| Time |
Nick |
Message |
| 00:25 |
|
w00tc0d3 joined #minetest-dev |
| 00:58 |
|
OldCoder joined #minetest-dev |
| 01:17 |
|
kaeza joined #minetest-dev |
| 01:44 |
ShadowNinja |
I made this get_core_version patch for the Minetest IRC mod. It's MUCH nicer than a API function because it's unlikely people will do version comparisons with this. ;-) http://ix.io/c8C/diff |
| 01:45 |
ShadowNinja |
It tries to find the bin path by using the modpath, or `which`, or assuming /usr/bin. It then tries to use `git describe` to find the version, if that fails it opens up the binary and looks for "VER=". |
| 01:46 |
ShadowNinja |
Foolproof! |
| 01:48 |
ShadowNinja |
It works quite well. It only fails for systemwide installs that aren't in the path or /usr/bin/. Or platforms where `which` is missing. |
| 01:48 |
ShadowNinja |
I should probably remove the git detection actually, since the binery can be a different version that what you're using. |
| 01:48 |
ShadowNinja |
binary* |
| 01:49 |
ShadowNinja |
Oh, and lod versions might not have a VER= set. |
| 01:49 |
ShadowNinja |
old* |
| 01:49 |
ShadowNinja |
I haven't tested it on Windows, but it should work there too. |
| 02:04 |
ShadowNinja |
Any objections to http://ix.io/c8F/diff ? I'll push it semi-soon. |
| 02:33 |
ShadowNinja |
I've created a proper version API. It returns something like this: {["patch"] = 9, ["minor"] = 4, ["string"] = "0.4.9-dev", ["hash"] = "0.4.9-214-gc80d67f-dirty", ["extra"] = "", ["major"] = 0} |
| 02:34 |
ShadowNinja |
The main menu API already had a simpler version of this for display. |
| 02:35 |
ShadowNinja |
The major/minor/patch fields could be removed if that would make strict version checks less likely. |
| 02:37 |
ShadowNinja |
I also have a patch adding write_json to the async API pending (I missed that before). |
| 03:36 |
|
Exio4 joined #minetest-dev |
| 05:28 |
sfan5 |
ShadowNinja: >it's unlikely people will do version comparisons with this. Don't assume modders won't do what they can't. |
| 05:28 |
sfan5 |
s/can't/can/ |
| 05:29 |
ShadowNinja |
sfan5: O.K. I'll assume that modders will write executable reading functions like this and use them for raw version checking. |
| 05:30 |
sfan5 |
maybe not that, but if it were to get added to the API |
| 05:30 |
ShadowNinja |
sfan5: But comments on a proper API function? |
| 05:31 |
ShadowNinja |
sfan5: I just did that, minus the version checking. |
| 05:32 |
* ShadowNinja |
is trying to implement luaL_typerror for a Lua 5.2 update. |
| 06:14 |
ShadowNinja |
\o/ Minetest compiles with Lua 5.2 now. 'Night. |
| 06:19 |
|
w00tc0d3 joined #minetest-dev |
| 06:51 |
|
smoke_fumus joined #minetest-dev |
| 07:12 |
|
darkrose joined #minetest-dev |
| 08:24 |
|
rsiska joined #minetest-dev |
| 08:26 |
|
tomreyn joined #minetest-dev |
| 08:58 |
|
ImQ009 joined #minetest-dev |
| 09:15 |
|
proller joined #minetest-dev |
| 09:16 |
|
nore joined #minetest-dev |
| 10:17 |
|
jin_xi joined #minetest-dev |
| 10:17 |
|
troller joined #minetest-dev |
| 10:24 |
|
troller joined #minetest-dev |
| 10:28 |
|
proller__ joined #minetest-dev |
| 10:53 |
|
ImQ009 joined #minetest-dev |
| 11:03 |
|
Shardvexz joined #minetest-dev |
| 11:35 |
|
Garmine joined #minetest-dev |
| 11:56 |
|
BrandonReese joined #minetest-dev |
| 12:29 |
|
Shardvexz joined #minetest-dev |
| 12:38 |
|
troller joined #minetest-dev |
| 13:14 |
|
hmmmm joined #minetest-dev |
| 13:55 |
|
Amaz_ joined #minetest-dev |
| 14:00 |
|
PilzAdam joined #minetest-dev |
| 14:22 |
|
Amaz_ joined #minetest-dev |
| 14:54 |
|
Jordach joined #minetest-dev |
| 14:57 |
|
zat joined #minetest-dev |
| 15:02 |
|
troller joined #minetest-dev |
| 15:25 |
|
Shardvexz joined #minetest-dev |
| 15:47 |
|
Amaz_ joined #minetest-dev |
| 16:08 |
|
ImQ009 joined #minetest-dev |
| 16:18 |
|
Calinou joined #minetest-dev |
| 16:19 |
|
PenguinDad joined #minetest-dev |
| 16:35 |
|
zyxwvuts joined #minetest-dev |
| 16:38 |
|
troller joined #minetest-dev |
| 16:41 |
|
Anchakor_ joined #minetest-dev |
| 16:42 |
celeron55 |
does someone oppose it if the "minetest" and "engine" namespaces in Lua are officially renamed to "core"? the reason for doing this is to not make forks of minetest need to support the name of every other fork, and the reason why i propose "core" is because it's short and sounds better than "engine" when combined with a dot and some function |
| 16:43 |
|
deltib joined #minetest-dev |
| 16:43 |
celeron55 |
also if someone can propose a better name, at least i can consider it; i'm not really happy with these alternatives |
| 16:46 |
celeron55 |
(i personally don't like this transition but it has to be done in order to not cause problems in the long run) |
| 16:46 |
Megaf |
Sound good to me |
| 16:53 |
|
Zeitgeist_ joined #minetest-dev |
| 16:53 |
|
Zeitgeist_ joined #minetest-dev |
| 16:55 |
|
rubenwardy joined #minetest-dev |
| 17:03 |
|
sapier joined #minetest-dev |
| 17:03 |
|
BlockMen joined #minetest-dev |
| 17:04 |
BlockMen |
celeron55, i prefer engine, not least because we already have engine |
| 17:06 |
* BlockMen |
still want opinons on this: https://github.com/minetest/minetest/pull/1265#issuecomment-42318782 |
| 17:07 |
sapier |
ShadowNinja: you didn't really understand why we do not support reading core version from api do you? |
| 17:07 |
ShadowNinja |
sapier: I do understand why. But it's a bad reason. |
| 17:08 |
sapier |
no it's the only correct reason, btw if you way of reading it was encouraged we'd never ever have a chance to really sandbox mods in order to get them fool proof |
| 17:09 |
sapier |
opening arbitrary files especially binaries is worst thing you can do if you want to get a safe environment |
| 17:11 |
ShadowNinja |
sapier: The binary reading way is encouraged by not offering another option (other than memory inspection) |
| 17:11 |
sapier |
celeron55: Yes I'm against renaming too, it's just adding another compatibility layer for no really understandable reason. saving 4 letters isn't a reason to me justifying modifying any mod out there or adding another compatibility alias |
| 17:11 |
celeron55 |
i propose that we expose an api that gives out a version string, in the format "Minetest x.y.z" |
| 17:11 |
sapier |
we already have way to much compatibility code we need to get rid of |
| 17:11 |
celeron55 |
the reason is that apparently some mods just want to tell the user on what they are running |
| 17:12 |
celeron55 |
sapier: what? |
| 17:12 |
celeron55 |
so "minetest = core" is too much code? |
| 17:12 |
celeron55 |
come on now |
| 17:12 |
sapier |
what's use for doing so? |
| 17:12 |
celeron55 |
i explained why it has to be done, please read it |
| 17:12 |
sapier |
you just add confusion because of mods use mixed mode |
| 17:12 |
celeron55 |
19:45:46 <@celeron55> does someone oppose it if the "minetest" and "engine" namespaces in Lua are officially renamed to "core"? the reason for doing this is to not make forks of minetest need to support the name of every other fork, and the reason why i propose "core" is because it's short and sounds better than "engine" when combined with a dot and some function |
| 17:12 |
sapier |
I read it but I can't follow your explanation |
| 17:12 |
celeron55 |
19:45:51 -!- mode/#minetest-dev [-o celeron55] by celeron55 |
| 17:13 |
sapier |
renaming something widely used because it "sounds" better isn't a reason I understand |
| 17:13 |
BlockMen |
core doesnt sound better than engine |
| 17:13 |
celeron55 |
---------------> "the reason for doing this is to not make forks of minetest need to support the name of every other fork" |
| 17:14 |
celeron55 |
are people blind or what |
| 17:14 |
celeron55 |
there will be a day when some fork decides that they don't want "minetest" in their code, and that will hurt everyone |
| 17:14 |
sapier |
and why do you expect forks not to customize their own api? |
| 17:14 |
BlockMen |
^ that |
| 17:14 |
celeron55 |
because the api is large and most things can be done with it without extensions |
| 17:15 |
sapier |
if they change name they most likely do it exactly to be incompatible |
| 17:15 |
BlockMen |
and they can keep backwards compatibility to minetest if they want to |
| 17:15 |
celeron55 |
no? they do it because it's weird to have a random name in an API |
| 17:16 |
celeron55 |
so we assume that no good mods will be developed primarily for any fork of minetest? |
| 17:16 |
sapier |
that's your opinion I can't follow your logic |
| 17:16 |
ShadowNinja |
BlockMen: Builtin should have a usable HUD. Other wise all games have to provide them and old or abandoned games will stop working. |
| 17:16 |
celeron55 |
that minetest users wouldn't want to use? |
| 17:16 |
Anchakor |
they might want to get backports from minetest (security fixes for example) which would random renaming & code style changes make PITA |
| 17:16 |
|
Exio4 joined #minetest-dev |
| 17:16 |
celeron55 |
Anchakor: oh god, i'm glad someone understands |
| 17:16 |
sapier |
basicaly renaming causes all mods out there to use a deprecated api. Any reasonable developer tries to replace deprecated api and therefore you cause a lot of work |
| 17:17 |
celeron55 |
it won't be changed ever again; you're going to cause a log more work in the long term otherwise |
| 17:17 |
BlockMen |
ShadowNinja, we already had this, e.g. tree/grass growing |
| 17:17 |
celeron55 |
lot* |
| 17:17 |
BlockMen |
it was also moved to games |
| 17:17 |
ShadowNinja |
find . -type f | sed -I s/minetest/core/ |
| 17:18 |
ShadowNinja |
-i* |
| 17:18 |
sapier |
I know nothing is as constant as change so don't expect me do believe "it won't be changed ever again" |
| 17:18 |
celeron55 |
why would it be? |
| 17:18 |
BlockMen |
and since we are talking about the dev (what master is) there is no need to do it that way |
| 17:18 |
BlockMen |
furthermore you dont need to read out the flags |
| 17:18 |
ShadowNinja |
BlockMen: Games don't need to have trees at all. All games need a HUD. |
| 17:18 |
celeron55 |
and if you think change is constant, why do you try to oppose it anyway |
| 17:19 |
sapier |
maybe because we realize core collides with namespace of some other lib we wanna use ... but I'm not concerned about those things I can imagine but because of those I cant |
| 17:19 |
BlockMen |
ShadowNinja, so you would call minetest_game without trees and grass not broken? |
| 17:19 |
BlockMen |
interesting |
| 17:19 |
celeron55 |
additionally, we don't have to deprecate "minetest" |
| 17:19 |
sapier |
and this is even a reason you added celeron55 |
| 17:20 |
ShadowNinja |
BlockMen: minetest_game needs it, but, eg, moontest doesn't. |
| 17:20 |
sapier |
of course we cann add engine core base minetest and whatever other things to namespace to but that's way messies do |
| 17:20 |
BlockMen |
ShadowNinja, games could also decide not to use default hud? |
| 17:20 |
BlockMen |
where is the difference? |
| 17:21 |
ShadowNinja |
BlockMen: I don't know what you're getting at. But see my compatibility point. |
| 17:21 |
celeron55 |
sapier: do you realize if at some point we realize that there is a fork that gets a lot of mod development and the mods happen to be compatible with minetest except for the fact that they use a different namespace, it will be too late to switch? |
| 17:22 |
BlockMen |
ShadowNinja, we are talking about a dev state. when minetest is released it is bundeled with _game, so there wont be an compatibility issue |
| 17:22 |
sapier |
and you believe those mods will suddenly switch to "core"? |
| 17:22 |
celeron55 |
suddenly? |
| 17:22 |
celeron55 |
they don't exist yet |
| 17:22 |
celeron55 |
but if the namespace is switched to "core" now, that's the namespace they will be using then in that fork too |
| 17:22 |
ShadowNinja |
BlockMen: There will for every other game out there. |
| 17:23 |
celeron55 |
because the fork sees no reason to use another namespace |
| 17:23 |
celeron55 |
because the namespace is neutral |
| 17:23 |
sapier |
well you did ask for oppinions, I can't follow your reasons that's why I still prefere not adding a second name |
| 17:23 |
celeron55 |
this is my point; i am not sure how high the chance of that situation is though |
| 17:23 |
BlockMen |
ShadowNinja, same situation was for trees and all games that wanted have trees |
| 17:23 |
BlockMen |
you apply double standards |
| 17:23 |
sapier |
and it's nothing different then a second name if you insist on not deprecating minetest |
| 17:24 |
ShadowNinja |
BlockMen: *shrug* trees could be in builtin if there was a way to easily disable them. |
| 17:24 |
celeron55 |
i am fine with not doing this, as long as somebody is able to convince me that there is no reasonable chance for that happening |
| 17:24 |
sapier |
so if you can't resist renaming imho the only sane way is deprecating minetest and force all (reasonable) developers to update their mods |
| 17:25 |
sapier |
and if you do so drop that silly minetest.env compatibility while doing so |
| 17:25 |
BlockMen |
ShadowNinja, you could add a way. like now a "strange" way gets added to hide builtin statbars |
| 17:25 |
BlockMen |
it makes more sense it is more easy to change hud if in default |
| 17:25 |
sapier |
any non reasonable developers will add the minetest = core alias themselfs anyway |
| 17:26 |
Calinou |
<ShadowNinja> BlockMen: Builtin should have a usable HUD. |
| 17:26 |
Calinou |
I reported an issue asking for empty heart/bubble textures in engine |
| 17:26 |
ShadowNinja |
sapier: This isn't about compatability. |
| 17:26 |
celeron55 |
sapier: but that's fine |
| 17:26 |
sapier |
no it's not |
| 17:26 |
sapier |
it's a lot of work for nothing |
| 17:26 |
ShadowNinja |
sapier: It will be fully compatible. |
| 17:26 |
BlockMen |
Calinou, huh? i dont get what you want now |
| 17:27 |
Calinou |
I'm for “coreâ€, short makes for much better code and saved fingers |
| 17:27 |
ShadowNinja |
sapier: I've already done it. It wasn't all that hard. |
| 17:27 |
Calinou |
BlockMen, right now, if you want a background texture for heart/bubble, you need lua HUD (= perceivable lag) |
| 17:27 |
celeron55 |
this is largely a political issue though, i have noted that sapier doesn't like figuring out poltiical issues 8) |
| 17:27 |
sapier |
of course ShadowNinja keep compatibility trash forever causing any change to get even more and more hard |
| 17:27 |
BlockMen |
Calinou, if its in default you can just remove the starbars, done |
| 17:27 |
BlockMen |
its the best way to reach your goal |
| 17:27 |
sapier |
no it's pure technical, if we did start on green meadow I'd fully agree to core |
| 17:28 |
sapier |
but that's not case here |
| 17:28 |
Calinou |
BlockMen, but you pereceive lag due to Lua HUD (and it increases server load) |
| 17:28 |
Calinou |
it's not a solution |
| 17:28 |
ShadowNinja |
We could just use minetest and have the egnine use 'core' internally. As having one name allowed me to clean up the code a bit. |
| 17:28 |
BlockMen |
Calinou, the builtin hud is also lua hud |
| 17:28 |
BlockMen |
there is no difference in lag |
| 17:28 |
BlockMen |
no, im wrong, in biltin there is more lag |
| 17:28 |
Calinou |
in minetest_next it is |
| 17:28 |
Calinou |
in minimal or minetest_game, is it? |
| 17:28 |
sapier |
if you want to use core celeron55 I insist in a clean exit strategy from old naming |
| 17:29 |
BlockMen |
Calinou, plz, look at the pull and read the code |
| 17:29 |
ShadowNinja |
Minetest should use one name internally, regardless of what mods use. |
| 17:29 |
sapier |
because I don't know how much work I had the last year to support stange behaviour present in minetest and everytime I didn't do this I got complains .. and none of you was there to support the cleanup ... contrary you insisted on keeping that insane scaling by 1280 y lines |
| 17:30 |
celeron55 |
do you mean that mods will be forced to use the new namespace after some time? |
| 17:30 |
BlockMen |
do the namespace switch with 0.5.x damnit |
| 17:31 |
sapier |
yes that's my opinion, call "minetest" deprecated now, warn them in next version and force them to change in follow up version |
| 17:31 |
sapier |
for example |
| 17:31 |
|
zyxwvuts left #minetest-dev |
| 17:31 |
celeron55 |
there are a lot of mods that nobody maintains anymore but which work just fine, so i'm not okay with fully breaking compatibility in a long time |
| 17:32 |
sfan5 |
hm |
| 17:32 |
sapier |
so what? if those mods are valuable to community there will be someone to fix it |
| 17:32 |
ShadowNinja |
I wouldn't completely drop support until there was another change that broke compatability because ^. |
| 17:32 |
ShadowNinja |
A warning could be done though. |
| 17:32 |
sapier |
dropping support is the only way to get rid of broken code |
| 17:33 |
celeron55 |
but it should be dropped after that long time |
| 17:33 |
sfan5 |
I know this is some work, but what if we have a 'minetest' namespace that won't get any new functions/things added & a 'core' namespace which does what the 'minetest' namespace did previously |
| 17:33 |
ShadowNinja |
A warning will slow down access to minetest.* though. |
| 17:33 |
celeron55 |
sfan5: nope |
| 17:33 |
BlockMen |
sfan5, even more messy |
| 17:33 |
sfan5 |
uh, ok |
| 17:34 |
sapier |
we wont be able to maintain that much different ways of doing things, and it's becoming more and more messy to do so |
| 17:35 |
celeron55 |
i think i'm going to ask an opinion from... ehm... "the fork" |
| 17:35 |
celeron55 |
it won't be representative of any other future fork though so i'm not sure what use is this of |
| 17:35 |
sapier |
well you can call freeminer freeminer |
| 17:37 |
sapier |
I'm not responsible for scriptapi so it's not my decision but what I know is it's gonna be quite some work to get mobf work again after you change it |
| 17:37 |
celeron55 |
anyway, i am basing this thinking on the facts that minetest will likely be around for years to come, and in those years probably many forks will come that people use for various reasons, and it is likely that at least one of those forks would like to do these two things at the same time: 1) not use the "minetest" namespace, 2) maintain compatibility with minetest |
| 17:37 |
Calinou |
ShadowNinja, post an informal warning on forums |
| 17:37 |
celeron55 |
and if minetest dies, a fork will be forked |
| 17:37 |
sapier |
and what's wrong with name minetest? there's a 3d engine out there called sauerbraten |
| 17:38 |
celeron55 |
dunno, ask those supposed forkers after 3 years |
| 17:38 |
celeron55 |
unfortunately i don't have a time machine 8) |
| 17:38 |
Calinou |
the engine is Cube 2, the game is Sauerbraten |
| 17:38 |
sfan5 |
^ |
| 17:39 |
sapier |
Calinou: the other way round |
| 17:39 |
|
ImQ009 joined #minetest-dev |
| 17:40 |
sapier |
hmm ok they recently switched names |
| 17:40 |
sapier |
Do what you can't stop doing celeron55 I'm not amused about having to fix all my mods |
| 17:40 |
Calinou |
...in 2009 |
| 17:40 |
Calinou |
replace minetest. to core. using your text editor? |
| 17:40 |
celeron55 |
sapier: i haven't decided yet |
| 17:40 |
Calinou |
like I replaced minetest.env: with minetest. it worked OOTB |
| 17:41 |
* BlockMen |
still prefers engine |
| 17:41 |
celeron55 |
sapier: i will try to decide soon |
| 17:41 |
sapier |
Calinou: a simple text replace did never work and it didn't work for minetest.env by that time too |
| 17:41 |
Calinou |
2 letters shorter matters a lot over time :P |
| 17:41 |
Calinou |
sapier, worked for me in throwing, carts, maptools |
| 17:41 |
BlockMen |
engine<minetest :P |
| 17:41 |
BlockMen |
thats enough |
| 17:42 |
sapier |
if you're that short on memory you probably should update your drive |
| 17:42 |
sapier |
celeron55: well to me it seems to be decided you just want to be able to tell "everyone agreed" later ;-) |
| 17:42 |
celeron55 |
no i don't |
| 17:43 |
Calinou |
BlockMen, we shouldn't settle for less |
| 17:43 |
Calinou |
core is 4 chars, minetest is 8 |
| 17:44 |
celeron55 |
i'm trying to dig up good reasonings from people but people tend to stick to repeating some thing again and again so it isn't very helpful |
| 17:44 |
sapier |
hmm guess I'm gonna add a loopup file in case of name change so all minetest stuff is done in a single file an I don't have to lookup those locations for next years renaming |
| 17:44 |
BlockMen |
Calinou, and engine is 6 chars.. |
| 17:44 |
BlockMen |
the golden middle |
| 17:44 |
sapier |
well celeron55 why should ppl do different to you do ;) |
| 17:45 |
sapier |
why don't we rename to e? .. I'd suggest c but abc and ijk are often used for index values |
| 17:46 |
sfan5 |
><sapier> why don't we rename to e? |
| 17:46 |
sfan5 |
I was about to propose the same |
| 17:46 |
celeron55 |
because it's too short and people like to use it for eg. a variable containing an entity |
| 17:46 |
sapier |
because it's even shorter then core ... 75% to be precise ... and e is as non saying as core |
| 17:46 |
sapier |
ok what about h |
| 17:47 |
sapier |
to close to k ... |
| 17:47 |
sfan5 |
or even better: what about modifying Lua so ".place_node(pos, {name='default:stone'})" accesses the api? </mostly joke> |
| 17:47 |
sfan5 |
that would be 0 chars |
| 17:48 |
BlockMen |
sfan, place_node has to be shorten too |
| 17:48 |
BlockMen |
so "._pn() |
| 17:48 |
sapier |
let's stop this discussion noone is giving new reasons |
| 17:48 |
sfan5 |
soon: .p_n(p, {n='d:s', p1=4}) |
| 17:48 |
rubenwardy |
^ excellent |
| 17:48 |
* BlockMen |
prefers that too |
| 17:48 |
rubenwardy |
Code golf for the win! |
| 17:49 |
rubenwardy |
That is an idea... |
| 17:49 |
sapier |
we've got some other controversial things to discuss, e.g. transmission of client os to server |
| 17:49 |
sapier |
any opinions to this? |
| 17:49 |
BlockMen |
or builtin vs default (HUD) |
| 17:49 |
sfan5 |
I was about to ask celeron55 for his opionion on that(sending client os to server) |
| 17:49 |
ShadowNinja |
sapier: I approve that. |
| 17:50 |
ShadowNinja |
+of |
| 17:50 |
sapier |
I'm against it ... even more then I'm against renaming |
| 17:50 |
sapier |
to be honest doing this would make me even think about stopping to support minetest |
| 17:50 |
sapier |
that much against am I against doing things like that :-) |
| 17:51 |
celeron55 |
i think it's a stupid idea unless someone tells a valid reason other than spying people |
| 17:51 |
sapier |
I don't want minetest to tell anything about my machine it is not required to tell |
| 17:52 |
Anchakor |
sapier: you can use E instead of e as lua is case sensitive |
| 17:52 |
celeron55 |
(i also think that if minetest really sticks to this kind of privacy standard, maybe we should advertise it a bit more) |
| 17:52 |
celeron55 |
(it's a feature, really) |
| 17:53 |
sapier |
Is it become that normal for applications to spy on user this has become a thing worth advertising? |
| 17:53 |
ShadowNinja |
Yes. |
| 17:53 |
Anchakor |
it's usually opt-in or opt-out |
| 17:53 |
sapier |
seems I'm getting old |
| 17:54 |
sfan5 |
opt-in makes no sense because nobody will enable it |
| 17:54 |
Calinou |
<celeron55> (i also think that if minetest really sticks to this kind of privacy standard, maybe we should advertise it a bit more) |
| 17:54 |
Calinou |
<celeron55> (it's a feature, really) |
| 17:54 |
Calinou |
+1 |
| 17:54 |
sapier |
opt out is for ppl doing non legit things only |
| 17:54 |
Calinou |
sfan5, a game telling an user's OS makes no sense |
| 17:54 |
Anchakor |
well I enabled the firefox spying feature because I want them to know how many addons I use that fix the unsatisfactory default firefox experience :) |
| 17:54 |
sfan5 |
sapier: we should add opt-out for telling the server the minetest version too |
| 17:55 |
sapier |
telling users os is only for spys nothing else |
| 17:55 |
sfan5 |
did you even read the discussion? |
| 17:55 |
ShadowNinja |
Calinou: Use something like c55+.+ (-.) instead of pasting. |
| 17:55 |
Calinou |
the bot recognizes it? |
| 17:55 |
sfan5 |
which bot? |
| 17:55 |
ShadowNinja |
Calinou: Yes. |
| 17:55 |
sapier |
not today sfan5, but till yesterday I haven't read a single non spy related legit reason |
| 17:56 |
ShadowNinja |
!plugin help Karma |
| 17:56 |
ShadowBot |
ShadowNinja: That plugin is loaded, but has no plugin help. |
| 17:56 |
sfan5 |
sapier: you could serve e.g. a low-res texture pack to clients running on Linux/arm (probably android) |
| 17:56 |
ShadowNinja |
:-( |
| 17:56 |
Anchakor |
sapier: what HTTP user-agent header do you use in your browsers? |
| 17:56 |
celeron55 |
sfan5: by adding to the protocol the information that the client wants a low-res texture pack |
| 17:56 |
celeron55 |
not the information that the client is running on ARM |
| 17:56 |
sapier |
well that's great sfan5 so my hell of a machine 8core 4gb tablet will get same texture pack as my old 1core 500mhz arm phone |
| 17:56 |
Calinou |
it makes no sense: it's up to the client to limit pack size |
| 17:57 |
Calinou |
(would be nice to have max_texture_size option, indeed) |
| 17:57 |
sapier |
http requests are not linked to a username Anchakor |
| 17:57 |
Calinou |
most sensible servers don't change the default texture pack; if they do, they often stick to 16× or 32× |
| 17:57 |
Calinou |
nothing to be scared of |
| 17:57 |
sfan5 |
sapier: send_client_sysinfo = false |
| 17:57 |
sapier |
unless ppl enter it but there's not a way a minetest user can't enter a username |
| 17:58 |
Anchakor |
sapier: they are linked to an IP |
| 17:58 |
sapier |
still you'll have to ask any german user to ask for permission to save that data |
| 17:58 |
sfan5 |
no you don't |
| 17:58 |
Anchakor |
and a running TCP connection |
| 17:58 |
sapier |
there's no clear legal situations about ip's in germany but for usernames there is no doubt it's personal data |
| 17:58 |
sfan5 |
>but for usernames there is no doubt it's personal data |
| 17:58 |
sfan5 |
[citation needed] |
| 17:59 |
Anchakor |
^ |
| 17:59 |
sapier |
and the ip is required in order to establisch communication while os type isn't |
| 17:59 |
sapier |
-c |
| 17:59 |
ShadowNinja |
sapier: If so we, and every website with registration, has been violating german law for a long time. |
| 17:59 |
sapier |
sfan5: do you really question if a name identifying a person is personal data? |
| 18:00 |
sfan5 |
name != nickname |
| 18:00 |
sapier |
any website requireing registration shows you some terms of service on join |
| 18:00 |
sfan5 |
if you use your real name as nickname it's not the websites fault |
| 18:00 |
ShadowNinja |
Spaces aren't even valid in player names. |
| 18:00 |
sfan5 |
it can't even detect that |
| 18:00 |
celeron55 |
so is there really no better reasons for your opinions than whether it's legal or not? |
| 18:00 |
celeron55 |
that's rather extreme |
| 18:01 |
sapier |
celeron55: so you'd add a feature not being legal for other reasons? |
| 18:01 |
Anchakor |
I don't think knowledge of OS is a big deal, it is useful for statistics |
| 18:01 |
ShadowNinja |
sapier: Then Minetest is illegal in germany. I don't plan to cater to their laws. |
| 18:01 |
celeron55 |
sapier: no |
| 18:01 |
ShadowNinja |
Anchakor: Exactly. |
| 18:01 |
sapier |
if you store data e.g. server logs might be illegal in germany yes |
| 18:02 |
sfan5 |
>a feature not being legal |
| 18:02 |
sfan5 |
[citation needed] |
| 18:02 |
ShadowNinja |
I'd like this to know, eg, how many of my users use Android. |
| 18:02 |
sfan5 |
^ +1 |
| 18:02 |
sapier |
spy |
| 18:02 |
Calinou |
do you really need it? does your life depend on it? |
| 18:02 |
Calinou |
stop having the “mesuriteX |
| 18:02 |
celeron55 |
i believe privacy laws germany are quite similar to that of finland; a nickname doesn't really count but of course it's kind of grey area because someone could use an official name |
| 18:03 |
celeron55 |
+in |
| 18:03 |
sapier |
sfan5 §3a BSDG |
| 18:03 |
celeron55 |
but then again, there is no requirement of supplying real information |
| 18:03 |
celeron55 |
i think that mostly makes the difference |
| 18:04 |
sapier |
celeron55: personal data is not linked to presence of real name |
| 18:05 |
sfan5 |
sapier: so minetest is potentially illegal in germany because the user *might* use his real name as username? |
| 18:05 |
sapier |
of course we're in a grey are, but if you've got nickname os, time, ip and whatever chances this data is enough to really identify a person grow drasticaly with any additional dataset |
| 18:05 |
celeron55 |
of course if it's linked to an IP, that is almost pin-pointable to a certain person |
| 18:05 |
celeron55 |
so yeah |
| 18:06 |
sfan5 |
I don't think time & IP counts as "Einzelangaben über persönliche oder sachliche Verhältnisse einer bestimmten oder bestimmbaren natürlichen Person" |
| 18:06 |
sfan5 |
( http://dejure.org/gesetze/BDSG/3.html point 1 ) |
| 18:06 |
sapier |
how much do you think is necessary to identify a person? |
| 18:07 |
sfan5 |
an IP does not identify a person |
| 18:07 |
sapier |
may I try to find out who you are sfan5? I right now I don't have more information than any server owner has, maybe even less |
| 18:08 |
Anchakor |
then I think to be sure you should tell on the loading screen that the game is sending possibly person identifying data to the servers, because of Germany |
| 18:08 |
Anchakor |
and just send OS info anyway |
| 18:08 |
sfan5 |
sapier: no, that won't count anyway |
| 18:08 |
sapier |
well once I have the ipv6 address of your mobile phone that's almost as good as having your tax number in germany |
| 18:08 |
sfan5 |
you could take the shortcut of using github etc. |
| 18:09 |
sfan5 |
AldiTalk does not provide IPv6 |
| 18:09 |
rubenwardy |
!title https://forum.minetest.net/viewtopic.php?f=9&t=9259 |
| 18:09 |
sapier |
your luck ... but it's just a matter of time |
| 18:09 |
sfan5 |
I could configure my phone to not use IPv6 |
| 18:09 |
sfan5 |
also my phone internet goes through an APN, you would only see the IP of the APN |
| 18:09 |
sapier |
you wont get anything else except ipv6 by some not to far distant future |
| 18:10 |
sapier |
of course you can not use a mobile phone at all |
| 18:10 |
celeron55 |
so, maybe we should talk about something useful instead of the law: what information are server owners looking for, then? |
| 18:10 |
sfan5 |
-.- that is not what I wanted to say |
| 18:10 |
celeron55 |
i'd guess not exact OS version |
| 18:10 |
sapier |
that's the final argument "you can avoid it" |
| 18:10 |
sfan5 |
well |
| 18:10 |
celeron55 |
if it's so important, maybe minetest could send whether it think it's running on a tablet or not, if that's so valuable |
| 18:11 |
celeron55 |
thinks* |
| 18:11 |
sfan5 |
the linux kernel version is probably not interesting except the major and minor version |
| 18:11 |
Anchakor |
they are looking for a/s/l obviously ;) |
| 18:11 |
sfan5 |
but i'd like to know how many people still use XP |
| 18:11 |
sapier |
there's no more then major and minor version of linux kernel |
| 18:11 |
celeron55 |
yeah so and then tax number, street address and paypal password |
| 18:11 |
celeron55 |
what else will you please? |
| 18:11 |
sapier |
patch levels are no longer part of official kernel version |
| 18:11 |
sfan5 |
googlemail password please :) |
| 18:12 |
sfan5 |
>3.13.0-24-generic |
| 18:12 |
sfan5 |
I mean the 3.13 part |
| 18:12 |
sapier |
.0 is patchlevel |
| 18:13 |
|
tomasbrod joined #minetest-dev |
| 18:14 |
sapier |
If we want to add this feature, ask user on first start if he wants to submit this data and don't provide detailed information, e.g. stick to winxp,vista,win7,win8, linux 2,3, |
| 18:14 |
celeron55 |
i'm not advocating this, but what would people think if minetest explicitly asked the user whether he wants to send extra information to the server (with the warning that it might be personally identifiable but might help the server owner improve their service) at startup when a configuration file wasn't yet found? |
| 18:14 |
sfan5 |
hm |
| 18:15 |
celeron55 |
s/the server/to servers/ |
| 18:15 |
sapier |
well so doing it silently is better celeron55? |
| 18:15 |
celeron55 |
i'm advocating not sending anything |
| 18:15 |
BlockMen |
how would knowing how many xp users there are help the server owner to improve service? |
| 18:16 |
celeron55 |
but i don't really know how much people could improve the experience of others by knowing some of that information so i can suggest stuff without sticking to it |
| 18:16 |
sfan5 |
how about "Do you want to allow Minetest to send information about your system to the servers it connects to? The informations sent looks like this for you: Linux/3.13.0-24-generic x86_64" |
| 18:16 |
sapier |
Imho if we don't have courage to tell users about it we shouldn't do it at all |
| 18:17 |
Anchakor |
it could help convince some server owners that "welp, nobody plays on linux anyway, so lets ignore this bug" isn't really true |
| 18:17 |
sapier |
I'd agree to do it that way too sfan5, as long as we tell user what is sent |
| 18:17 |
sfan5 |
sapier: the sysinfo for windows system (how it is now) only tell the server owner winxp, vista, win7 (win8 is put into win7 category) and the architecture |
| 18:17 |
sfan5 |
+s |
| 18:17 |
celeron55 |
regular gamers are OK with this amount of spying: http://store.steampowered.com/hwsurvey/ |
| 18:18 |
sapier |
celeron55: do they have a joice? |
| 18:18 |
sapier |
choice? |
| 18:18 |
celeron55 |
apparently "Participation in the survey is optional, and anonymous" |
| 18:18 |
celeron55 |
i don't know how they ask it though |
| 18:19 |
sfan5 |
they ask you nicely and allow you to look at whats send |
| 18:19 |
celeron55 |
coould be opt out, or a yes/no pop-up dialog every month or something |
| 18:19 |
sfan5 |
(can confirm because I use steam) |
| 18:19 |
sapier |
really? I don't use steam for exactly that reason I don't wanna install a trojan on my system |
| 18:19 |
celeron55 |
so it's a yes/no dialog |
| 18:19 |
sfan5 |
I don't remember exactly, but yes, probably |
| 18:19 |
celeron55 |
also: http://steamcharts.com/ |
| 18:20 |
celeron55 |
those aren't disableable i think |
| 18:20 |
celeron55 |
(because it's part of the DRM) |
| 18:20 |
sapier |
I don't think steam is a good example do we wanna be evil just because someone else is already evil? |
| 18:21 |
ShadowNinja |
Lua 5.2 support: http://ix.io/c9r/diff |
| 18:21 |
sfan5 |
ShadowNinja: add depracation warnings to unpack & package.loaders |
| 18:21 |
BlockMen |
Anchakor, server owner should update minetest recently so such behavior wouldt really be an issue (if i understood you correct) |
| 18:21 |
celeron55 |
sapier: i don't think we want to be evil, or at least i don't |
| 18:21 |
sfan5 |
and all the other stuff |
| 18:22 |
ShadowNinja |
I'll have to add compatability with LuaJit's 5.1 C API. |
| 18:22 |
* sfan5 |
did not intend "evil stuff" when making that pull request |
| 18:23 |
celeron55 |
i definitely won't use a minetest client that sends unnecessary information to servers |
| 18:23 |
sapier |
it's not about what you intended sfan5, I believe you don't want to be but the mechanisms you add are more likely used for evil reasons then for good ones |
| 18:23 |
sfan5 |
if evil = trying to get malware on the clients computer |
| 18:23 |
sapier |
I wont use that client either |
| 18:23 |
sfan5 |
we've dicussed that before |
| 18:24 |
sfan5 |
if someone is still using XP opening a browser is way riskier than opening minetest |
| 18:25 |
sapier |
dancing on highway is riskier too ... but is as related to this as using xp |
| 18:25 |
sfan5 |
2nd 'opening' -> playing && ... + on servers |
| 18:25 |
celeron55 |
what if server owners would just add a pop-up formspec if they want to know what their players use |
| 18:25 |
celeron55 |
"what os are you using" and a few buttons |
| 18:26 |
celeron55 |
not a big deal and very explicit |
| 18:26 |
sfan5 |
what if players don't know which os they are using? |
| 18:26 |
celeron55 |
add a button for it and assume android |
| 18:26 |
sfan5 |
I think that is a litte too annoying |
| 18:27 |
celeron55 |
there should be a cost in spying |
| 18:27 |
sfan5 |
imagine quickly joining a few servers and each one asks you what OS you are using first |
| 18:27 |
sfan5 |
how is that spying? |
| 18:28 |
sfan5 |
Pretty much every company is a spy by that definition |
| 18:28 |
Anchakor |
on a side note I really dislike the annoying "we are using cookies" (and there is nothing you can do) website notifications |
| 18:29 |
celeron55 |
yes, they are absolutely useless |
| 18:29 |
celeron55 |
and they only do harm to those who switch cookies off (because then they see the message always) |
| 18:29 |
celeron55 |
it's stupid |
| 18:30 |
BlockMen |
sfan5, i still dont get what do you want to do with the information? |
| 18:30 |
BlockMen |
*except statistics |
| 18:31 |
sapier |
sfan5 options "windows, linux, android, apple, other" |
| 18:31 |
sapier |
sfan5: why? you don't have to ask user everytime, just e.g. once a month |
| 18:31 |
sapier |
ohh add another option "don't wanna tell you" |
| 18:31 |
sapier |
If you wanna know what os your players use play with open cards |
| 18:31 |
sapier |
it's up to you when to ask a player |
| 18:31 |
sapier |
you don't have to ask him on first login |
| 18:31 |
sapier |
e.g. you could ask on 10th too |
| 18:31 |
sapier |
maybe that data is even better as you'd only get data from regular players |
| 18:31 |
sapier |
#1265 I'd like to merge that fix, any opinions to it? |
| 18:31 |
ShadowBot |
https://github.com/minetest/minetest/issues/1265 -- Fix heart + bubble bar size on different texture packs by sapier |
| 18:32 |
BlockMen |
https://github.com/minetest/minetest/pull/1265#issuecomment-42461347 sapier |
| 18:32 |
sfan5 |
sapier: what about quickly joining 10 servers you've never been to before? |
| 18:32 |
sapier |
I know your opinion BlockMen but I don't agree to it ;) |
| 18:33 |
sapier |
yes sfan5 what's about it? I don't get the question by now? |
| 18:33 |
BlockMen |
so explain me why we should run eventhandler two times (when there is a changed hud), sapier |
| 18:33 |
sfan5 |
sapier: you would get asked 10 times |
| 18:33 |
BlockMen |
and why we need the useless get_flags_ function? |
| 18:34 |
sapier |
we need it to be compatible to current mods anyway |
| 18:34 |
BlockMen |
no? have no get_flags currently? |
| 18:35 |
sapier |
but you need it to be able to evaluate the values set by set_flags |
| 18:35 |
sapier |
which is available right now |
| 18:35 |
BlockMen |
if there is no c++ statbar anymore you cant show or hide it either |
| 18:35 |
sapier |
you can |
| 18:36 |
sapier |
the current code does behave exactly as before (related to this feature) |
| 18:36 |
BlockMen |
so its better to keep the bad way instead to improve things? |
| 18:36 |
sapier |
we can mark it to be deprecated and drop in next version but for time beeing we need to be compatible |
| 18:37 |
sapier |
n-1 compatibility is a must |
| 18:37 |
BlockMen |
well, fine, keep the flags |
| 18:37 |
BlockMen |
but there is still a pro when moving it to default |
| 18:37 |
sapier |
so if there's an agreement to drop support for disabling statbar by set_hud_flags I'm gonna mark it deprecated and remove it as soon as next version is released |
| 18:38 |
BlockMen |
sapier, no. flags are also used for crosshair, hudbar, etc |
| 18:38 |
sapier |
well in mid term those should be replaced by luahud too |
| 18:38 |
BlockMen |
wut? no way |
| 18:39 |
sapier |
why not? luahud already supports those features |
| 18:39 |
BlockMen |
useless traffic |
| 18:39 |
BlockMen |
there is no need to move everything to lua |
| 18:39 |
sapier |
do you really want to complain about 100 bytes? :-) |
| 18:40 |
BlockMen |
i think to prevent lags the border for the current wielded item is send kinda often |
| 18:40 |
sapier |
well I don't intend to do anything there the next time so we don't need to discuss this now |
| 18:40 |
sapier |
well wielded item is a bad example as this is the one not supported by luahud |
| 18:41 |
BlockMen |
ok, then you just need to convince me why we need to event handlers and we are done |
| 18:41 |
sapier |
but crosshair and hotbar could be |
| 18:41 |
BlockMen |
*s/to/two |
| 18:41 |
sapier |
two? |
| 18:41 |
|
proller joined #minetest-dev |
| 18:41 |
sapier |
whe need the changed to update if those statbars are show |
| 18:42 |
BlockMen |
if you have an costum lua hud you add another handler |
| 18:42 |
sapier |
and we need health and breath to not have to lookup both everytime |
| 18:42 |
BlockMen |
but since you have just the flags you have still the other handler |
| 18:42 |
sapier |
custom lua hud? |
| 18:42 |
sapier |
tell me more about it |
| 18:42 |
BlockMen |
a mod that modifies the statbars? |
| 18:43 |
sapier |
mods need something to cause them to do something |
| 18:43 |
BlockMen |
if it is in default you can just override the ids and use the already running handler |
| 18:43 |
sapier |
isn't that handler called on each step? |
| 18:43 |
celeron55 |
sapier: you need to document the protocol change in clientserver.h |
| 18:43 |
celeron55 |
looks like world_pos is missing from there too |
| 18:43 |
sapier |
ok I'm gonna add this |
| 18:44 |
celeron55 |
i don't have time for studying this properly but the basic idea seems fine to me |
| 18:44 |
celeron55 |
i hope that it doesn't contain problems 8) |
| 18:45 |
sapier |
BlockMen: I'm looking for this callback can you tell me where to find it? |
| 18:45 |
sapier |
well celeron55 it's been reason for major conflicts I assume if it did have any reasonable bugs I'd already been punched to fix them |
| 18:45 |
BlockMen |
sapier, hm? |
| 18:46 |
sapier |
that handler in default I assumed to be called on_step? |
| 18:47 |
sapier |
I'd like to verify my assumption |
| 18:47 |
BlockMen |
there is nothing called on_step. it is using the player events aswell |
| 18:47 |
sapier |
ok tell me where to find it if I don't need to a the new callback mechanisn I'm gonna remove it gladly |
| 18:48 |
BlockMen |
sapier, i would be fine with it if you make the id's overwriteable |
| 18:48 |
BlockMen |
then this issue woulndt occur |
| 18:48 |
sapier |
I don't know what you're talking about now if you don't tell me where to look at |
| 18:49 |
BlockMen |
sapier, https://github.com/minetest/minetest/pull/1265/files#diff-a5826972d46496671f7266e00d6e31ecR24 |
| 18:49 |
BlockMen |
^remove the local and get a useable name |
| 18:49 |
BlockMen |
s/get/find |
| 18:50 |
sapier |
well I did this this exactly to not risk it to be messed up |
| 18:50 |
BlockMen |
what should be messed up? mods can hide statbars already |
| 18:50 |
BlockMen |
so what could the make "bad" with the ids? |
| 18:50 |
BlockMen |
*they |
| 18:51 |
sapier |
yes but if this is available they can place anything there and other mods can't rely on it |
| 18:51 |
BlockMen |
same with flags? |
| 18:51 |
BlockMen |
other mods cant know what a mod does with the flags |
| 18:51 |
sapier |
and how is our own handling supposed to handle those things if player replaxces e.g. a statbar by a image? |
| 18:52 |
sapier |
ok not player but another mod |
| 18:52 |
|
iqualfragile joined #minetest-dev |
| 18:53 |
sapier |
our callbacks are registred and we rely on those Id's to be what they are supposed to be |
| 18:53 |
BlockMen |
it cant. but you cant prevent mods misbehave |
| 18:53 |
BlockMen |
mods can always do "bad things" |
| 18:53 |
sapier |
I don't see the benefit by now |
| 18:53 |
BlockMen |
the can destroy your whole world |
| 18:54 |
BlockMen |
sapier, that you dont need to add another handler that does excatly the same |
| 18:54 |
BlockMen |
while the builtin still runs and does in fact nothing |
| 18:54 |
BlockMen |
thats just stupid |
| 18:54 |
sapier |
no it doesn't run if flags disable it |
| 18:55 |
|
werwerwer joined #minetest-dev |
| 18:55 |
BlockMen |
ofc it does. it just not updates the bar |
| 19:14 |
|
Amaz_temp joined #minetest-dev |
| 19:14 |
|
Amaz_temp left #minetest-dev |
| 19:17 |
|
Amaz_ joined #minetest-dev |
| 19:24 |
|
w00tc0d3 joined #minetest-dev |
| 19:32 |
|
EvergreenTree joined #minetest-dev |
| 19:36 |
ShadowNinja |
Uhh... I think `git submodule` just deleted all or my branches and staged changes. |
| 19:36 |
ShadowNinja |
of* |
| 19:36 |
ShadowNinja |
And my remotes. |
| 19:37 |
ShadowNinja |
*whew* no, it didn't. |
| 19:40 |
sapier |
can someone please test #1265 again? |
| 19:40 |
ShadowBot |
https://github.com/minetest/minetest/issues/1265 -- Fix heart + bubble bar size on different texture packs by sapier |
| 19:46 |
|
Calinou joined #minetest-dev |
| 19:52 |
|
kaeza joined #minetest-dev |
| 19:58 |
ShadowNinja |
Adds Lua 5.2 support: http://ix.io/c9F/diff It's based on my builtin-organization branch. |
| 20:04 |
sapier |
ShadowNinja: basically I'm positive about it, do we have any issues with builtin lua versions of distributions? |
| 20:04 |
VanessaE |
whoa whoa whoa... I thought there was a HUGE insistence on avoiding even touching 5.1 (let alone 5.2)? |
| 20:04 |
VanessaE |
or whatever the version was outside of what we're using now |
| 20:05 |
kaeza |
we use 5.1 ... |
| 20:05 |
VanessaE |
ok then 5.2 |
| 20:05 |
VanessaE |
you get my point :P |
| 20:05 |
VanessaE |
when did that change? |
| 20:05 |
sapier |
well if there is a problem with lua versions of distributions we're blocked |
| 20:05 |
VanessaE |
(that 5.2 was to be considered acceptable) |
| 20:05 |
sapier |
and what about luajit? |
| 20:06 |
sapier |
is it 5.1 or 5.2? |
| 20:06 |
ShadowNinja |
I talked about it before, and I think some prople didn't like the new goto feature. But it's got other advantages. |
| 20:06 |
VanessaE |
LuaJIT is just-north-of-5.1-but-not-quite-5.2 I guess |
| 20:06 |
ShadowNinja |
sapier: It's C API is 5.1-like. But it supports many Lua 5.2 features. |
| 20:06 |
ShadowNinja |
^ |
| 20:07 |
sapier |
I still don't like the goto feature but as long as we keep it out of minetest core as well as minetest_game |
| 20:07 |
ShadowNinja |
Just about the only major thing it's missing from 5.2 is _ENV. |
| 20:07 |
sapier |
I'm against calling goto a reason to add it |
| 20:07 |
sapier |
it's more a reason NOT to add it ... but it's obviously only one new thing in lua |
| 20:07 |
sapier |
5.2 |
| 20:07 |
ShadowNinja |
sapier: goto isn't my reason. It has bit32 and a dozen other features. |
| 20:08 |
VanessaE |
sapier: well it's not like you HAVE to use every feature it has. |
| 20:08 |
kaeza |
plus 5.3 is almost out, so ... |
| 20:08 |
sapier |
that's what I meant with " ...as long as we keep it out of..." |
| 20:08 |
VanessaE |
right. |
| 20:08 |
sapier |
still ... what about distribution compatibility |
| 20:09 |
VanessaE |
(why couldn't I initially remember the version number? *feels stupid*) |
| 20:09 |
ShadowNinja |
LuaJIT has one or two Lua 5.2 features too. |
| 20:09 |
sapier |
lua 5.2 is of no use if luajit or major distributions are in conflict to it's features |
| 20:09 |
ShadowNinja |
5.3* |
| 20:09 |
VanessaE |
sapier: solve that the way it always was - bundle it? |
| 20:09 |
ShadowNinja |
sapier: LuaJIT supports most of Lua 5.2's features. |
| 20:10 |
sapier |
then we need to add a check for lua 5.1 and disallow it's system usage |
| 20:10 |
VanessaE |
right. Isn't there some way to "fall back" to interpreted Lua for when JIT misses a feature? |
| 20:10 |
sapier |
most of? |
| 20:10 |
ShadowNinja |
VanessaE: Nope. |
| 20:10 |
sapier |
I tend to "most of" isn't enough but I'd prefere to know exactly what is missing |
| 20:10 |
kaeza |
does MT use any strictly-5.1 stuff? |
| 20:11 |
kaeza |
i.e. anything not available in 5.2? |
| 20:11 |
sapier |
it's not about mt kaeza can you tell for sure no mod uses 5.1 stuff? |
| 20:11 |
VanessaE |
ShadowNinja: christ in a cartoon, 23000 lines for that patch??????? |
| 20:11 |
ShadowNinja |
LuaJIT has features that Lua doesn't have and vice versa. We'll just have to use the subset that both support, or use "if jit then" or the like around interpreter-specific code. |
| 20:12 |
ShadowNinja |
VanessaE: About 99% is Lua, since we include the whole thing. |
| 20:12 |
kaeza |
sapier, only "advanced" thing I've seen used by mods is setmetatable() |
| 20:12 |
VanessaE |
good point |
| 20:12 |
sapier |
for what I know luajit does support full set of lua5.1 functions am I wrong about this? |
| 20:12 |
sapier |
can you tell FOR SURE kaeza? |
| 20:12 |
ShadowNinja |
sapier: No, LuaJIT fully supports 5.1. |
| 20:12 |
VanessaE |
ShadowNinja: it just makes it look like you practically had to rewrite the whole engine to do it is all |
| 20:13 |
sapier |
ok ShadowNinja then we need to find out what would be missing in luajit for 5.2 |
| 20:13 |
kaeza |
sapier, how many big, complex mods are out there? maybe technic, MOBF, and IRC |
| 20:13 |
ShadowNinja |
sapier: I went over the official incompatabilities list and added patches to builtin/compat/52compat.lua or src/script/common/c_internal.(cpp/h) |
| 20:13 |
VanessaE |
sapier: the only thing LuaJIT doesn't support in 5.1 is anything where there's simply a temporary bug in its implementation like with anything else. otherwise it just works out of the box. |
| 20:13 |
sapier |
if no mod can use 5.2 features because of luajit lacks those features I don't see any use in updating lua |
| 20:14 |
sapier |
bugs don't count as "not support" |
| 20:14 |
ShadowNinja |
This might break MT_IRC beacuse it uses a shared library. |
| 20:15 |
VanessaE |
sapier: unless they persist in a "won't fix" state for months and months :P but honestly I think it would be better to stick with 5.1 and import whatever 5.1-compatible libraries and such that we feel modders (or engine devs) would benefit from |
| 20:15 |
kaeza |
I believe luasocket/LuaIRC already support 5.2 |
| 20:15 |
sapier |
without luajit support this might be of almost no use :-( |
| 20:15 |
ShadowNinja |
sapier: LuaJIT doesn't fully support 5.2, but it mostly supports it. |
| 20:15 |
kaeza |
so only mt_irc core would need to support it, if there's anything to change (require() maybe) |
| 20:15 |
sapier |
you can't use the new functions if they break a mod on luajit and if you can't use it I don't know what they're good for |
| 20:16 |
ShadowNinja |
kaeza: Yes, I know that they do. I use them for my Lua IRC bot. |
| 20:16 |
sapier |
according to luajit webpage luajit is 5.1 api + BitOp + FFI (whatever ffi is) |
| 20:16 |
ShadowNinja |
kaeza: The only thing to change would be to compile a Lua 5.2 SO and use it when needed. But maybe we should use a system lib. |
| 20:17 |
ShadowNinja |
sapier: http://luajit.org/extensions.html "Extensions from Lua 5.2." |
| 20:17 |
sapier |
"the only thing ... to compile" ... well guys ... I think a lot of users may not believe this to be a easy task ;) |
| 20:18 |
ShadowNinja |
sapier: MT_IRC is probably the most advanced Minetest mod, feature-use wise. |
| 20:18 |
kaeza |
also, any reason not to use a Lua DLL on Windows? |
| 20:18 |
ShadowNinja |
(Due to it's compiled library requirement) |
| 20:19 |
kaeza |
I believe this would also fix the mt_irc crashes on Win |
| 20:19 |
sapier |
"Other features are only enabled, if LuaJIT is built with -DLUAJIT_ENABLE_LUA52COMPAT:" so we need a special version of luajit? |
| 20:19 |
ShadowNinja |
kaeza: No? |
| 20:19 |
kaeza |
ShadowNinja, ? |
| 20:19 |
ShadowNinja |
sapier: You should probably assume that those features aren't available. |
| 20:20 |
ShadowNinja |
kaeza: Doesn't it do that already? |
| 20:20 |
sapier |
well there are almost the most interesting things in there |
| 20:21 |
sapier |
actually ... there's only one feature I'd like to have ... break ;-) but that'd not be available |
| 20:21 |
kaeza |
^ and continue |
| 20:22 |
kaeza |
wait... break is available... |
| 20:22 |
sapier |
no only with compat switch enabled |
| 20:24 |
sapier |
I'm generally positive about 5.2 but I think it's to early to do that switch by now. I think best time would be once debian enables that compat define |
| 20:24 |
ShadowNinja |
sapier: Break is available, you can just use it anywhere in Lua 5.2. |
| 20:24 |
sapier |
no I can't use it |
| 20:24 |
sapier |
luajit doesn't enable it without additional compat define |
| 20:25 |
|
SoniEx2 joined #minetest-dev |
| 20:25 |
ShadowNinja |
kaeza: Continue isn't supported, but continue-like-goto can be done. |
| 20:25 |
sapier |
see your link |
| 20:25 |
ShadowNinja |
sapier: Re-read it. |
| 20:25 |
sapier |
Other features are only enabled, if LuaJIT is built with -DLUAJIT_ENABLE_LUA52COMPAT: |
| 20:25 |
sapier |
goto is a keyword and not a valid variable name anymore. |
| 20:25 |
sapier |
break can be placed anywhere. Empty statements (;;) are allowed. |
| 20:25 |
SoniEx2 |
does minetest change the print function? |
| 20:26 |
sapier |
break isn't available on luajit |
| 20:26 |
ShadowNinja |
"break CAN BE PLACED ANYWHERE." It's still AVAILABLE with or without that setting. |
| 20:26 |
ShadowNinja |
sapier: break is a basic Lua 5.1 feature. |
| 20:27 |
ShadowNinja |
SoniEx2: Yes, it redirects it to minetest.debug. |
| 20:27 |
SoniEx2 |
ShadowNinja, do you tostring? |
| 20:27 |
sapier |
you tell me break is available in lua 5.1? argh |
| 20:27 |
SoniEx2 |
I mean this is what your print function should look like: print = function(s,...) local t={n=select('#',...},...} io.stdout:write(tostring(s)) for i=1,t.n do io.stdout:write("\t"..tostring(t[i])) end io.stdout:write("\n") end |
| 20:28 |
SoniEx2 |
(just replace io.stdout and make a local copy of select) |
| 20:28 |
kaeza |
SoniEx2, https://github.com/minetest/minetest/blob/master/builtin/builtin.lua#L9 |
| 20:28 |
ShadowNinja |
SoniEx2: Yep. https://github.com/minetest/minetest/blob/master/src/script/lua_api/l_util.cpp#L32-L58 |
| 20:28 |
kaeza |
wee.. wrong link |
| 20:29 |
sapier |
seems to ... ok so break isn't a reason ... well actually that's not a positive reason for 5.2 as then I don't have any feature in 5.2 I'd need ;-) |
| 20:29 |
SoniEx2 |
ugh that additional branching... |
| 20:29 |
SoniEx2 |
(line 50) |
| 20:30 |
ShadowNinja |
SoniEx2: How else would you do it? |
| 20:30 |
sapier |
ShadowNinja: I'd wait with 5.2 I don't see a pressing need for it atm |
| 20:30 |
SoniEx2 |
ShadowNinja, look at my Lua code, skip the first part (io.stdout:write(tostring(s))) if s is nil |
| 20:31 |
ShadowNinja |
sapier: Do you see a reason not to add it? Because I don't want to rebase this in three months. |
| 20:31 |
|
Mimilus joined #minetest-dev |
| 20:31 |
ShadowNinja |
SoniEx2: I see. Patches welcome. ;-) |
| 20:32 |
SoniEx2 |
ShadowNinja, then you don't have to compare every loop iteration and don't branch on the first thingy... |
| 20:32 |
ShadowNinja |
It isn't a very big optimization, and it will probably lead to code duplication for the tostring call. |
| 20:32 |
Mimilus |
plop |
| 20:32 |
sapier |
Usually switching to a new api version causes various compatibility issues you don't see right from beginning, comparing this to lack of real need for new features makes me suggest not to merge it |
| 20:32 |
Mimilus |
local one = 1 , local two = 0, if .... then two = one doesn't work ??? but why |
| 20:33 |
ShadowNinja |
Even a simple branch predictor should do well with that branch. |
| 20:33 |
SoniEx2 |
ShadowNinja, then do it in Lua? |
| 20:33 |
sapier |
btw we still don't merge things because of not having a reason against it but because of HAVING A GOOD REASON to merge it |
| 20:34 |
ShadowNinja |
SoniEx2: As I said, patches welcome. ;-) |
| 20:34 |
sapier |
talking about merge I'm merging #1265 in a few minutes |
| 20:34 |
ShadowBot |
https://github.com/minetest/minetest/issues/1265 -- Fix heart + bubble bar size on different texture packs by sapier |
| 20:35 |
sapier |
Mimilus: can you paste your full code to something like pastebin? |
| 20:35 |
Mimilus |
how must i do to assign the value of a local varible to another variable |
| 20:36 |
sapier |
post your code I can't help you on showing a incomplete sniplet |
| 20:37 |
sapier |
most likely you did something wrong about variable scopes but I'm just guessing without further knowledge about your code |
| 20:37 |
ShadowNinja |
celeron55: Thoughts on a Lua 5.2 upgrade? |
| 20:37 |
Mimilus |
sapier, http://pastebin.com/MnVZ8x4E |
| 20:37 |
SoniEx2 |
ShadowNinja, do you have a bootloader or something? |
| 20:38 |
celeron55 |
ShadowNinja: what's the benefit? |
| 20:38 |
ShadowNinja |
SoniEx2: Yes... |
| 20:38 |
celeron55 |
ShadowNinja: luajit is mostly lua 5.1 |
| 20:38 |
celeron55 |
so you can't use lua 5.2 features anyway |
| 20:38 |
sapier |
and what happens Mimilus? |
| 20:38 |
ShadowNinja |
celeron55: http://luajit.org/extensions.html "Extensions from Lua 5.2" These features can be used, +bit32. |
| 20:39 |
SoniEx2 |
ShadowNinja, where? |
| 20:39 |
Mimilus |
it doesn't work :) |
| 20:39 |
ShadowNinja |
SoniEx2: /boot/ |
| 20:39 |
ShadowNinja |
SoniEx2: How is this relevant? |
| 20:39 |
sapier |
that's a little bit to simplified Mimilus ;-) |
| 20:40 |
sapier |
what do you expect it to do? |
| 20:40 |
|
kahrl joined #minetest-dev |
| 20:40 |
SoniEx2 |
ShadowNinja, uhh I mean a Lua script to setup things... |
| 20:40 |
sapier |
btw your notation of writing then in front of instruction is quite stange |
| 20:41 |
Mimilus |
sapier, the node doesn't have texture for front behind right left faces |
| 20:41 |
ShadowNinja |
SoniEx2: echo $LUA_INIT -> "" |
| 20:41 |
celeron55 |
ShadowNinja: so lua 5.2 enables those things that luajit says "can break existing code if unconditionally enabled" when compared to 5.1 |
| 20:41 |
celeron55 |
have you figured out if those are a problem for existing mods? |
| 20:42 |
sapier |
what do those prints show? |
| 20:42 |
ShadowNinja |
celeron55: No, it allows modders to use those features without worrying about Lua users not having those features available. |
| 20:42 |
Mimilus |
they have to show the value of the local variable |
| 20:42 |
Mimilus |
sapier, it was just for me |
| 20:43 |
ShadowNinja |
celeron55: I've patched incompatabilities in builtin/common/51compat.lua. |
| 20:43 |
celeron55 |
well that probably helps |
| 20:43 |
ShadowNinja |
http://ix.io/c9S/lua |
| 20:43 |
Mimilus |
sapier, the print was just for anderstand |
| 20:44 |
celeron55 |
ShadowNinja: i demand serious testing for it before it's merged |
| 20:44 |
sapier |
still celeron55 if we add it we have to define our lua api level is luajit not lua 5.2 |
| 20:44 |
ShadowNinja |
-unpack infinite recursion. ;-) |
| 20:44 |
celeron55 |
sapier: i am aware of that, that's why i'm really questioning this |
| 20:44 |
celeron55 |
it makes things messy |
| 20:44 |
celeron55 |
currently it's simple; it's just simply lua 5.1 |
| 20:44 |
|
tomreyn joined #minetest-dev |
| 20:45 |
khonkhortisan |
I'm familiar with the difference |
| 20:45 |
ShadowNinja |
celeron55: I tested it with Lua and LUAJIT (with LUA52COMPAT). |
| 20:45 |
celeron55 |
ShadowNinja: which mods? |
| 20:45 |
Mimilus |
sapier, what do you want to mean by strange for my notation, how will i have to do ? |
| 20:46 |
celeron55 |
khonkhortisan: you mean 5.1 vs. 5.2 vs. luajit? |
| 20:47 |
celeron55 |
anyway, i'm not really a fan of this and i think getting comments from more actual modders would help |
| 20:48 |
khonkhortisan |
I compiled an old version of quake and had to update a few lines for my current lua, then downgrade some for luajit because it didn't do some 5.2 things |
| 20:48 |
celeron55 |
if people want the 5.2 features that luajit can provide, it might make sense |
| 20:48 |
|
grrk-bzzt joined #minetest-dev |
| 20:48 |
celeron55 |
also how do we plan to make sure that people who develop on 5.2 don't make things that don't run on luajit and the other way around? |
| 20:49 |
celeron55 |
should we disable the things that are only on the other one? |
| 20:49 |
celeron55 |
probably yes |
| 20:50 |
ShadowNinja |
celeron55: Only basic mods. |
| 20:50 |
celeron55 |
whether minetest uses an official dialect of lua doesn't make a huge difference because the code isn't really reusable outside of it anyway, but the dialect must be clearly specified |
| 20:51 |
sapier |
I'm pushing 1265 now |
| 20:53 |
celeron55 |
(it's sad that i can't even use the word "lua version"... luajit is kind of messed up) |
| 21:09 |
VanessaE |
ShadowNinja, celeron55: test it with dreambuilder. That seems stable under 5.1 |
| 21:09 |
|
Exio4 joined #minetest-dev |
| 21:10 |
VanessaE |
celeron55: I would agree with that - only allow the subset of things that both 5.2 and LuaJIT(+bit32+whatever else) can do |
| 21:11 |
ShadowNinja |
VanessaE: You can try it. :-) I'll rebase it soon. |
| 21:15 |
sapier |
does anyone have a 3d screen to check the top/bottom or side by side 3d modes? |
| 21:17 |
VanessaE |
not I, sorry. |
| 21:18 |
sapier |
hmm maybe I can use bino to show a captured screencast sidebyside video in interlaced mode ... sounds strange but could work |
| 21:18 |
ShadowNinja |
I'd like to get #1258 merged soon as it's in constant need of rebasing. |
| 21:18 |
ShadowBot |
https://github.com/minetest/minetest/issues/1258 -- Organize builtin into subdirectories and use "core" namespace by ShadowNinja |
| 21:19 |
sapier |
there's still no official decision about core or not core shadow |
| 21:20 |
VanessaE |
[05-07 12:42] <celeron55> does someone oppose it if the "minetest" and "engine" namespaces in Lua are officially renamed to "core"? the reason for doing this is to not make forks of minetest need to support the name of every other fork, and the reason why i propose "core" is because it's short and sounds better than "engine" when combined with a dot and some function |
| 21:20 |
celeron55 |
i haven't decided it yet |
| 21:22 |
VanessaE |
my thought is "core" is okay, I routinely call it "engine core". On-disk, it's minetest_core . |
| 21:22 |
VanessaE |
(plus I see the same nomenclature used pretty much everywhere else, e.g. quassel core vs. client) |
| 21:34 |
sapier |
that strange way of recording in top/bottom as well as sidebyside and displaying with bino did really work |
| 21:37 |
sapier |
what's the opinion about #1269, I don't know why travis gcc build failed there's no usable result either |
| 21:37 |
ShadowBot |
https://github.com/minetest/minetest/issues/1269 -- Scene draw cleanup + add various 3d modes by sapier |
| 21:38 |
sapier |
and the last file compiled in that log isn't even related to what is changed |
| 21:43 |
sapier |
I'm gonna continue testing 1269 to the end of this week. Unless someone finds bugs or new issues caused by it. It's not supposed to change anything for 2d mode and adds new features only |
| 21:44 |
sapier |
please check the genericCAO header commit beeing part of it too |
| 21:47 |
|
cg72 joined #minetest-dev |
| 21:55 |
sapier |
does anyone know who crazyR14 is? ... the name seems to be program ;-) |
| 22:02 |
|
jin_xi joined #minetest-dev |
| 22:06 |
|
sapier left #minetest-dev |
| 22:07 |
proller |
again 2000+ lines changed without sense.. |
| 22:11 |
|
EvergreenTree joined #minetest-dev |
| 22:17 |
VanessaE |
why doesn't it make sense? |
| 22:17 |
VanessaE |
a lot of people have stereoscopic 3D TV's |
| 22:17 |
proller |
its maybe 100 lines, other - just spaces change |
| 22:22 |
|
crazyR joined #minetest-dev |
| 22:23 |
VanessaE |
I don't see where that's a big problem, part of the purpose of the commit is "cleanup" |
| 22:24 |
VanessaE |
remember, even just moving a few lines of code can sometimes make git act stupid about the true nature of the change and bulk it up to hundreds of lines. |
| 22:25 |
VanessaE |
there ARE some spacing changes in here but not a whole lot |
| 22:26 |
VanessaE |
and remember too that editors with decent tabulating/indenting functions can move whole blocks of text back and forth in one shot. |
| 22:26 |
VanessaE |
what to he or I is a single keystroke may become thousands of lines of changes when committed to git. |
| 22:26 |
VanessaE |
so don't bitch about it too much |
| 22:30 |
|
RealBadAngel joined #minetest-dev |
| 23:04 |
RealBadAngel |
ive made a pull with some logging tweaks |
| 23:04 |
RealBadAngel |
#1278 |
| 23:04 |
ShadowBot |
https://github.com/minetest/minetest/issues/1278 -- Map irrlicht log level to minetest. Allow write them to debug file. by RealBadAngel |
| 23:05 |
RealBadAngel |
patch allow disable both irrlicht and mt spamming the console with action messages |
| 23:06 |
VanessaE |
s/console/terminal/ |
| 23:09 |
RealBadAngel |
well, mt source uses "console" |
| 23:18 |
|
jin_xi joined #minetest-dev |
| 23:20 |
|
pi2 joined #minetest-dev |
