embeded stuff before and its way different
01:14 RealBadAngel yes, at least two devs have to agree on the change before merging
01:15 RealBadAngel we are all humans, we do mistakes, such rule means at least two folks have to understand the code
01:16 RealBadAngel and even after that we are still hunting the bugs ;)
01:29 ShadowNinja ''tell sapier This is needed to maintain compatability when table-based formspecs are added.  And it doesn't matter if it's ten times as big, it is still easily managable, and if you're really concerned about size we can pass it through ZLib.
01:29 HLuaBot I'll tell that to "sapier" next time I see them around.
01:29 VanessaE haha
01:30 domtron joined #minetest-dev
01:37 RealBadAngel also please tell sapier that tables are the only solution
01:37 RealBadAngel its lua native and nothing else could be taken into account
01:37 RealBadAngel HOWGH
01:37 Zeno` additionally, please tell sapier that there is a character called "Sapier" in the book I just finished reading
01:39 RealBadAngel lol
01:40 RealBadAngel btw, wheres BlockMan?
01:41 RealBadAngel i want plants rotations to be merged
01:41 OldCoder joined #minetest-dev
01:42 RealBadAngel https://github.com/minetest/minetest_game/pull/290
01:42 RealBadAngel that waits for 2 weeks already
01:42 RealBadAngel as for such small change its just a bit too long
01:47 Zeno` easy fix, use math.random(0, 359) and then divide by two when setting param2 and then everybody is happy :3
01:47 RealBadAngel no, we cant
01:47 RealBadAngel param2 is 8bits only
01:47 RealBadAngel and we wont have any more bits
01:47 VanessaE so?
01:48 VanessaE divide by 2 he said.
01:48 Zeno` I mean.... param2 =  math.random(0, 359) / 2
01:48 VanessaE that gives a range 0-180.
01:48 VanessaE (well, 179.5)
01:48 RealBadAngel we have to pass that value
01:48 VanessaE divide by 2 in Lua first silly
01:48 Zeno` minetest.set_node(p, {name="default:papyrus", param2 = rot/2})
01:49 VanessaE like that ^^^^^^
01:49 RealBadAngel jeez, result is the same
01:49 Zeno` it makes it more cunning
01:49 Zeno` yes, the result is exactly the same, but maybe more self documenting?
01:49 Zeno` Not sure
01:49 RealBadAngel not really
01:50 RealBadAngel c++ side multiplying by 2 is way faster
01:50 Zeno` Possibly what SN is referring to is that 0..179 seems like a strange range; it's not obvious what that range represents
01:51 RealBadAngel thats lack of docs
01:51 Zeno` correct
01:51 RealBadAngel atm drawtypes are not docummented at all
01:51 RealBadAngel even nodeboxes are marked as experimental
01:52 Zeno` https://github.com/minetest/minetest/blo​b/master/src/content_mapblock.cpp#L1119 <--- possibly SN is suggest that that be fixed instead
01:52 Exio4 you can't make 360 integers fit into 8 bits without a lost of precission
01:52 RealBadAngel damn, you cant say or even see 1 deg
01:52 RealBadAngel 2 degs are more than enough
01:53 RealBadAngel make a screwdriver to change param2
01:53 Exio4 imo 0-179 (aka how it is right now) is k
01:53 RealBadAngel you will see that its smooth
01:53 RealBadAngel and thats not even the point of the change
01:53 Zeno` yeah it's ok
01:54 RealBadAngel its just about givin plants some variety
01:54 Zeno` "but it's wasting bits" lol
01:54 Zeno` </sarcasm>
01:54 RealBadAngel i asked for more bits but no go
01:54 Zeno` so why can't it be merged? I'd like to see it
01:55 domtron joined #minetest-dev
01:55 RealBadAngel 4 bytes we have and all is occupied
01:55 RealBadAngel because BM is not here?
01:55 Zeno` you're always one step ahead of me RBA
01:55 Zeno` sometimes 2!
01:56 Zeno` 2 degrees
01:57 RealBadAngel i hate having such small changes waitn in line while i have way bigger ones to merge
01:57 RealBadAngel i have coded 2 new drawtypes. new shaders
01:57 Zeno` make it larger then? Add lots of comments and redundant code. That way it looks impressive :3
01:58 RealBadAngel and im blocked with such foolish plants ;)
01:59 Zeno` I have a very important question... why does EAI_ADDRFAMILY appear to be undefined on linux?!
01:59 RealBadAngel wtf is that?
01:59 Zeno` it's even mentioned in the man pages
01:59 Zeno` socket stuff
01:59 RealBadAngel not familiar with that
02:00 Zeno` works fine on BSD... I'll come back to the issue later
02:00 RealBadAngel if it doesnt wave, scroll or flash or bounce
02:00 RealBadAngel thats not my stuff ;)
02:00 Zeno` I can make it bounce
02:01 Zeno` lol
02:02 Zeno` bouncing sockets... my great new thingy to add to the latest compositing window manager
02:02 Zeno` I'll enable it at the same time as my windows melt when they close; will be awesome
02:02 * VanessaE pours her junk box on a jumping:trampoline
02:03 VanessaE here, there should be some sockets bouncing around in that mess.
02:04 RealBadAngel rotfl
02:05 cg72 0.o
02:09 ^v joined #minetest-dev
02:13 Zeno` RealBadAngel, since you're bored: https://github.com/minetest/minetest/issues/1542
02:14 RealBadAngel i hate assertions
02:14 RealBadAngel why not code default values?
02:15 cg72 wow thats out of my normal stuff
02:15 Zeno` may as well do it at the same time as https://github.com/minetest/minetest/issues/1525 since there needs to be some mechanism added to return values from https://github.com/minetest/minetes​t/blob/master/src/treegen.cpp#L121 to the Lua environment anyway
02:15 * cg72 runs away
02:15 Exio4 assertions are nice to know shit is fucked up
02:15 Zeno` I do agree that people misuse assertions a lot though
02:16 Exio4 all the people abuses programming features...
02:16 RealBadAngel point is to not make a shit in your sandbox in the first place
02:16 Zeno` assert() should be used to check invariants, preconditions, postconditions, stuff like that
02:16 Exio4 you can see goto being abused too
02:16 Zeno` not as a replacement for setting proper defaults or checking function return values
02:16 Zeno` but anyway that's not the issue here I don't think
02:17 VanessaE Zeno`: #1542....wait what?
02:17 RealBadAngel its not the first time that assertions become a problem
02:17 VanessaE um, that's gonna crash EVERYTHING I RUN
02:17 Zeno` VanessaE, make a treedef but do not set a default value for random_level (i.e. leave it out of the Lua table)
02:17 VanessaE well no, that's not true
02:18 Zeno` ^^^ to reproduce the crash
02:18 VanessaE I guess I use random_level all over the place
02:18 VanessaE but of course it's useless because of #1469
02:19 Zeno` I've changed my mind on #1469... I think that having it default to "no seed" unless one is explicitly provided is the way to go now
02:19 Zeno` mainly for debugging convenience... i.e. it's easier to debug with a constant seed
02:19 VanessaE unless any of the random features are used.
02:19 VanessaE then at least a seed of 0 or something needs to be assumed
02:20 Zeno` oh they will be random, it's just that every tree will have the same random sequence
02:20 VanessaE then they're not random
02:20 Zeno` they're random for the individual tree :)
02:20 VanessaE not random enough for the random values to actually be useful in practice :P
02:20 Zeno` then when finished debugging you explicitly set a random seed (not a constant seed)
02:21 domtron joined #minetest-dev
02:22 VanessaE plugging in a random seed for each call is heavy.
02:22 Zeno` e.g. if I call rand() from C (without first calling srand(/*somerandomvaluemaybetime*/) then I will always get the same sequence of random numbers so things are, relatively, reproducible
02:22 VanessaE and makes the Lua side of the code more complicated than it needs to be
02:22 Zeno` yeah, so maybe a different approach needs to be taken... I dunno, maybe a flag
02:23 Zeno` the random seed will have to be set somewhere though, whether it's best in C++ or in Lua requires some thought I think
02:23 Zeno` probably C++, but I think there does need to be some way to get the "less random" behaviour as well
02:24 VanessaE I think it's obvious that if the call uses any of the functions that require a random seed, it should be called forth in C++
02:24 Zeno` probably, yes
02:24 VanessaE the less-random behavior is easy:  expressly set a seed of 0 in Lua.
02:25 VanessaE "debug mode" should be something you have to explicitly add into your code,
02:25 VanessaE so really, if the seed is not set, then just have C++ plug in the current time or some other lightweight value there.
02:25 Zeno` so... something like if (seed != 0) { /* set a random seed */ }   C++ side?
02:25 VanessaE anything that's always changing
02:26 VanessaE well, if seed != NULL
02:26 Zeno` well, the seed could even be pos :)
02:26 VanessaE (or nil or whatever it would be called once C++ gets ahold of it)
02:26 Zeno` it would be called 0 :D
02:26 VanessaE \0 then
02:27 VanessaE not 0
02:27 Zeno` O/
02:27 VanessaE seriously?  a seed with a decimal value of 0 is treated as no seed?
02:27 Zeno` no, 0 is a valid seed
02:28 VanessaE oh, just the way the math works out?
02:28 Zeno` yeah
02:28 Zeno` a seed of 0 will always give the random sequence of, say, 8 81 131 1331 11 444
02:28 Zeno` a seed of 1 will always give the random sequence of, say, 24 242 11 0
02:29 Zeno` and it's definitely 0, not \0 :p
02:29 Zeno` https://github.com/minetest/minet​est/blob/master/src/treegen.h#L52
02:29 cg72 seed = pr.next(0,10) :P
02:30 VanessaE well either way, without a seed specified from the lua side, the randomness stuff just. does. not. work.
02:30 Zeno` that's right
02:31 Zeno` what I'm saying is that since 0 is technically a valid seed, should spawn_ltree() instead treat it as a "seed not set" value
02:31 VanessaE yes
02:32 VanessaE it's not much trouble for me to fix this in my code, but the real question is should I be fucking with my code at all or should this be fixed in C++?
02:33 VanessaE and if I fix by changing my code, will I then run into #1532?
02:33 VanessaE er 1542
02:33 Zeno` I don't think so
02:33 Zeno` if you weren't already setting random_level you'd have already run into it :)
02:39 Wolfzie left #minetest-dev
02:44 cg72 left #minetest-dev
02:52 VanessaE bbl
02:56 Zeno` wait!
02:56 Zeno` how much later? :)
03:06 Zeno` VanessaE, https://github.com/minetest/minetest/pull/1543
03:06 Zeno` I umm think my editor tabs/spaces are fscked
03:06 Zeno` always something going wrong
03:12 Zeno` How do I delete the pull request?
03:12 Zeno` RealBadAngel, https://github.com/minetest/minetest/pull/1543   <--- am I able to delete that?
03:13 Zeno` somehow I still have a previous pull request in there
03:15 Zeno` nvm
03:24 Miner_48er joined #minetest-dev
03:33 cg72 joined #minetest-dev
05:07 RealBadAngel also, folks, its been a year since a monster of demoscene allowed us to us his music
05:07 RealBadAngel i wont just let it flow
05:08 RealBadAngel https://www.youtube.com/watch?v=cjHNPzv9sGY
05:08 Zeno` RealBadAngel, have you got SDL installed?
05:08 RealBadAngel going to to merge that as main menu title
05:09 cg72 joined #minetest-dev
05:10 Zeno` I gotta fix my text editor
05:10 Zeno` no idea why it wants to indent randomly :/
05:10 RealBadAngel use geany
05:10 cg72 sumlime text
05:10 cg72 ;)
05:11 cg72 sublime**
05:12 RealBadAngel if nobody wakes up in 1 hour Mercury Rain got merged
05:12 RealBadAngel and become mt main theme
05:12 Nate__ joined #minetest-dev
05:13 RealBadAngel fuck that, its too good to be left behind
05:13 Zeno` there is no reason for qtcreator to suddenly start doing this though
05:13 Zeno` I'll use vim and start again
05:13 cg72 good night RealBadAngel, Zeno`  o/
05:13 RealBadAngel geany...
05:13 cg72 ps vim = yuck
05:13 Zeno` I don't mind vim
05:14 Zeno` I only like qtcreator because it makes finding things more quickly while I am new to the codebase
05:26 RealBadAngel 20 minutes passed
05:27 * RealBadAngel is moving Merury Rain into main_menu.ogg now
05:38 Zeno` ShadowNinja, ok, pull request fixed
05:38 Zeno` sorry about the indentation
05:38 Zeno` I think I'll just stick to vim mode now
05:39 Zeno` https://github.com/minetes​t/minetest/pull/1543/files   <-- I'd normally do this differently, but I'm trying to follow the style of the rest of that file
05:40 Zeno` I'd probably use ternary operator truth be told
05:40 RealBadAngel 30 minutes
05:40 Hunterz joined #minetest-dev
05:41 RealBadAngel half way
05:42 Zeno` half way to where?
05:42 Zeno` mexico?
05:48 RealBadAngel to mt has music
05:49 Zeno` I dunno... I want the plasma lava
05:50 RealBadAngel too late, already done :_
05:54 Zeno` and nice water
05:55 Zeno` I want nice water as well
05:55 Zeno` and clouds
05:59 Zeno` ponies?
06:02 RealBadAngel lol
06:02 RealBadAngel complin the pull
06:02 RealBadAngel just to be sure
06:03 Zeno` which pull
06:03 RealBadAngel Mercury Rain
06:03 RealBadAngel the score that gonna become mt main title
06:04 Zeno` Is it a tracker style composition?
06:04 RealBadAngel yes
06:04 Zeno` I won't complain then
06:04 RealBadAngel made by Skaven/ Fairlight
06:05 Zeno` Fairlight are still around? :-o
06:05 RealBadAngel he gave us permission to use all of his tracks
06:05 Zeno` https://www.youtube.com/watch?v=cjHNPzv9sGY <--- that one? Use it!
06:06 RealBadAngel thats the one
06:07 Zeno` love it
06:07 RealBadAngel sorry i said Fairlight, hes a member of Future Crew
06:07 Zeno` yeah
06:08 Zeno` when I moved from Amiga to PC demo stuff 2nd reality was inspiration for a lot of my stuff
06:08 Zeno` err.... showing my age :(
06:08 RealBadAngel and i should say former member
06:08 RealBadAngel how old are you?
06:08 Zeno` 17
06:08 RealBadAngel :P
06:09 RealBadAngel and really?
06:09 Zeno` and really it's in your PM :P
06:09 RealBadAngel hehehe
06:09 RealBadAngel pretty close
06:09 RealBadAngel im just a yr older ;)
06:11 RealBadAngel ok, compiled, checked, good
06:11 RealBadAngel pushing
06:12 Zeno` I made a mistake.... as of july the 11th I am 1 year older than what I told you
06:12 Zeno` I don't really keep track of my age anymore lol
06:23 grrk-bzzt joined #minetest-dev
07:07 werwerwer joined #minetest-dev
07:11 Krock joined #minetest-dev
07:40 Nerotic joined #minetest-dev
08:14 PenguinDad joined #minetest-dev
08:29 casimir joined #minetest-dev
08:44 Calinou joined #minetest-dev
08:55 loggingbot_ joined #minetest-dev
08:55 Topic for #minetest-dev is now Minetest core development and maintenance. Chit-chat goes to #minetest. Consider this instead of /msg celeron55. http://irc.minetest.ru/minetest-dev/ http://dev.minetest.net/
08:57 proller joined #minetest-dev
09:33 proller joined #minetest-dev
09:48 ImQ009 joined #minetest-dev
10:39 proller joined #minetest-dev
10:56 Taoki[laptop] joined #minetest-dev
11:27 Taoki[laptop] joined #minetest-dev
11:48 proller joined #minetest-dev
11:49 PilzAdam joined #minetest-dev
11:50 Jordach joined #minetest-dev
12:07 Garmine joined #minetest-dev
12:10 CraigyDavi` joined #minetest-dev
13:28 Eater4 joined #minetest-dev
13:57 proller joined #minetest-dev
13:57 kaeza joined #minetest-dev
14:56 proller joined #minetest-dev
15:05 NakedFury joined #minetest-dev
15:32 cg72 joined #minetest-dev
15:44 Calinou joined #minetest-dev
15:49 Taoki[mobile] joined #minetest-dev
16:03 Hunterz joined #minetest-dev
16:21 zat joined #minetest-dev
16:38 sapier joined #minetest-dev
16:39 sapier ShadowNinja: who decided to add "table based formspecs" I will not agree to it for sure as they are same crap as the ones we have right now just 10 times as big
16:43 sapier there's only one (maybe) benefit of tables compared to declaration style they may be more changeable later ... but on the other hand you add two major issues 1) order of evaluation ... what's gonna be "on top" and 2) size
16:44 VanessaE sapier: what would you suggest then?
16:44 VanessaE current formspecs are just plain nasty
16:44 VanessaE you know as well as anyone
16:44 sapier and tables will be as nasty as formspecs too
16:44 sapier and you can't even read them
16:44 sapier even less then current formspecs
16:45 VanessaE so, what would you use?
16:45 sapier my suggestion is choosing a existing gui language ... and if you really want full features take one with a gui designer ... you'll never get a full featured gui description in a human readable way
16:45 VanessaE if you had to throw formspecs out completely and start from scratch?
16:46 sapier that's why formspecs are limited
16:46 grrk-bzzt joined #minetest-dev
16:47 sfan5 sapier: good luck finding a gui language that is not hated by everyone
16:48 sapier sfan5 that's exactly why formspec isn't as bad as ppl claim, it's easy for beginners, regular forms can be done quite simple and still there IS a way to do more complex things ... and yes it is limited ... that's the price to pay to not need a graphical designer
16:49 Calinou suggestion: in Configure menu, show pre-installed mods in a light blue rather than blue, which is hard to read.
16:50 sfan5 sapier: what is the problem with order of eval. with lua tables?
16:50 sapier and the form declaration is only half of gui as you need a api to handle signals and events too
16:51 sapier table elements are unsorted if they wouldn't be you loose the only benefit tables have in comparison to formspecs ... editability
16:52 sapier e.g. if you specify buttons in current formspec by definition they are drawn "in order"
16:53 sapier if you do this by table you either specify the draw order manual ... quite ugly, use automatic table indexing ... less ugly
16:53 sapier or don't specify draw order at all
16:53 sapier in first cases you need to reorder all elements to insert a button in between (same as now)
16:53 sapier in later case you can't assume any order of drawing
16:54 sapier buttons are a bad example ... replace button by image
16:55 sapier e.g. icons on top of a map
16:56 sapier Calinou good idea can you create a pull request?
16:59 sapier answering the question what I'd choose if I had to implement a full featured gui  ... I'd try to find a way to enable irrlicht to display html code ... implementing a html renderer on our own is not an option but maybe there is one we can use
17:00 sfan5 sapier: how about in-order by default, but allow a z-index to be set?
17:00 sapier and where's the benefit to current formspecs?
17:01 sapier we don't need a new gui language requireing same crapy syntax as old one
17:01 Calinou I can make an issue, but no time to make pull request :/
17:02 sfan5 sapier: I'm not talking about syntax
17:02 sapier and when I'm talking about "size" I don't mean "size in bytes" but "size on screen"
17:02 sfan5 just add width and height params
17:02 sfan5 or use default one if none specified
17:03 sapier I was talking about size of table style code compared to formspec style code ... different topic ShadowNinja mentioned
17:03 sapier back to order how do you mean z-index?
17:04 sapier like { draw_index=0 ... } or something different?
17:04 sfan5 z-index like in html
17:05 sapier isn't z-index css?
17:05 sfan5 right
17:05 sfan5 like in css them
17:05 sfan5 then*
17:06 sapier wouldn't change need for reordering z-indexes on adding items in between
17:07 sapier would be better then relying on implicit order for sure yes ... but adding more complexity too
17:07 sapier Am I the only one who thinks lua tables are way to bloated to show a gui?
17:08 Exio4 they're bloated but just lua code
17:09 Exio4 metaprogramming is probably easier with lua tables too! ;P
17:09 sfan5 uh
17:09 sfan5 lua tables are just a way of transporting data
17:09 sapier sorry but improved metaprogramming support at cost of easy entry in form design isn't actually what I'd consider progress
17:10 Exio4 that was a joke
17:10 sapier if we did switch to table there'd be no other way then hiding that bloated lua tables behind some sort of dom model ... we'd need that one anyway in order to get a better signal/event handling
17:10 sapier << away
17:13 sfan5 dom model
17:13 sfan5 why not just use xml then? /s
17:13 sfan5 (xml sucks)
18:12 Krock Was/is there already a feature request for resizing overlaid images to the base image's size?
18:12 Krock because mixing 16 and 64px can look very ugly
18:13 VanessaE such a request did exist yes
18:14 VanessaE RBA was working on it at one time
18:14 VanessaE bbl
18:14 Krock ah
18:14 GhostDoge Krock: but there should still be a way to overlay without resizing
18:15 Krock GhostDoge, reason?
18:18 GhostDoge hmmm… I don't know of any good reason right now
18:20 Krock :)
18:38 proller joined #minetest-dev
18:43 kahrl joined #minetest-dev
19:01 proller joined #minetest-dev
19:17 Taoki joined #minetest-dev
19:23 ShadowNinja The Lua formspec tables could be structured like DOM objects, but we don't need to generate those objects from a string like wuth XML.
19:24 ShadowNinja And yes, it's more code.  But it's also much more readable.  You could wrote all of Minetest on one line if you wanted too, and current formspecs are almost always one line.
19:26 Taoki joined #minetest-dev
19:38 sapier ShadowNinja: may I mention your yesterdays example? Imho it's a perfect proof of formspec tables not beeing more readable as generic attribute ... as for that example formspec is way more readable then table syntax
19:38 sapier but yes if you represent a table in table syntax it is more readable ...well guess that's not really a surprise ;-)
19:39 ShadowNinja sapier: No, the table us more readable.
19:40 sapier As I said I don't say table syntax doesn't have any benefit, I just say there's negative points to ... how you weight those aspects is highly subjective
19:40 ShadowNinja You don't know what's x/y/w/h without looking at docs for the first one.  The second is clear wuthout and extra documentation to memorize.
19:41 sapier sorry ShadowNinja but to me a 4 line compact syntax is way more readable then a 40 line syntax where I have to collect parameters of a single element from multiple lines
19:42 sapier or                                                                           do
19:42 sapier you                                  think                                                       this
19:42 sapier is more                           read                         able
19:42 sapier then this small sentence at the end
19:43 Krock or do you think this is more able to read
19:44 sapier of course this is a extreme example but some table style formspecs will look this way ... the more simple the more likely you end up in bloated ones
19:45 sapier I wouldn't have any issue in having a table as internal syntax (in case of beeing designed well and complete) ... but I'd hate to have to use it in mods
19:45 Krock My two pennies to this: tables eat up space, old one is fine
19:46 sapier and of course you'll have to find a good solution for the implicit ordering issue ... there are plenty of possible solutions true, but finding a good one really fitting all our needs is more complex ... most likely we'd need some iterations to find it
19:48 ShadowNinja sapier: You're welcome to find a better solution, but this is the best one so far.
19:48 sapier if this is the best one imho it's not worth to spend any time on it
19:49 sapier you'd have to rewrite code of about 3 months of full time development for almost no benefit
19:50 sapier don't tell me you don't have to do it, as if it's not supposed to replace formspec by some time it's even less worth working on it
20:03 pitriss About formspecs I agree with Krock.
20:05 sapier http://offscreengecko.sourceforge.net/ this might be a way to implement html based gui's ... at least license wise
20:05 sfan5 >html
20:05 sfan5 no
20:05 sfan5 as long as it's not strict xml, no
20:06 sapier better full featured html than another arbitrary complex and error prone own language
20:06 celeron55 i still think we should do our own in lua (menu and client side lua) which would use primitive draw calls implemented in the engine and which can be distributed as a very useful self-contained lua library too
20:07 celeron55 a middle ground between that and formspecs is hard to find
20:07 sfan5 sapier: html is bad
20:07 sfan5 html is half-broken xml
20:07 sapier imho if we replace formspecs we should do the full thing and not another half boiled one ... but to be honest I don't se a pressing need to replace formspec at al
20:07 sapier l
20:08 sapier well no matter if it's bad or not defining our own xml based language is insane
20:09 sfan5 that doesn't mean we should use a bad one
20:09 sapier defining our own most likely will end up in a even more bad one after some time
20:10 sapier still imho we should evolve formspec adding a api for signal/event handling and let the old relics die by some time
20:11 sapier benefit of this way of doing it would be each individual step is small and if some of them are wrong we don't loose a lot of work at once, thus reverting/dropping a change doesn't mean wasting months of developer time
20:12 sapier And I don't believe we're capable of designing even an acceptable gui language at first shot
20:12 sfan5 improving the current thing won't work
20:13 sapier did you look at the fstk thingy?
20:13 sfan5 no
20:13 celeron55 >I don't believe we're capable of designing even an acceptable gui language at first shot
20:13 celeron55 it should definitely be modeled by something that already exists and works well
20:13 celeron55 s/by/based on/
20:14 celeron55 (there are a ton of these kinds of things out there)
20:14 celeron55 i'm fine with iterating the existing one, IF there is a good way to do it
20:15 celeron55 i don't see any obvious way so maybe sapier could make a more detailed proposal? 8)
20:15 sapier sfan5: compare old mobf settings to new one using fstk, difference in code is about 100 lines (including comments) ... but new one is about 3 times of features
20:16 * sfan5 away
20:16 sapier well imho if we use something like the table thingy we need to put something similar to fstk on top of it to hide the bloated tables ... but same way we could hide formspec behind it
20:18 celeron55 what is fstk?
20:19 sapier the encapsulation used in mainmenu to consolidate the tab and dialog handling
20:20 sapier for next mobf version I did those changes required to make it available for in game too ... minor issues where left to link a form ot a player
20:20 sapier https://github.com/minetest/mine​test/blob/master/doc/fst_api.txt
20:20 sapier not much has changed since that definitition
20:23 khonkhortisan joined #minetest-dev
20:23 sapier fstk api doesn't really care about the format the formspec is told to core but of course it has to provide that format
20:23 sapier so while it's not visible in api it's present in implementation ... the user provided part of it
20:24 sapier mostly
20:26 celeron55 i do have to admit i don't know almost anything of this, but good luck anyway 8)
20:27 sapier well the main issue is we want a full featured all capable gui description language with simplicity of a one dimensional list :)
20:27 sapier quite hard to get that done :)
20:32 celeron55 well that's not going to happen
20:33 sapier as always we need to find some path in between
20:33 celeron55 either one uses a single dimensional list to get this kind of simple static positioning of elements or alternatively one uses a deeper structure in order to nest elements to get fancier layouting
20:35 celeron55 i still think QML is one of the best languages for writing GUI stuff; it's just not very intuitive to regular programmers
20:36 celeron55 also we can't really have that in MT due to technical reasons
20:36 celeron55 so, how do you choose the path?
20:37 celeron55 just evolve the existing one and hope that it some day finally gets to where it should be?
20:37 sapier as long as there's noone supporting that thing yes ... at least I don't have spare time to implement a qml parser ;-)
20:38 celeron55 well QML wasn't really a suggestion; anyway the only way to use it would be to use Qt in MT 8)
20:38 sapier especially as I don't know qml at all ;-)
20:39 celeron55 it's quite integrated to Qt's scripting and other systems
20:39 sapier that'd be a problem
20:39 celeron55 it even uses a custom javascript parser
20:40 sapier uargh
20:40 celeron55 that's how domain specific it is 8)
20:40 ImQ009 joined #minetest-dev
20:40 celeron55 to pick up the scripts from between the UI definitions
20:40 sapier why not add gecko engine and just display html ? ;-)
20:41 celeron55 html is the worst designed thing ever and gecko is probably larger than core Qt 8)
20:41 sapier true ... but we'd get css too ;-)
20:42 sapier and I thing we'll find way more html/css developers then qml developers
20:43 celeron55 i have heard so many bad comments from someone that tried to use the embeddable version of webkit in his game that i don't want that route
20:43 sapier still my prefered way would be improving formspec step by step as we get requirements from modders, right now I do only know one feature we can't handle
20:43 celeron55 most html/css developers are shit and can't actually do anything, especially in a limitedly debuggable environment like this would be
20:43 sapier font sizes
20:43 celeron55 now the same guy is making his own UI engine
20:44 sapier another ui enginge would be another option ... yet we'd need something we wouldn't have to maintain in about a year ourselfs
20:45 sapier I think formspec discussion is mixing up at least three things
20:45 sapier 1) how are ui elements drawn
20:45 sapier 2) how is ui defined
20:46 sapier and 3) how is user input handled and linked to ui definition
20:46 celeron55 yes
20:47 sapier gecko/webkit address 1) and 2) (maybe 3 too but I'm not sure about that)
20:47 celeron55 maybe designing this should be an issue on github which would list those subparts in the description
20:47 sapier fstk is most 3)
20:47 sapier and table formspec is 2 only
20:52 sapier celeron55 ... we actually have https://github.com/minetest/minetest/issues/1399
20:52 Calinou did we talk about libRocket?
20:52 Calinou this thing: http://librocket.com/ was suggested on GitHub
20:53 sapier please add it to that issue
20:53 Calinou no WebKit stuff involved, but you can use HTML/CSS
20:53 Calinou we should make a specific issue for potential replacements? this one is about re-designing formspecs
20:53 celeron55 hmm, i think it was librocket that the guy was using
20:53 sapier that's what we do
20:53 celeron55 that he ended up hating
20:53 Calinou if you use it well, it's pretty cool
20:54 celeron55 it doesn't use webkit? well whatever
20:54 Calinou no need to learn yet another language, big upside
20:54 Calinou I don't think it does.
20:54 Calinou not 100 % sure though.
20:54 celeron55 anyway, i've heard very bad stuff about it, be cautious
20:55 Calinou I've heard very bad stuff about anything, it's like programming languages…
20:55 sapier I assume we're just collecting options right now
20:55 celeron55 sapier: comment those three things in there, they don't seem to be clearly mentioned
20:58 sapier done
21:04 Jordach has anyone in here noticed fps drops while in the main inventory
21:04 Jordach other formspecs do not drop fps
21:05 sapier other formspecs with same amount of inventory slots too?
21:05 Jordach sapier, chests have more and do not
21:05 sapier that's interesting true
21:06 Jordach solid 64fps in any formspec except the inventory, which plays at 6fps
21:07 sapier 6?
21:07 sapier that's hard to believe
21:07 Jordach opening my inventory drops my fps to 6
21:07 Jordach sapier, the graphics on my tower are slowed by it
21:08 sapier hmm I see a drop too but only about 2 fps
21:08 nore_ joined #minetest-dev
21:08 Jordach sapier, this is on a multiplayer server
21:09 sapier and I have way more on my mobf menu, there it's 10 fps ... but that's a way more complex one too
21:09 Jordach Chest Formspec: https://cdn.mediacru.sh/CSj39es1bPbh.png
21:09 Jordach Inventory: https://cdn.mediacru.sh/ANg0g4yXFOL_.png
21:09 sapier wow
21:09 sapier what graphics card do you have?
21:10 Jordach sapier, AMD Radeon HD 8670D
21:10 Calinou I'll try
21:10 Jordach i can maintain 60fps with full shaders
21:10 Calinou creative or not, Jordach ?
21:10 sapier thanks calinou I was about to ask someone to test it
21:10 Jordach Calinou, noty
21:11 sapier does it happen on default game too jordach?
21:11 Jordach sapier, no
21:11 Jordach (untested, while my own game shows it well)
21:11 Calinou going from 240 to 190 in heavy scene, 600 to 340 in light scene (looking at sky)
21:11 sapier can you try to find out which mod causes the issue?
21:11 Calinou not too bad :)
21:11 Calinou preload item is off
21:11 Jordach sapier, using xgui
21:11 Calinou game is Carbone, default inventory, only slight tweaks to the look (hotbar highlight)
21:12 sapier I'd guess something in xgui is causing that issue
21:12 Jordach sapier, has not shown up in singleplayer until now
21:12 Jordach (and hasn't on multiplayer as well until now)
21:12 sapier can you try to pinpoint the commit?
21:12 Jordach sapier, it's in builds since july
21:13 Calinou https://cdn.mediacru.sh/mxVWQhuIB5cE.png
21:13 sapier so maybe a commit combined with xgui
21:13 Jordach okay, does not show with singleplayer in survival mode
21:13 Jordach it seems to be multiplayer specific
21:14 sapier there's not much multiplayer specific in formspec handling
21:14 Jordach sapier, solid 60 in singleplayer (advanced and singleplayer tab)
21:15 sapier same mods?
21:15 Jordach yup
21:15 sapier so you do one time start singleplayer next time multiplayer server?
21:16 nore_ joined #minetest-dev
21:16 sapier or do you connect to a server?
21:16 Jordach sapier, connect to a server (jordach.minetest.net : 30000) shows it for me
21:16 Jordach https://cdn.mediacru.sh/sp2W7iYlTeEe.png
21:17 sapier ok but you ARE 100% sure that server does run exactly same mods and same version of those mods?
21:17 Jordach yes
21:17 Jordach i havent changed the game mod wise
21:17 sapier ok
21:18 Jordach same performance drops even with it removed
21:18 Jordach (preload items)
21:18 sapier I try to connect to that server
21:18 Calinou I'll try server
21:18 sapier you know you have curl external server configured not having those testures?
21:19 Jordach curl is enabled
21:19 sapier yes but that server doesn't provide the textures
21:19 Jordach even with an empty inv, still occurs
21:19 sapier 23:18:00: ERROR[CurlFetchThread]: http://minetest.digitalaudioc​oncepts.com/bfdmedia/address = 0.0.0.0wool_dark_grey.png not found (HTTP response code said error) (response code 404)
21:19 Jordach sapier, index.mth is missing
21:19 Jordach ignore it
21:19 Jordach (VE generated this - not my issue)
21:19 sapier gonna take quite some time to get all of those textures
21:19 Calinou media doesn't seem to download
21:20 sapier they do calinou but extremely slow
21:20 sapier never ever seen them download that slow :-)
21:20 Jordach AHA!
21:20 Jordach i see what's going on
21:20 Jordach 60fps with a brand new account, the player file must be either slightly damaged for full of extra junk
21:21 sapier ok so I don't have to wait for media to complete till tomorrow? ;-)
21:21 Jordach sapier, ~26mb
21:21 sapier what's line speed of your server?
21:21 Calinou since the download bar is small, I can't see progress
21:22 Calinou Jordach, look at the screenshot I posted, my inventori ys full
21:22 Calinou sapier, the upload speed is always capped when downloading from server
21:22 Jordach restarted server with new player file
21:22 sapier for what reason? minetest is supposed to handle upload itself
21:23 sapier that's been one of the major improvements when I fixed the protocol ;-)
21:23 Jordach sapier, seems to after a while the player files under players seem to get clogged with extra data and useless junk
21:23 Calinou to not congest bandwidth, to not make all players lag
21:24 sapier that's been the thing fixed by my protocol fix
21:24 Jordach there might be an issue [22:24:19] <JD-BFD> <crazyginger72> jordach it drops 1 fps now for me
21:24 sapier we tested on different servers and didn't even get noticable lag on crapy dsl lines (if configured correct)
21:24 Calinou problem is, server with very good upload (eg 8 Mb/s, optic fiber) will send slowly
21:24 Calinou it won't end at eg. 2 Mb/s
21:25 Calinou which would be welcome
21:25 Calinou (UDP limit?)
21:25 sapier no
21:26 sapier max_packets_per_iteration is the parameter you have to adjust to your player and line speed
21:26 Jordach sapier, hang on
21:26 Jordach something's wrong
21:27 Jordach why are the positions saved as *thousands* and not the actual positions
21:27 sapier I've got about 1/3 of textures right now :)
21:27 sapier ok stopped downloading
21:27 _Esteban_ joined #minetest-dev
21:27 Jordach sapier, http://paste.debian.net/114048/
21:27 _Esteban_ left #minetest-dev
21:28 sapier I don't know I didn't change player files recently
21:28 Jordach sapier, cg's last position is 22:25:28: ACTION[ServerThread]: crazyginger72 places node deco:glass at (-142,7,220)
21:29 Jordach that file is recording location wrong
21:29 Jordach (and i hit ctrl+c sorry)
21:29 sapier no problem ... are you sure it's in position and not in blocksize positions?
21:29 Jordach sapier, read the paste, the postions are in thousands and not hundreds as stated by node placement
21:30 sapier yes as I said it could be blocksize positions instead of lua positions
21:30 Jordach sapier, but why would it do that.
21:31 sapier don't ask me but the one who wrote it ;-)
21:31 Jordach sapier, does not occur with BlockMen's last build
21:32 Jordach server is running Xubuntu 14.04 with a build dated to a few weeks
21:33 sapier ok if it's not for all build's my assumption about blocksize positions is most likely wrong
21:34 Jordach sapier, this is just afer RBA's plantlike stuff
21:34 Jordach iirc
21:34 Jordach and resumable digging
21:35 sapier resumable digging wás added?
21:36 VanessaE since when?
21:36 Jordach sapier, switch to food: node stays dug
21:36 Jordach partially
21:36 VanessaE eh...that hardly counts.
21:36 VanessaE you're just exploiting a glitch
21:38 sapier I don't see any commit that might cause this directly .... of course it could be a side effect but you'll have to find out this issue as by now you're the only one who can reproduce it
21:38 Jordach sapier, only on Linux, not Windows
21:39 VanessaE sapier: I can confirm a slight drop of FPS while the inventory is open - 5 to 10 fps, but nothing remotely similar to Jordach's experience.
21:39 Jordach VanessaE, cg72 also confirmed it
21:39 VanessaE and I tried it with current minetest_game
21:39 Jordach VanessaE, read the paste and then the placed node
21:39 VanessaE and almost-current engine.
21:39 sapier VanessaE: yes that's about what I see too ... and the drop depends on complexity of menu too ... you can see this quite good in mobf settings menu
21:40 Jordach sapier, it seems player files are damaged by saving
21:40 sapier I've updated my git view yesterday so it should be quite recent
21:40 Jordach (it adds a 0 onto positions where it shouldn't be)
21:40 VanessaE Jordach: no.
21:41 VanessaE player files have been measured in tenths for a LONG time now.
21:41 Jordach VanessaE, on windows it's not like that
21:41 Jordach reported as lua table values
21:41 sapier wait
21:41 VanessaE there's no logic in it, but I've seen this behavior for at least a year now
21:41 sapier windows build?
21:41 Jordach sapier, BlockMen's latest
21:41 sapier there's a commit doing mingw cleanup
21:42 Jordach um oh
21:42 Jordach there it is
21:42 Jordach that broke player saving?
21:42 Jordach MSVC isn't affected by this
21:42 sapier I still don't understand how this can break it but it's worth a try
21:42 sapier https://github.com/minetest/minetest/commi​t/17345404bad71989238d492ed2e23c97f5681065
21:43 Jordach Windows output https://cdn.mediacru.sh/gzpOfeQ56pP2.png
21:43 sapier very strange
21:43 Jordach and if i move the world locally, and not in a VM, that should remove the thousands
21:44 VanessaE Jordach: position = (-758.94,125,2520.19)  <----  my position on Vanilla.  As reported by the client, I am really at -75.9,12.5,252.0
21:44 Jordach VanessaE, that's wrong
21:44 VanessaE (not far from the spawn)
21:44 Jordach windows doesn't add a thousand
21:44 VanessaE Jordach: it's been this way for a LONG ass time
21:44 Jordach VanessaE, it isn't
21:44 Jordach windows clients don't report that
21:45 VanessaE if windows isn't dividing it down, it's reporting wrong.
21:45 VanessaE or if it isn't multiplying it up, it's saving wrong
21:45 VanessaE tell you what, go on Vanilla please.
21:45 sapier jordach I'm at *10 positions in all of my worlds on windows too
21:45 Jordach VanessaE, decimal point is wrong
21:45 sapier -windows + linux
21:45 sapier so most likely the *10 positions are correct
21:46 VanessaE Jordach: the *10 positions are correct.
21:46 VanessaE and if they're not, well, they are now.
22:01 sapier left #minetest-dev
22:03 Jordach VanessaE, cg72 stopped reporting it after the server got recompiled
22:05 VanessaE the fps drop?
22:05 Jordach yup
22:05 * VanessaE shrugs
22:06 VanessaE odd that it happened at all
22:11 CraigyDavi` joined #minetest-dev
22:12 SmugLeaf joined #minetest-dev
22:12 SmugLeaf joined #minetest-dev
23:13 swaaws joined #minetest-dev
23:38 ShadowNinja Hmmm, that's curious.  It's probably because it's stored multiplied by BS (10).
23:39 ShadowNinja I'd be fine with Qt if we could get it to integrate well.
23:40 ShadowNinja Qt has loads of other stuff too, which might be usefull.
23:41 ShadowNinja But it will probably slow down compilation, and Qt may be a bit bloated.
23:55 zat joined #minetest-dev

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