Minetest logo

IRC log for #minetest-dev, 2015-04-27

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

All times shown according to UTC.

Time Nick Message
00:02 paramat joined #minetest-dev
00:07 paramat now pushing game#310
00:07 ShadowBot https://github.com/minetes​t/minetest_game/issues/310 -- Allow overriding papyrus and cactus grow functions by HybridDog
00:14 paramat done
00:24 paramat anime time!
00:24 paramat left #minetest-dev
00:49 hmmmm ShadowNinja: I don't like how dstream was changed.  I want my old dstream back
00:57 Wayward_Tab joined #minetest-dev
01:01 ShadowNinja hmmmm: That would involve a big `find src -type f | xargs -IX sed -i 's/dstream/nonestream/g' X` though.
01:03 ShadowNinja That's about 160 lines of changes.
01:03 hmmmm you mean in your patch?
01:03 ShadowNinja Yes.
01:03 hmmmm well why'd you change it all to begin with
01:04 ShadowNinja hmmmm: Look at the old debug.cpp.  It's quite ugly and it uses the C-style FILE* API.
01:05 hmmmm C-style isn't necessarily ugly
01:05 ShadowNinja dstream was the C++-style public interface for the stderr/file combo.
01:05 ShadowNinja hmmmm: Yes, but this was.
01:05 hmmmm maybe i'm misunderstanding, lemme look at the patch...
01:07 ShadowNinja dstream used to be completely seperate from errorstream, infostream, actionstream, etc.  dstream was a much lower-level API.  Instead of adding a StreamLogOutput(std::cerr) and a FileLogOutput("debug,txt") you added a StreamLogOutput for either dstream or dstream_no_stderr.
01:08 hmmmm different question
01:08 ShadowNinja (Although there were two seperate classes for dstream and dstream_no_stderr that were near;y identical, I combined those.)
01:08 hmmmm print() in lua is now the system lua print?
01:08 hmmmm i.e. it's not being overridden anymore
01:09 ShadowNinja hmmmm: Yes, so it goes to stdout like one would expect.
01:09 hmmmm okay
01:11 hmmmm oh eww
01:11 hmmmm why do you need to have a using namespace directive inside of the code block
01:12 ShadowNinja I don't need to, but it reduces a lot of nomespace specialization and I don't want to do it globally.
01:12 ShadowNinja I'd love to have C++11's using x = y; statement and enum classes though.
01:13 Taoki joined #minetest-dev
01:31 Miner_48er joined #minetest-dev
03:05 hmmmm god dammit shadowninja
03:05 hmmmm for as much as you bitch about code style, your code is not compliant with the style
03:06 hmmmm and then there's the unnecessary operator overloading
03:23 hmmmm alright upon closer inspection i don't approve of this commit anymore
03:24 hmmmm this adds just as many style issues as it cleans up in terms of style
03:25 hmmmm it adds unnecessary added abstractions that don't have any solid benefit
03:25 hmmmm I'm getting really weary about these "cleanup" commits
03:25 hmmmm they usually don't fix any actual problems aside from preference
03:29 hmmmm for example:  what is the point of LogOutputEntry?  all that does is wrap around ILogOutput and move bool silenced to that instead
03:29 hmmmm how does it make it cleaner to add a new wrapper class
03:29 hmmmm this is so bloaty
03:39 Hunterz joined #minetest-dev
04:05 Player_2 joined #minetest-dev
04:24 est31 joined #minetest-dev
04:31 Zeno` joined #minetest-dev
05:15 crazyR_ joined #minetest-dev
05:25 hmmmm https://github.com/kwolekr/minetest/commit​/cd1d625ab21e741e91be7d2190bb4fd59fab3200
05:28 est31 looks good
05:28 hmmmm k coo
05:44 sfan5 hmmmm: are you fine with merging #2641 now or do we wait for some reason?
05:44 ShadowBot https://github.com/minetest/minetest/issues/2641 -- Revert the upper-case PROJECT_NAME nonsense that was part of #2402 by sfan5
05:44 hmmmm yeah, sure.
05:44 hmmmm I agree with it too, even if I don't care too much about it
05:44 hmmmm it seems silly to optimize for the uncommon case.
05:45 hmmmm the common case is lower
05:46 * est31 still waits for reviews for #2620
05:46 ShadowBot https://github.com/minetest/minetest/issues/2620 -- SRP based login by est31
05:46 hmmmm sorry man
05:46 hmmmm it's really big and we're busy
05:46 hmmmm i'll have to schedule a time to review it
05:46 est31 good
05:46 sfan5 I'll merge 2641 now the
05:46 hmmmm yeah
05:46 sfan5 n
05:46 hmmmm for larger commits I think we should explicitly allocate time for reviewing
05:47 hmmmm because otherwise nobody's gonna do it
05:47 sfan5 error: patch failed: misc/winresource.rc:37 (╯°□°)╯︵ ┻━┻
05:47 Hunterz joined #minetest-dev
05:47 sfan5 guess it's gonna wait until after school...
05:47 * sfan5 afk
05:47 hmmmm hah okay
05:47 hmmmm well that flopped
05:47 hmmmm i thought it passed win32 on travis...
05:49 hmmmm est31, what time zone are you?
05:53 est31 tough question
05:53 est31 there are physical time zones and actually lived time zones
05:53 est31 my physical is CEST
05:54 hmmmm well my actual goal is to find what times I am most likely to catch you around
05:57 est31 for today, I think I'll be online in 12 hours again.
05:58 hmmmm hmm ok
05:59 hmmmm any times after 8:00 PM EST tomorrow??
06:03 est31 ok
06:03 hmmmm mostly in case if i have to ask you questions
06:04 Player_2 joined #minetest-dev
06:04 * Zeno` glances about
06:04 Zeno` prng exception?!
06:04 hmmmm i need it for unit tests
06:05 Zeno` yuck :(
06:05 Zeno` oh well
06:05 hmmmm hey got any better ideas
06:05 hmmmm apart from 'let's crash it weeeeee'
06:07 Zeno` Not really. But I loath exceptions with continuation semantics. But nvm
06:07 est31 or no sorry
06:07 est31 day after tomorrow would be ok
06:07 hmmmm continuation semantics?
06:08 Zeno` I'd rather the program just stopped
06:08 hmmmm well see
06:08 hmmmm this effectively does that
06:08 hmmmm nobody else aside from unit tests catch PrngException
06:09 Zeno` yet :)
06:09 Zeno` anyway, it's too silly to argue about. I just don't like it. Others do. That's how it is :D
06:10 hmmmm the only solid excuse i can think of to catch it is in LuaPseudoRandom so you can convert it into a LuaError
06:11 Zeno` hmm
06:12 Zeno` Why is RANDOM_RANGE so small? Limitation of the algorithm?
06:13 hmmmm because it sucks
06:13 Zeno` heh
06:13 hmmmm lol i don't know
06:14 hmmmm there's a better variant of this same exact LCG in glibc
06:14 hmmmm it populates all 32 bits
06:18 Zeno` hmm
06:22 Zeno` So... the range -20, -10 is not valid?
06:22 jin_xi joined #minetest-dev
06:22 hmmmm for PseudoRandom::range()?  yes of course
06:24 Zeno` for inline int range(int min, int max)
06:25 hmmmm not sure what you're asking
06:25 Zeno` -20 - -10 is 4294967286 isn't it?
06:25 Zeno` (when cast to u320
06:25 Zeno` )
06:26 hmmmm i'm pretty sure PseudoRandom uses int everywhere
06:26 hmmmm oh that cast
06:26 Zeno` https://github.com/minetest/minetest/commi​t/cd1d625ab21e741e91be7d2190bb4fd59fab3200​#diff-0adfb87b0afcca4db97b6e67f11ecc91R67
06:26 Zeno` I dunno... I'm probably wrong
06:28 hmmmm min = -20, max = -10.  -10 - -20 = 10
06:29 hmmmm max - min is always guaranteed to be a non-negative number if min < max
06:29 hmmmm that's just how math works
06:30 Zeno` lol
06:31 Zeno` I was doing -20 - -10 :/
06:31 Zeno` I really need to sleep more often
06:31 hmmmm i feel like there is something horribly wrong with randNormalDist()
06:31 hmmmm don't think i ever looked at output until now:  http://fpaste.org/215691/11594214/
06:31 * hmmmm shudders
06:32 hmmmm that my friend does not look like a bell curve to me
06:32 Zeno` well a range of 32768/10 seems awfully small to me (the only reason I was really looking at that line in the first place)
06:33 hmmmm it's because the algo sucks.  if you use PcgRandom this won't happen
06:33 Zeno` oh that's the old crappy one?
06:33 * Zeno` looks at the whole file
06:34 Zeno` pfft... PseudoRandom
06:34 Zeno` pity it cannot be blown up
06:34 Zeno` without breaking stuff heh
06:35 hmmmm yeah :(
06:36 hmmmm man I really have no idea what I'm doing wrong
06:36 hmmmm unless I totally 100% misunderstand statistics
06:36 hmmmm an improbable amount of values are bunching up in the center
06:43 hmmmm i think i know what's going on, it's working normally but the peak is so sharp because i essentially squish the standard deviation so the probability is effectively 0 by the time it gets to either end
06:46 hmmmm how on earth do i even address this
07:02 kilbith joined #minetest-dev
07:10 Darcidride joined #minetest-dev
07:20 nore joined #minetest-dev
07:44 hmmmm acually nope, just stretched it and the problem persists
07:45 hmmmm http://i.imgur.com/seiPaNr.png
07:59 hmmmm ahh this is interesting
07:59 hmmmm it only happens if part of the range is negative
07:59 hmmmm sounds like i have a bug rather than a flawed algorithm
07:59 hmmmm :D i love unit testing.  i would've never found this otherwise
08:03 hmmmm :( it was the float-to-int truncation in the negative case.
08:04 Yepoleb joined #minetest-dev
08:04 blaze` joined #minetest-dev
08:08 hmmmm https://github.com/minetest/minetest/commi​t/415167b228dea8b4108d7fe164c7fec8c061dcf5
09:22 Megaf_ joined #minetest-dev
09:43 kilbith joined #minetest-dev
10:06 Elinvention joined #minetest-dev
10:21 selat joined #minetest-dev
10:34 leat joined #minetest-dev
10:42 est31 joined #minetest-dev
10:47 leat1 joined #minetest-dev
10:58 leat1 joined #minetest-dev
11:08 leat1 joined #minetest-dev
11:47 proller joined #minetest-dev
11:50 cib0 joined #minetest-dev
12:20 Zeno` joined #minetest-dev
12:30 proller joined #minetest-dev
12:35 Elinvention joined #minetest-dev
12:38 leat1 joined #minetest-dev
12:48 leat1 joined #minetest-dev
13:05 Wayward_One joined #minetest-dev
13:25 Zeno` Anyone want to donate to me $600 to keep my server running? :p
13:32 FR^2 joined #minetest-dev
13:39 exio4 I may be able to donate 0.60USD Zeno`
13:39 exio4 would it be enough?
13:42 geheimnis joined #minetest-dev
13:59 sfan5 Zeno`: what type of server costs $600?
14:05 Zeno` VPS
14:09 sfan5 Zeno`: i hope you don't mean $600/month
14:10 Zeno` no, no, year
14:11 sfan5 $50/month is still much..
14:19 sfan5 merging 2641 in 5 minutes, hmmmm agreed to it this morning
14:36 Elinvention joined #minetest-dev
14:41 cib0 joined #minetest-dev
15:29 Krock joined #minetest-dev
15:34 Calinou joined #minetest-dev
15:37 proller joined #minetest-dev
15:49 leat1 joined #minetest-dev
16:01 blaze joined #minetest-dev
16:04 kilbith ShadowNinja, can you inflict a penalty to the old versions of MT on the serverlist, please ?
16:05 kilbith they're not likely to be maintained anymore and gives a bad impression of MT
16:05 Calinou we shouldn't support old versions in master at all, like Red Eclipse does
16:05 Calinou let them be played outside of master server but don't show old servers on list at all
16:06 Calinou we don't support old versions of software, this is a game
16:13 est31 joined #minetest-dev
16:13 est31 pushing #2632
16:13 ShadowBot https://github.com/minetest/minetest/issues/2632 -- Add russian translation to minetest.desktop by mich1
16:15 est31 pushed
16:24 hmmmm joined #minetest-dev
16:24 Hunterz joined #minetest-dev
17:05 rubenwardy joined #minetest-dev
17:39 Sokomine the server list does need some work. it's too long to be convenient for most people. and it's difficult to locate a particular server in the list even though you know how it's called
17:40 Calinou fix: remove old servers, prevent them from registering
17:40 Calinou that'll prune 50 % of the servers at least, whose admins can't be bothered to use git pull
17:44 rubenwardy old meaning before stable, of course
17:45 Calinou yeah, < 0.4.12
17:45 kilbith and if the versionning is broken like now ?
17:46 kilbith i still have 0.4.11 after compiling
17:46 Krock it's not broken, use use dev
17:46 sfan5 in other words, you want to throw 14 server off the list?
17:46 sfan5 Krock: it is
17:46 sfan5 the 0.4.12 isn't on the master branch
17:46 Krock ummm.. okay, if you say so, it must be that way.
17:46 sfan5 which is why git detects 0.4.11
17:46 est31 Calinou, what about forks?
17:46 Calinou the Minetest master server is for Minetest, not other games.
17:46 Calinou — eihrul (edited)
17:47 sfan5 (stefan@stefan-pc) /tmp/mtof % git describe --tags --always --dirty
17:47 sfan5 0.4.11-501-g4ea5a96
17:47 Calinou yeah, Git versioning is broken
17:47 Calinou so we might have to keep supporting 0.4.11
17:47 Krock The master server doesn't need a change but the mainmenu does. A search box or a filter would help a lot
17:47 est31 there is the favourite list
17:48 est31 but yes there needs to be improvement
17:48 rubenwardy See freeminer's server menu
17:49 Krock Not connection should be added to the favourite list, let's do it manually like bookmarks in the browser.
17:49 leat1 joined #minetest-dev
17:49 Krock Sometimes I have trouble finding a server again because there are just 10 new on top
17:50 est31 yes thats good
17:50 est31 when encryption gets added, servers can get unique id numbers
17:52 sfan5 yay making a _game_ server as complicated to maintain as a banking app
17:53 est31 there are things in between
17:53 est31 banking apps are corporate world
17:54 est31 I don't see any reasonable argument here sfan5. You don't even know what I propose
17:54 est31 is ssh as difficult to maintain like a banking app
17:54 est31 ?
17:55 sfan5 anytzhing with uniquite id sounds complicated
17:55 sfan5 wtf
17:55 sfan5 anything* unique*
17:55 sfan5 it's going to be pub
17:55 sfan5 it's going to be public key crypto, right?
17:55 est31 yes
17:55 est31 or like with onion addresses a part of the hash
17:56 est31 nobody likes a 60 character base64 string as id
17:56 sfan5 ssh is more complicated to maintain than ssh
17:56 sfan5 uh
17:56 sfan5 telnet
17:56 est31 for me it isnt
17:56 est31 I have to remember pesky passwords for telnet
17:57 est31 (ok admit its unrelated)
17:59 est31 the issue is, if we ever want to do some reliable storing for the master list, we need unique ids
18:00 est31 so for example verify a server's age
18:02 sfan5 this is exactly what i mean
18:02 sfan5 this is _game_
18:02 sfan5 we don't need verification of the servers "age", number of players or any of those other things
18:03 Sokomine kilbith: same here. my version doesn't grow over 0.4.11 for some unknown reason
18:03 kilbith it happened during the last releasing
18:04 est31 sfan5, perhaps one day we can show statistics for the servers together with how many users connected etc?
18:04 Sokomine i don't think throwing servers off the list is a solution. what we need is more structure so that players can find servers that fit to their needs
18:04 sfan5 est31: we can do that already, there jist isn't any code that does that
18:04 Sokomine builders have diffrent requirements than tablet players who just want to peek in
18:05 est31 sfan5, so how identify a server? by its address?
18:05 est31 address can change
18:05 est31 by the name?
18:05 est31 "there is no name"
18:05 Sokomine just counting locally how often a player connected to a server and showing that in the favorite tab might help a bit
18:05 sfan5 i would identify the server by hostname and port
18:05 est31 that can change
18:05 Sokomine at least for long-time players. new ones still have a problem
18:06 sfan5 but it's less likely to change than any other thing
18:06 est31 in the current list yes
18:08 est31 the thing about encryption is that most people will ignore it anyway. so it should be made as easy to ignore as possible. but when somebody likes that feature why take it away?
18:08 est31 I mean why do we need passwords the first place?
18:09 est31 "its just a game", don't we trust the players?
18:10 Sokomine it's all not so much about security. the troublie is that there are more servers than can be memorized. and throwing a long list at people where the list is sorted...some way...just doesn't really help. it's still far better than the old way where there was no list at all
18:10 Hijiri joined #minetest-dev
18:11 Sokomine hmm. server imposters are slightly less likely than player imposters. takes more to run a server
18:11 est31 yea perhaps we can add some interactivity on the client side
18:11 Sokomine maybe make a confirmed id optional? for the larger, well maintained servers?
18:11 Sokomine most of this ought to happen client-side, yes
18:11 Sokomine maybe a sort-by-name option
18:12 est31 uh no
18:12 est31 either real search or leaving it this way
18:12 Sokomine yes, trouble is that that'd again favor some servers. but right now, finding one again is not easy
18:12 est31 because you know its MinetestForFun, but is it [EN] MinetestForFun, or [Fr/En] etc etc
18:13 est31 why would that favour severs?
18:13 est31 I mean if servers want to game they just themselves AAAAAAAA
18:13 est31 or some better variation thereof
18:14 Sokomine there's client/serverlist/favoriteservers.txt right now. that's sufficient as a base. it's the presentation that needs change. well, and maybe some more information from the server
18:14 Sokomine servers playing mt against each other? :-)
18:15 Wayward_One joined #minetest-dev
18:15 est31 no game the system
18:15 Sokomine tss
18:15 est31 I mean its really no effort
18:15 est31 the cryptographic id
18:15 est31 you just have one file more in the world's dir
18:16 est31 is it that painful?
18:16 Sokomine anyway. information that's needed is suitability for builders, acceptance of pvp (damage is not an indicator) and suitability for mobile clients i'd say. that would be helpful extra information
18:16 est31 suitability for mobile clients?
18:16 est31 how would you make that
18:16 est31 ?
18:16 Sokomine just make it optional, est31. a server that provides encrypted information is probably one where the server owner is less clueless than others
18:17 est31 why do people think encryption is a hassle?
18:17 kilbith also force the line-breaking please : https://lut.im/owEpEgm6/l8J65AE0  so that the row is not shifted
18:17 est31 see ^ thats a https url
18:17 Sokomine there are a lot of server that don't work well with the android client. they have many mods installed with many complex textures. that's not very phone-friendly
18:18 est31 yea perhaps that would be nice, but the issue with that flag is, how set it?
18:18 est31 also, perhaps thats rather a workaround than a proper fix
18:18 Sokomine est31: some server operators ARE clueless and don't know much. those servers ought to show up on the list as well. it's not bad if it's shown that that server operator did not put too much effort in yet
18:19 Sokomine i don't think there's a way to "fix" that. at least not that smartphones and tablets are not entirely equal to high-end-machines. well, except maybe wait a copule of years
18:20 est31 so you want to make encryption difficult so that only experienced people enable it?
18:20 Sokomine the server could indicate those values, and we could provide an faq explaining which setting would be suitable in which cases
18:21 Sokomine est31: no. but if there are complaints that it's too complicated, optional encryption is perfectly fine
18:22 Sokomine but all that is of little help regarding the inconvenient current server list
18:23 est31 it is related
18:23 est31 I mean encryption itself is very lean for users
18:23 Sokomine yes. hard to sort somethiing where there are no usful ids
18:23 est31 its basically just a function which accepts a key, a value and outputs something encrypted
18:23 est31 other way round for decryption
18:24 est31 and about the keys, we have to bind them somehow onto something else
18:24 est31 the way the www does it is with certificates
18:24 est31 I agree that is horrible
18:25 est31 really, even for that usecase
18:25 est31 now we have ssh
18:25 est31 there we connect to a certain server, and when we are connecting first time, we are asked whether to store the key, and then never bothered again
18:27 est31 what I propose is even simpler than that (no question)
18:27 est31 ssh is very easy on the server side
18:27 est31 basically you only have to generate keys
18:27 est31 but that can be done automatically too
18:28 est31 but this discussion is quite unproductive
18:29 est31 (about encryption)
18:30 Sokomine hm, yes. encryption certainly won't hurt
18:30 est31 so, Sokomine you have my support when you want to code a patch that filters the server's name inside the list.
18:30 Sokomine just make it optional
18:30 Sokomine yes. i'm just not entirely sure how it could be presented in a better way
18:31 Sokomine as this is a very visual game, screenshots of servers might be helpful. like what some operation systems or browsers do nowadays...make a tiny screenshot of what you saw last when you visited a world
18:31 Sokomine might help identifying a server
18:33 Sokomine and servers on the public list could provide a tiny screenshot...
18:33 Calinou sounds like a potential security issue
18:33 Calinou people could put scary pics or worse
18:33 est31 thats no security issue
18:34 rubenwardy screenshots would be good
18:34 kilbith they still can put scary title/description as well...
18:34 rubenwardy They could do that in the game with content, as well
18:34 kilbith no reason to scare about the screenshots specifically then
18:35 Sokomine mmdb already stores pictures for mods. maybe a similar mechanism could be used? a server sends in a picture, and if one forum dev approves of it, the picture can get used
18:35 Calinou no, approving is too cumbersome
18:35 Sokomine that'd be a delay of maybe a few days, but that doesn't matter. and there are not *that* many servers
18:35 est31 Sokomine, minetest is a game no banking app
18:35 est31 sfan5 is right in that point
18:36 est31 this just makes making a server even harder
18:37 Sokomine i'm pretty sure that tiny screenshots of the server and some more general information might help players. some are pretty young
18:37 Sokomine right now, the interface is ok for a banking app as far as usability goes - but not for a game. that's the trouble
18:38 est31 the client tab is already very cramped
18:38 Sokomine it *is* hard to make a good server. servers which did not yet put that much effort in ought to be listed as well of course
18:38 Sokomine it is
18:38 Sokomine we need to improve that somehow
18:39 * est31 should perhaps better go offline now (I'm tired)
18:39 est31 bye
18:42 Miner_48er joined #minetest-dev
18:44 MinetestForFun joined #minetest-dev
18:50 cib0 joined #minetest-dev
18:55 rubenwardy_ joined #minetest-dev
19:11 leat1 joined #minetest-dev
19:15 paramat joined #minetest-dev
19:22 paramat hmmmm and all, here's my first attempt at a mod that converts a lua table defined schematic into a .mts file and saves it in a schems folder in the mod https://gist.github.com/pa​ramat/78dfd2e44b7839b1f7da it works, thanks for this
19:22 paramat left #minetest-dev
19:27 Robert_Zenz joined #minetest-dev
19:33 FR^2 joined #minetest-dev
19:38 Fritigern joined #minetest-dev
20:02 Hunterz1 joined #minetest-dev
20:06 AnotherBrick joined #minetest-dev
20:14 Wayward_Tab joined #minetest-dev
20:14 Sokomine i've also written a converter for worldedit .we to .mts, which is part of mg_villages. i've ignored the lua table definition of schematics completely so far. .mts is a good storage format
20:31 Hijiri joined #minetest-dev
20:35 sofar Sokomine: MT we or MC we?
20:37 AnotherBrick joined #minetest-dev
20:40 leat1 joined #minetest-dev
20:49 Sokomine sofar: MT we. do you happen to have a documentation for mc we?
20:56 sofar Sokomine: no, looked into converting my kids' world from MC to MT, which ended up okay.....ish
20:56 sofar but I converted the whole world in the end, not just parts
21:07 AnotherBrick joined #minetest-dev
21:28 proller joined #minetest-dev
21:48 Hijiri joined #minetest-dev
21:48 shadowzone joined #minetest-dev
22:53 Wayward_Tab joined #minetest-dev
23:03 proller joined #minetest-dev
23:11 crazyR joined #minetest-dev
23:45 proller joined #minetest-dev
23:46 Wayward_One joined #minetest-dev
23:54 Hijiri joined #minetest-dev

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