| Time |
Nick |
Message |
| 00:02 |
SoniEx2 |
ShadowNinja, let me show you... |
| 00:03 |
SoniEx2 |
local k,o,d=next,type d=function(a,r,t)if o{}~=o(a)then return a end t={}r=r or{}r[a]=t for x,y in k,a do t[r[x]or d(x,r)]=r[y]or d(y,r)end return t end table.copy={shallow=function(a,b)b={}for x,y in k,a do b[x]=y end return b end,deep=d} |
| 00:04 |
SoniEx2 |
use the deep copy with a custom r and you can do fancy hax |
| 00:07 |
SoniEx2 |
for example, table.copy.deep(sometable, setmetatable({}, {__index = function(t,k) return k end})) does a shallow copy |
| 00:07 |
SoniEx2 |
(which can be done with table.copy.shallow) |
| 00:07 |
SoniEx2 |
well |
| 00:07 |
SoniEx2 |
sort of |
| 00:07 |
SoniEx2 |
you need a no-op __newindex too |
| 00:08 |
SoniEx2 |
but still |
| 00:08 |
SoniEx2 |
anyway, basically you combine that code with metatables and you can easily do recursive pretty print, recursive dump, etc |
| 00:09 |
SoniEx2 |
I should probably add metatable copying... |
| 00:11 |
SoniEx2 |
ShadowNinja, anyway got it? |
| 00:12 |
SoniEx2 |
plus you get a really simple (aka fast) table copy function... |
| 00:19 |
Exio4 |
implement that in haskell and it will be easier to read |
| 00:20 |
|
diemartin joined #minetest-dev |
| 00:22 |
ShadowNinja |
SoniEx2: Um, yes, that's totally the cleanest and most readable way to implement it. </sarcasm> |
| 00:24 |
ShadowNinja |
SoniEx2: Also, these are dumping functions, not copying functions. |
| 00:35 |
SoniEx2 |
ShadowNinja, why not both? |
| 00:37 |
ShadowNinja |
SoniEx2: A copy function may be included in the future, but that's irrelivant to my patch and it certainly won't be the code you pasted. |
| 00:38 |
SoniEx2 |
well my code fits IRC :3 |
| 00:43 |
|
zsoltisawesome left #minetest-dev |
| 00:48 |
SoniEx2 |
ShadowNinja, this should be human readable: https://gist.github.com/SoniEx2/fc5d3614614e4e3fe131 |
| 00:48 |
jin_xi |
ShadowNinja: http://paste.org.ru/?xwauzc ? |
| 00:50 |
SoniEx2 |
jin_xi, I looked at that if ... else if ... else if ... and thought "hmm I wonder how that would handle NULs"... then I saw you're removing it... |
| 00:53 |
jin_xi |
basically this commit https://github.com/minetest/minetest/commit/2b1eff772524fca8249fd64028e5cbfeabc127a0 is very fishy |
| 00:55 |
ShadowNinja |
SoniEx2: No, that just has some spacing. It needs sane naming and structure. |
| 00:56 |
SoniEx2 |
ShadowNinja, oh ok |
| 00:56 |
SoniEx2 |
well it has sane structure now |
| 00:56 |
SoniEx2 |
:P |
| 00:56 |
|
diemartin joined #minetest-dev |
| 00:57 |
SoniEx2 |
naming? replace "k" with "next", "o" with "type", "d" with "deep", "r" with "recursion", "a" with "in", "b" and "t" with "out" |
| 01:06 |
|
pro joined #minetest-dev |
| 01:07 |
SoniEx2 |
ShadowNinja, how about this: https://gist.github.com/SoniEx2/fc5d3614614e4e3fe131#file-tablecopy-sane-lua |
| 01:08 |
SoniEx2 |
(as you asked for a sanity I also added recursion table validation to make sure it's a table) |
| 01:09 |
ShadowNinja |
SoniEx2: Like so: https://gist.github.com/ShadowNinja/6a6c0ac7141c0812c3b2 |
| 01:10 |
SoniEx2 |
ShadowNinja, you use pairs |
| 01:10 |
SoniEx2 |
5.2 has __pairs |
| 01:11 |
SoniEx2 |
and I don't like "copies" being an internal parameter |
| 01:11 |
ShadowNinja |
SoniEx2: So we should use ugly internal functions that aren't available in out current Lua version? |
| 01:11 |
SoniEx2 |
not enough flexibility |
| 01:12 |
SoniEx2 |
uhh |
| 01:12 |
ShadowNinja |
SoniEx2: There's no external use for it. |
| 01:12 |
SoniEx2 |
"next" exists in 5.1 |
| 01:12 |
ShadowNinja |
Yes, and? |
| 01:12 |
SoniEx2 |
I have a use for it |
| 01:12 |
SoniEx2 |
5.1, 5.2, 5.3 they all have "next" |
| 01:13 |
SoniEx2 |
and please make local copies of the stuff |
| 01:13 |
SoniEx2 |
(for sanity) |
| 01:14 |
ShadowNinja |
SoniEx2: Yes, and? |
| 01:14 |
SoniEx2 |
my function supports modding and metatables properly |
| 01:15 |
ShadowNinja |
SoniEx2: Looks like it ignores metatables. |
| 01:15 |
SoniEx2 |
it's moddable |
| 01:15 |
SoniEx2 |
so it can easily support metatables |
| 01:21 |
SoniEx2 |
ShadowNinja, here's a better example: https://gist.github.com/SoniEx2/fc5d3614614e4e3fe131#file-tablecopy-sane-lua |
| 01:23 |
SoniEx2 |
(also I prefer to make a table.copy table and put the copy functions there because I feel it looks better...) |
| 01:24 |
SoniEx2 |
hmm... I forgot to make a local copy of rawset and setmetatable... |
| 01:24 |
SoniEx2 |
but still |
| 01:49 |
SoniEx2 |
ShadowNinja, so... |
| 02:21 |
VanessaE |
why does Minetestserver strip out things like /usr/lib/x86_64-linux-gnu/lua/5.1/ from lua module search path? |
| 02:21 |
ShadowNinja |
VanessaE: It doesn't? |
| 02:21 |
VanessaE |
it does. |
| 02:22 |
VanessaE |
./?.lua;/usr/local/share/luajit-2.0.2/?.lua;/usr/local/share/lua/5.1/?.lua;/usr/local/share/lua/5.1/?/init.lua;/home/minetest/.minetest/worlds/MG_World/worldmods/irc/?/init.lua;/home/minetest/.minetest/worlds/MG_World/worldmods/irc/?.lua |
| 02:22 |
ShadowNinja |
SoniEx2: So, why not pairs? |
| 02:22 |
VanessaE |
that's from earlier, before I re-built everything to be on the right search path. |
| 02:22 |
ShadowNinja |
VanessaE: AFAIK Minetest doesn't touch any search paths. |
| 02:22 |
VanessaE |
that's what I get at the start of the IRC mod. |
| 02:23 |
VanessaE |
so minetest or something it does is altering the search paths |
| 02:23 |
VanessaE |
the IRC entries aside, this ^^^ does not match what I get from lua on the command line |
| 02:23 |
SoniEx2 |
ShadowNinja, because the code is supposed to be compatible with _at least_ Lua 5.1, LuaJIT, Lua 5.2, Lua 5.3 |
| 02:24 |
VanessaE |
default search path for command line Lua on the same box: ./?.lua;/usr/local/share/lua/5.1/?.lua;/usr/local/share/lua/5.1/?/init.lua;/usr/local/lib/lua/5.1/?.lua;/usr/local/lib/lua/5.1/?/init.lua;/usr/share/lua/5.1/?.lua;/usr/share/lua/5.1/?/init.lua |
| 02:24 |
VanessaE |
./?.so;/usr/local/lib/lua/5.1/?.so;/usr/lib/x86_64-linux-gnu/lua/5.1/?.so;/usr/lib/lua/5.1/?.so;/usr/local/lib/lua/5.1/loadall.so |
| 02:24 |
VanessaE |
(and cpath) |
| 02:24 |
ShadowNinja |
SoniEx2: pairs is available in 5.1. |
| 02:24 |
SoniEx2 |
ShadowNinja, so basically, if you change the Lua version you probably won't have to change that piece of code |
| 02:24 |
SoniEx2 |
ShadowNinja, so is next |
| 02:25 |
ShadowNinja |
VanessaE: This is Minetest compiled with non-JIT Lua? |
| 02:25 |
VanessaE |
with JIT. |
| 02:25 |
SoniEx2 |
in Lua 5.1, pairs doesn't use the __pairs metamethod |
| 02:25 |
SoniEx2 |
but in Lua 5.2+, it does |
| 02:25 |
SoniEx2 |
so I can't use pairs in 5.2+ |
| 02:26 |
ShadowNinja |
SoniEx2: Why would we have to change that? With 5.2 it respects __pairs if set. |
| 02:26 |
SoniEx2 |
ShadowNinja, simply because I wanna copy all entries from a table |
| 02:26 |
ShadowNinja |
VanessaE: Then check LuaJIT's paths. |
| 02:26 |
SoniEx2 |
__pairs might combine the table and the metatable's __index together |
| 02:26 |
SoniEx2 |
and I don't want that |
| 02:26 |
VanessaE |
http://pastebin.ubuntu.com/7470915/ |
| 02:27 |
SoniEx2 |
or it might skip numbers |
| 02:27 |
SoniEx2 |
that would be bad too |
| 02:27 |
ShadowNinja |
SoniEx2: Hmmm, O.K. |
| 02:27 |
|
cheapie joined #minetest-dev |
| 02:27 |
VanessaE |
strange, a few minutes ago, the x86_64 bit was in those paths. now it isn't. |
| 02:27 |
SoniEx2 |
it's easier to copy tables in C/C++ |
| 02:28 |
SoniEx2 |
in Lua you have to use "hacks" |
| 02:28 |
VanessaE |
(well it was in the package.path anyway) |
| 02:28 |
ShadowNinja |
VanessaE: So, is the LuaJIT interpreter's search path consistent with Minetest's? |
| 02:28 |
SoniEx2 |
(altho it's more like an "exploit" or a "feature abuse") |
| 02:29 |
VanessaE |
ShadowNinja: http://pastebin.ubuntu.com/7470923/ |
| 02:29 |
VanessaE |
(also ShadowNinja, you have many /msgs waiting) |
| 02:29 |
ShadowNinja |
SoniEx2: Well, in C++ "tables" are std::map, and since C++ has reference operators it makes copying easier. |
| 02:30 |
SoniEx2 |
ShadowNinja, in C you just copy the memory and fix the pointers |
| 02:30 |
VanessaE |
ok so luajit took the x86_64 stuff out of the path. |
| 02:30 |
ShadowNinja |
VanessaE: So is it or isn't it? |
| 02:31 |
ShadowNinja |
SoniEx2: Not with linked lists and similar structures. |
| 02:31 |
VanessaE |
http://pastebin.ubuntu.com/7470932/ |
| 02:31 |
VanessaE |
ok so it looks like it IS consistent |
| 02:31 |
VanessaE |
which makes this a LuaJIT problem. |
| 02:32 |
SoniEx2 |
ShadowNinja, with linked lists you still copy memory and fix pointers |
| 02:33 |
ShadowNinja |
VanessaE: You said you compiled LuaJIT, so check luaconf.h or the docs to add the needed paths. |
| 02:33 |
SoniEx2 |
but you copy entry 1, then entry 2, then fix entry 1's pointer to entry 2, etc |
| 02:33 |
SoniEx2 |
or something like that |
| 02:33 |
SoniEx2 |
Lua tables are hash map + array |
| 02:33 |
SoniEx2 |
so basically array + array |
| 02:34 |
ShadowNinja |
SoniEx2: That isn't just a memcpy though. C++ just affers copy constructors eince it supports OOP. |
| 02:34 |
VanessaE |
ShadowNinja: checking. |
| 02:34 |
ShadowNinja |
since* |
| 02:34 |
ShadowNinja |
offers* |
| 02:34 |
SoniEx2 |
where/when did I say "you just do a memcpy", tho? |
| 02:34 |
SoniEx2 |
:P |
| 02:36 |
ShadowNinja |
"in C you just copy the memory and fix pointers" |
| 02:36 |
SoniEx2 |
I didn't say how |
| 02:37 |
ShadowNinja |
And pointers are a relatively advanced concept. But this is off-topic. |
| 02:40 |
SoniEx2 |
so anything against a built-in simple yet powerful (and moddable) table copy function? |
| 02:41 |
SoniEx2 |
other than the fact that you don't like it, that is... |
| 02:41 |
SoniEx2 |
(and the fact that it's probably not gonna get included...) |
| 02:42 |
ShadowNinja |
I'd support that. I'll finish my implementation and provide a patch later. But what about my *dump*() patch? |
| 02:42 |
SoniEx2 |
it looks better |
| 02:42 |
SoniEx2 |
I didn't go too deep into it tho |
| 02:43 |
ShadowNinja |
SoniEx2: O.K. good. Anyone else? |
| 02:43 |
ShadowNinja |
(I technically don't need approval for this, but I like to have comments for a patch li |
| 02:44 |
ShadowNinja |
longer than a few dozen lines) |
| 02:45 |
SoniEx2 |
heh I see what you did there |
| 02:45 |
SoniEx2 |
you basically did the same thing I did |
| 02:46 |
SoniEx2 |
the dump() is just like my deepcopy function |
| 02:46 |
SoniEx2 |
except you use pairs and stuff |
| 02:47 |
SoniEx2 |
I guess it's not supposed to handle metatables... |
| 02:48 |
ShadowNinja |
Um, no. It returns a human-readable representation of an object. |
| 02:48 |
ShadowNinja |
It doesn't touch metatables yet. |
| 02:49 |
SoniEx2 |
I know it does a dump and not a copy |
| 02:49 |
SoniEx2 |
but it kinda does the same I do in terms of scanning the table and stuff... |
| 02:49 |
ShadowNinja |
Unless you mean that it does the same thing that you did in your own project... |
| 02:50 |
ShadowNinja |
The dumped/copied table usage is similar, yes. |
| 02:51 |
SoniEx2 |
it just recursively calls itself with a recursion table to avoid infinite recursion |
| 02:51 |
SoniEx2 |
but it also checks for the type and returns basic_dump(in) if it's not a table |
| 02:52 |
SoniEx2 |
which's pretty much the same I do |
| 02:52 |
SoniEx2 |
in a way |
| 02:53 |
ShadowNinja |
Yes, it has it's similarities. |
| 02:53 |
SoniEx2 |
it would look kinda funny if done with a copy function and metatables... but it would work... |
| 02:54 |
SoniEx2 |
anyway, I'm tired |
| 02:54 |
SoniEx2 |
good night |
| 03:02 |
VanessaE |
ShadowNinja: had to change line 62 of luaconf.h to this: |
| 03:02 |
VanessaE |
#define LUA_CPATH_DEFAULT "./?.so;/usr/lib/x86_64-linux-gnu/lua/5.1/?.so;" LUA_LCPATH1 LUA_RCPATH LUA_LCPATH2 |
| 03:03 |
VanessaE |
that makes it work again. |
| 03:03 |
ShadowNinja |
:-) |
| 03:03 |
VanessaE |
but editing luajit's source, even for a one-line change, feels wrong. |
| 03:03 |
VanessaE |
and I see no clear way to add that extra atom from the command line |
| 03:04 |
ShadowNinja |
VanessaE: Use the -D flag to gcc. |
| 03:05 |
ShadowNinja |
And check if changing LUA_[RL]?CPATH%d? |
| 03:05 |
VanessaE |
ShadowNinja: it's unclear which of these 47 bazillion variables is actually accessible from there. |
| 03:05 |
ShadowNinja |
+ would be more appropriate. |
| 03:06 |
ShadowNinja |
I can't help much now though. Maybe tomorow. |
| 03:12 |
VanessaE |
I give up. |
| 03:12 |
VanessaE |
I'll leave it like this, at least it runs. |
| 03:24 |
|
kaeza joined #minetest-dev |
| 03:55 |
|
Megaf joined #minetest-dev |
| 04:24 |
|
crazyR joined #minetest-dev |
| 04:40 |
|
Eater4 joined #minetest-dev |
| 04:42 |
|
stormchaser3000 joined #minetest-dev |
| 04:42 |
|
Eater4 joined #minetest-dev |
| 05:11 |
|
RealBadAngel joined #minetest-dev |
| 05:12 |
RealBadAngel |
hi |
| 05:12 |
RealBadAngel |
anybody checked #1303 ? |
| 05:12 |
|
ffoxin_work joined #minetest-dev |
| 05:12 |
ShadowBot |
https://github.com/minetest/minetest/issues/1303 -- Unite nodes shaders. by RealBadAngel |
| 05:14 |
RealBadAngel |
also i want to merger this code into misc_helpers.lua |
| 05:14 |
RealBadAngel |
http://pastebin.com/xrM4nULS |
| 05:14 |
RealBadAngel |
*merge |
| 05:16 |
RealBadAngel |
handy way to check if function exists. it checks all elements of namespace for nil |
| 05:16 |
RealBadAngel |
so no crash possible |
| 05:43 |
|
SoniEx2 joined #minetest-dev |
| 06:03 |
|
grrk-bzzt joined #minetest-dev |
| 06:04 |
|
Mimilus_ joined #minetest-dev |
| 06:44 |
|
CraigyDavi_ joined #minetest-dev |
| 06:44 |
|
VargaD joined #minetest-dev |
| 06:45 |
|
darkrose joined #minetest-dev |
| 07:36 |
|
arjun joined #minetest-dev |
| 08:07 |
|
CraigyDavi joined #minetest-dev |
| 08:13 |
|
arjun joined #minetest-dev |
| 08:36 |
|
restcoser joined #minetest-dev |
| 08:49 |
|
jin_xi joined #minetest-dev |
| 09:33 |
|
ImQ009 joined #minetest-dev |
| 09:53 |
|
darkrose joined #minetest-dev |
| 09:53 |
|
pro joined #minetest-dev |
| 10:04 |
|
Jordach joined #minetest-dev |
| 10:52 |
RealBadAngel |
so, #1305 |
| 10:52 |
ShadowBot |
https://github.com/minetest/minetest/issues/1305 -- Safe way to check if function of name given exists. by RealBadAngel |
| 10:52 |
RealBadAngel |
any comments on that? |
| 11:06 |
|
proller joined #minetest-dev |
| 11:34 |
|
Megaf joined #minetest-dev |
| 11:37 |
|
PenguinDad joined #minetest-dev |
| 12:15 |
|
kahrl joined #minetest-dev |
| 12:25 |
|
PilzAdam joined #minetest-dev |
| 12:50 |
|
CraigyDavi_ joined #minetest-dev |
| 13:07 |
|
hmmmm joined #minetest-dev |
| 13:14 |
|
PenguinDad joined #minetest-dev |
| 13:27 |
|
Mimilus joined #minetest-dev |
| 13:33 |
|
ffoxin_work joined #minetest-dev |
| 13:46 |
|
catninja joined #minetest-dev |
| 13:46 |
|
lanxu joined #minetest-dev |
| 13:49 |
|
Likeyn joined #minetest-dev |
| 13:58 |
|
smoke_fumus joined #minetest-dev |
| 14:01 |
|
CraigyDavi joined #minetest-dev |
| 14:30 |
|
Piggybear87 joined #minetest-dev |
| 14:33 |
|
Mimilus joined #minetest-dev |
| 14:56 |
|
CraigyDavi joined #minetest-dev |
| 15:10 |
|
Zeitgeist_ joined #minetest-dev |
| 15:11 |
|
Exio4 joined #minetest-dev |
| 15:20 |
|
zat joined #minetest-dev |
| 15:49 |
ShadowNinja |
RealBadAngel: I use a function like resolve_name in a Lua IRC bot that I wrote, but "a" should be passable as a parameter. And it needs a more descriptive name, like "cur". |
| 15:50 |
ShadowNinja |
Also, globals are generally bad. |
| 15:51 |
ShadowNinja |
And it needs documentation. |
| 15:51 |
|
us`0gb joined #minetest-dev |
| 15:53 |
ShadowNinja |
Also: t={x={["Hi. :-)"]={f=function()end}}} print(function_exists("t.x['Hi. :-)'].f") won't work. |
| 15:54 |
ShadowNinja |
Actually, "t.x.Hi. :-).f" would almost work. The . Just needs an escape. |
| 15:56 |
ShadowNinja |
Also, the use case for function_exists seems rare. And it's simple to do if resolve_name is available. |
| 15:57 |
ShadowNinja |
Normally I'd use > if t and t.x and t.x["Hi. :-)"] and t.x["Hi. :-)"].f then <...> end |
| 16:00 |
|
PenguinDad joined #minetest-dev |
| 16:01 |
|
grrk-bzzt joined #minetest-dev |
| 16:01 |
ShadowNinja |
Much faster than parsing a string too. |
| 16:04 |
SoniEx2 |
function_exists(something) = if x then if x.y then if x.y.z then <stuff> end end end? |
| 16:04 |
SoniEx2 |
(and type(x.y.z)=="function"?) |
| 16:06 |
ShadowNinja |
SoniEx2: Yes, but Lua does proper short-circuiting, so you only need one if statement. |
| 16:18 |
|
CheapSeth joined #minetest-dev |
| 16:30 |
|
sapier joined #minetest-dev |
| 16:31 |
SoniEx2 |
ShadowNinja, uhh what do you mean? |
| 16:31 |
dsimon |
if x and x.y and x.y.z |
| 16:31 |
SoniEx2 |
oh that |
| 16:32 |
dsimon |
lua will not try to evaluate anything to the right of an 'and' if the left is false |
| 16:32 |
SoniEx2 |
yeah I forgot about that |
| 16:32 |
SoniEx2 |
.-. |
| 16:32 |
dsimon |
whoa, how did you put your mouth above your eyes? |
| 16:32 |
SoniEx2 |
why does it need a function then? |
| 16:32 |
dsimon |
that's a neat trick :-) |
| 16:33 |
dsimon |
though i'd worry about my face getting stuck |
| 16:33 |
dsimon |
hm, but the ideal would be if we could get a ruby-like 'try' function |
| 16:34 |
SoniEx2 |
we have pcall |
| 16:34 |
sapier |
http://pastebin.com/xrM4nULS RBA what exactly does this do? |
| 16:35 |
SoniEx2 |
select(2,pcall("return type(x.y.z.a.b.c.etc) == "function" and x.y.z.a.b.c.etc")) |
| 16:35 |
SoniEx2 |
uhh wait |
| 16:35 |
SoniEx2 |
that would return false, errmsg on error... |
| 16:35 |
SoniEx2 |
select(3,pcall("return nil, type(x.y.z.a.b.c.etc) == "function" and x.y.z.a.b.c.etc")) |
| 16:35 |
SoniEx2 |
:3 |
| 16:35 |
SoniEx2 |
(cheap hax ftw) |
| 16:36 |
SoniEx2 |
nil on error, or the function if it exists |
| 16:36 |
sapier |
RealBadAngel: hope it's not possible to get any function within lua stack that easy |
| 16:39 |
SoniEx2 |
uhh wait that shouldn't be a string... .-. |
| 16:39 |
SoniEx2 |
here: https://github.com/minetest/minetest/pull/1305#issuecomment-43352559 |
| 16:42 |
sapier |
still SoniEx2 noone did answer xyzz's question by now |
| 16:42 |
|
Megaf joined #minetest-dev |
| 16:43 |
SoniEx2 |
sapier, how about groovy? |
| 16:43 |
sapier |
not a valid reason for a merge ;-P |
| 16:43 |
SoniEx2 |
http://groovy.codehaus.org/Null+Object+Pattern |
| 16:44 |
SoniEx2 |
let me quote it |
| 16:44 |
SoniEx2 |
"Secondly, with the ?. operator, an expression like p?.job?.salary will be equal to null if salary is equal to null, or if job is equal to null or if p is equal to null. You don't need to code a complex nested if ... then ... else to avoid a NullPointerException." |
| 16:44 |
SoniEx2 |
uhh that was longer than I expected... |
| 16:44 |
sapier |
I don't want a theory lesson but a practical example |
| 16:44 |
SoniEx2 |
well |
| 16:44 |
sapier |
theory is good but most time quite useless |
| 16:45 |
SoniEx2 |
you don't need an if then else chain |
| 16:45 |
sapier |
I can't tell how much code I've seen to theoreticaly work but fail terribly for it's real usecase |
| 16:45 |
SoniEx2 |
or if x and x.y and x.y.z and x.y.z.a and x.y.z.a.b and x.y.z.a.b.c and x.y.z.a.b.c.etc then ... end |
| 16:46 |
|
Jordach joined #minetest-dev |
| 16:46 |
sapier |
still for what I read this code I'd not be sure if a function halled 123456789 wouldn't make function_exists(456) tell true |
| 16:47 |
SoniEx2 |
yeah it should use pcall and loadstring |
| 16:48 |
sapier |
why? |
| 16:48 |
SoniEx2 |
because pcall and loadstring would support x[y] and stuff |
| 16:48 |
sapier |
still it's called "function_exists" not "tryecute" |
| 16:48 |
sapier |
+x |
| 16:48 |
VanessaE |
RBA's idea arose from a sanity check I had to add to plants_lib yesterday |
| 16:49 |
SoniEx2 |
the code only supports x.y.z.etc |
| 16:49 |
sapier |
ok what do you check vanessae maybe I understand better once I know what this is used for |
| 16:49 |
SoniEx2 |
you can't use objects as indexes |
| 16:49 |
VanessaE |
he thought a more generic solution to the problem I had to solve might be a good idea for the future. |
| 16:50 |
SoniEx2 |
my code lets you do x.y.z.etc, x[y][z][etc], x(y).z.etc |
| 16:50 |
SoniEx2 |
and all that |
| 16:51 |
VanessaE |
sapier: well in plants_lib, in one of its API calls, you can pass a field that contains either a node name, a function name, or a table (L-system), but the program has to check if that string points to a valid function if it doesn't appear to be a node and isn't a table |
| 16:51 |
|
ImQ009 joined #minetest-dev |
| 16:51 |
sapier |
well to me "resolve_name" means "get ip address for dns name" ... and function_exists(name) I'd expect me to tell if a function called exactly name exists ... but that'd be way more easy by if name ~= nil and type(name) == "function" then |
| 16:51 |
VanessaE |
the function name is passed as a string containing the name |
| 16:52 |
sapier |
why do you pass it as string? |
| 16:52 |
VanessaE |
I eventually solved it in a simpler way: https://github.com/VanessaE/plantlife/blob/master/plants_lib/init.lua#L186 |
| 16:52 |
SoniEx2 |
VanessaE, select(3,pcall(loadstring("return nil, type(x.y.z.a.b.c.etc) == \"function\" and x.y.z.a.b.c.etc end"))) |
| 16:52 |
VanessaE |
sapier: old code. |
| 16:52 |
VanessaE |
sapier: when it was written, I didn't know how to properly pass actual functions back and forth, and now that part of the API is sorta set in stone. |
| 16:53 |
sapier |
that's really hacky ;-) why don't you clean it up ? ;-) |
| 16:53 |
VanessaE |
SoniEx2: not too different from that. |
| 16:53 |
VanessaE |
sapier: you wanna rewrite the other mods that use that feature? :) |
| 16:53 |
dsimon |
you could make it type-sensitive for backwards compatibility |
| 16:53 |
VanessaE |
dsimon: I could, but I'm lazy :P |
| 16:53 |
sapier |
I'd add a big fat warning and remove it in about half a year |
| 16:54 |
SoniEx2 |
^ |
| 16:54 |
VanessaE |
anyway, the above is how I solved it, similar to SoniEx2's idea. |
| 16:54 |
VanessaE |
RBA's idea might be useful in some rare corner cases too |
| 16:55 |
sapier |
well I'm not against the feature although I believe the intention to be wrong, but the naming of those functions is quitr missleading to me |
| 16:55 |
SoniEx2 |
VanessaE, your thing does too many loadstrings... |
| 16:55 |
VanessaE |
SoniEx2: I know. |
| 16:56 |
SoniEx2 |
so mine is better :P |
| 16:56 |
SoniEx2 |
(well almost) |
| 16:56 |
|
rubenwardy joined #minetest-dev |
| 16:58 |
VanessaE |
the sanity check, fwiw, came about in a weird way: |
| 16:58 |
VanessaE |
the check for a *node name* failed in one case because the node being called for was undefined |
| 16:58 |
VanessaE |
so it fell through with the assumption that it's a function :P |
| 17:11 |
|
cerulean256 joined #minetest-dev |
| 17:22 |
|
Calinou joined #minetest-dev |
| 17:31 |
RealBadAngel |
hey |
| 17:32 |
VanessaE |
hi |
| 17:33 |
RealBadAngel |
to get why the code checks namespace part by part try to run: print(type(minetest1.swap_node)) |
| 17:36 |
RealBadAngel |
and idea for that is easy way to configure your mods, just by checking if something exists, which is kinda opposite to publish features |
| 17:36 |
sapier |
RealBadAngel: I'm to lazy to do that now I guess what you wanna tell is it does really look for an exact match? |
| 17:36 |
RealBadAngel |
it checks each part of a.b.c.d.etc for nil |
| 17:37 |
RealBadAngel |
so it wont crash |
| 17:37 |
RealBadAngel |
example above will crash |
| 17:38 |
RealBadAngel |
ofc each mod could write a hax to check for functions, but i wanted a generic way, provided by engine |
| 17:38 |
sapier |
well while that feature could be usefull for some mods I'd see this more in some sort of support library then in builtin ... hmm we don't have something like that maybe we should create one |
| 17:38 |
RealBadAngel |
could be usefull? its necesary |
| 17:38 |
sapier |
especially as I expect "some" to be more "a few" then "a lot" |
| 17:39 |
sapier |
we've got exactly one usecase by now and I'm not even sure this is a really valid one |
| 17:39 |
RealBadAngel |
i can then easily configure my mods to work with all the engine versions out there |
| 17:39 |
sapier |
no you can't |
| 17:39 |
sapier |
well you shouldn't |
| 17:40 |
sapier |
basically for same reason once the feature flags have been added ... ok noone uses them true |
| 17:40 |
RealBadAngel |
thats example |
| 17:41 |
RealBadAngel |
another example, im using mesecons. but i want specific function from it and i want to be sure that is aviable |
| 17:42 |
sapier |
well anyway as function_exists seems to do what it's called it might be fine, but I suggest asking some ppl what they'd expect resolve_name to do because I'd expect it to do something completely different |
| 17:42 |
|
Anchakor joined #minetest-dev |
| 17:43 |
RealBadAngel |
resolve name just checks a then b then c until last part, function name |
| 17:43 |
RealBadAngel |
and btw, parameter is function_name, not example usage of function |
| 17:43 |
sapier |
well "resolve_name" is somehow connected to dns names |
| 17:44 |
VanessaE |
test_function_name ? |
| 17:44 |
sapier |
at least to me, that's why I suggest asking other people what they'd expect a function called this way to do (without telling them what it does first) |
| 17:45 |
RealBadAngel |
it should be made local |
| 17:45 |
RealBadAngel |
its in fact internal thingy for function_exists |
| 17:45 |
sapier |
in this case the name would be almost irrelevant |
| 17:46 |
sapier |
but as ppl tend to remove local from functions they consider usefull I'd still prefere it to have a more specific name. But of course that's a personal opinion |
| 17:47 |
RealBadAngel |
any ideas for name? |
| 17:47 |
VanessaE |
[05-16 13:44] <VanessaE> test_function_name ? |
| 17:48 |
sapier |
#1241 if noone knows about problems this adds I'm gonna merge it tonight (in about 4-5h) |
| 17:48 |
ShadowBot |
https://github.com/minetest/minetest/issues/1241 -- Add formspec toolkit and refactor mainmenu to use it by sapier |
| 18:01 |
|
nore joined #minetest-dev |
| 18:03 |
|
CraigyDavi_ joined #minetest-dev |
| 18:07 |
|
CraigyDavi__ joined #minetest-dev |
| 18:08 |
|
khonkhortisan joined #minetest-dev |
| 18:08 |
VanessaE |
~tell sapier [05-16 14:06] <Jordach> http://www.theregister.co.uk/2014/05/16/boffins_run_ios_apps_on_android_hardware/ |
| 18:08 |
ShadowBot |
VanessaE: O.K. |
| 18:09 |
sfan5 |
VanessaE: how is this relevant for android stuff? |
| 18:10 |
|
CraigyDavi joined #minetest-dev |
| 18:12 |
VanessaE |
sfan5: it may be helpful in developing an iOS port of minetest, as sapier lacks any iOS devices, as I recall. |
| 18:12 |
sfan5 |
I have an iOS device |
| 18:12 |
VanessaE |
oh ok, |
| 18:13 |
VanessaE |
well since he's been kinda at the forefront of it, that's why I figured he'd be interested. |
| 18:13 |
sfan5 |
ah |
| 18:14 |
sfan5 |
VanessaE: I even did some iOS stuff, but the non-portable way: https://github.com/sfan5/minetest/tree/ios |
| 18:16 |
VanessaE |
non-portable? O_o |
| 18:16 |
VanessaE |
(given the target platform, that comment confuses me a bit :) ) |
| 18:16 |
sfan5 |
s/portable/cmake/ |
| 18:16 |
VanessaE |
ah |
| 18:36 |
|
EvergreenTree joined #minetest-dev |
| 18:38 |
|
nore joined #minetest-dev |
| 18:51 |
RealBadAngel |
one handy thing for testing (waiting for comments) |
| 18:51 |
RealBadAngel |
http://pastebin.com/6tiLfaSu |
| 18:52 |
RealBadAngel |
make a mod out of it (paste it into empty init.lua) |
| 18:53 |
RealBadAngel |
it replaces built it item behaviour |
| 18:53 |
RealBadAngel |
so items on the ground do stack themselves when hit the ground |
| 18:54 |
VanessaE |
^^^^^ I tested the above, it works quite nicely. |
| 18:56 |
RealBadAngel |
*built in |
| 18:57 |
|
Jordach_ joined #minetest-dev |
| 19:13 |
|
khonkhortisan joined #minetest-dev |
| 19:17 |
RealBadAngel |
anybody else tested it? |
| 20:01 |
PenguinDad |
RealBadAngel: animated water is broken with #1303 http://www.directupload.net/file/d/3624/x6nz7iss_png.htm |
| 20:01 |
ShadowBot |
https://github.com/minetest/minetest/issues/1303 -- Unite nodes shaders. by RealBadAngel |
| 20:01 |
|
kaeza joined #minetest-dev |
| 20:07 |
|
tomreyn joined #minetest-dev |
| 20:17 |
cerulean256 |
quick question.. I am trying to use the on_punch callback and would like to access the original entity from within the call back... |
| 20:17 |
cerulean256 |
obj:get_luaentity().on_punch = function(self, puncher, time_from_last_punch, tool_capabilities, dir) blah blah blah |
| 20:18 |
cerulean256 |
maybe I am misunderstanding.. but isn't the self parameter a reference back to the luaentity? |
| 20:19 |
cerulean256 |
it's described in a little more detail here: http://dev.minetest.net/LuaEntitySAO |
| 20:20 |
cerulean256 |
an example would help a lot :) |
| 20:21 |
PenguinDad |
cerulean256: write obj:get_luaentity():on_punch = function(self, puncher, time_from_last_punch, tool_capabilities, dir) --some code end |
| 20:22 |
PenguinDad |
s/write/try/ |
| 20:22 |
cerulean256 |
only difference being the ':' |
| 20:23 |
cerulean256 |
ok.. I'll see if that's the cause.. now you mention it... it makes sense |
| 20:23 |
kaeza |
wat |
| 20:24 |
kaeza |
PenguinDad, that's invalid syntax (you can only use ':' to *call* functions) |
| 20:25 |
kaeza |
like obj:func(), but you cannot do obj:field = foo |
| 20:25 |
cerulean256 |
ok.. that didn't work |
| 20:25 |
PenguinDad |
kaeza: https://github.com/PilzAdam/carts/blob/master/init.lua#L24 |
| 20:25 |
cerulean256 |
using the . works |
| 20:26 |
cerulean256 |
I just made a test to send a chat message when I punch my entity |
| 20:26 |
cerulean256 |
but I want to be able to manipulate the entity that is being punched.. so I need access to the luaentity from within the punch func |
| 20:27 |
cerulean256 |
and all I get are nils |
| 20:27 |
kaeza |
PenguinDad, function foo:bar(...) is syntactic sugar for foo.bar = function(self, ...), so my comment still applies |
| 20:29 |
kaeza |
cerulean256, can you pastebin a little bit of code? |
| 20:29 |
cerulean256 |
sure |
| 20:29 |
kaeza |
also, can we take this to #minetest-mods / #minetest ? |
| 20:30 |
cerulean256 |
sure |
| 20:37 |
|
werwerwer_ joined #minetest-dev |
| 20:43 |
|
sapier joined #minetest-dev |
| 20:48 |
|
tomreyn joined #minetest-dev |
| 20:50 |
SoniEx2 |
a set node event should be added |
| 20:51 |
|
alexxs joined #minetest-dev |
| 20:51 |
SoniEx2 |
and place node, break node, etc |
| 20:52 |
kaeza |
there's register_on_(dig|place)node |
| 20:58 |
|
paramat joined #minetest-dev |
| 21:00 |
SoniEx2 |
kaeza, what about worldgen etc? |
| 21:01 |
kaeza |
for that, on_generated/ABMs |
| 21:07 |
|
Jordach_ joined #minetest-dev |
| 21:08 |
sapier |
ShadowNinja: you tapped in the "kill" trap ;-) |
| 21:18 |
SoniEx2 |
kaeza, so register_on_generated? |
| 21:18 |
kaeza |
yep |
| 21:18 |
SoniEx2 |
how do you use it? |
| 21:19 |
kaeza |
ask paramat :) |
| 21:19 |
SoniEx2 |
uhh wait... |
| 21:20 |
sapier |
I'm gonna merge 1241 in about half an hour if there are no issues about those things I changed |
| 21:20 |
|
Jordach joined #minetest-dev |
| 21:28 |
VanessaE |
kaeza: he's talking about some kind of callback that could be used to detect when something called by on_generated adds a node to the map |
| 21:31 |
kaeza |
adding such a callback would undermine the huge speeds you get from VManips, or make Lua-based generation even slower in the general case |
| 21:31 |
kaeza |
or dunno |
| 21:31 |
sapier |
that'd be basically a first grade performance killer |
| 21:31 |
sapier |
any changed node no matter who changes it will call that callback |
| 21:32 |
sapier |
abms switching nodes, lua mapgens ... and and and |
| 21:34 |
|
blaise joined #minetest-dev |
| 21:35 |
kaeza |
SoniEx2, what is your goal? i.e. what do you need the callback for? |
| 21:39 |
paramat |
celeron55, please can we have an option to make "air" do drowning damage? im currently using 'airlike' vacuum nodes which have problems |
| 21:43 |
sapier |
pushing #1241 now |
| 21:43 |
ShadowBot |
https://github.com/minetest/minetest/issues/1241 -- Add formspec toolkit and refactor mainmenu to use it by sapier |
| 21:43 |
SoniEx2 |
kaeza, ask VanessaE :P |
| 21:43 |
SoniEx2 |
VanessaE, I'm a they .-. |
| 21:45 |
VanessaE |
ehm, ok |
| 21:47 |
VanessaE |
kaeza: the gaol, or rather the theory, is that plants_lib could be sped up somewhat if I could evaluate the mapblock for surfaces just once, and then just populate the result, instead of having to run through the entire evaluator for every single node type that's being added ( https://github.com/VanessaE/plantlife/blob/master/plants_lib/init.lua#L119 ) |
| 21:48 |
VanessaE |
problem is, I can scan for surfaces, but it's impossible to know that, for example, a big tree was grown over here, or a chunk of the map was just erased over there by one of the mods that hooks into my mapgen code |
| 21:48 |
VanessaE |
so I have no choice but to run through that entire search_for_surfaces() call (I know the name is misleading) for every place that hooks in. |
| 21:49 |
VanessaE |
for Dreambuilder, I counted 120 such hooks, but I am not sure how many are duplicates (I'd guess about half) |
| 21:49 |
VanessaE |
s/gaol/goal/ |
| 21:53 |
sapier |
vanessae could you check #1297 is fixed for you too? |
| 21:53 |
ShadowBot |
https://github.com/minetest/minetest/issues/1297 -- Public serverlist failure |
| 21:54 |
VanessaE |
sapier: it's fixed, but check the Settings menu npow |
| 21:54 |
VanessaE |
now* |
| 21:54 |
VanessaE |
grey backgrounds behind the options, is this intentional? |
| 21:54 |
VanessaE |
also, btw, the brown background is fixed as well |
| 21:55 |
VanessaE |
http://digitalaudioconcepts.com/vanessa/hobbies/minetest/screenshots/Screenshot%20-%2005162014%20-%2005%3a55%3a36%20PM.png |
| 21:55 |
VanessaE |
is this intentional? |
| 21:56 |
sapier |
yes the grey background is intentionally |
| 21:56 |
VanessaE |
ok, but do note how the words overflow it. |
| 21:57 |
VanessaE |
this looks... eh.. well not good :P |
| 21:57 |
sapier |
yes that's caused by your font |
| 21:57 |
VanessaE |
that's not "my font" :P that's just my having turned the default size up so mybad eyes can read it |
| 21:57 |
sapier |
the texts overflow follow up checkbox too so this isn't a issue of the boxes but a generic issue about text lengths |
| 21:58 |
VanessaE |
right. may I suggest some sort of clipping be added? or better, abbreviate the text when it overflows |
| 21:58 |
sapier |
well worldcraft added a feature to make formspecs scale related to fontsize |
| 21:58 |
sapier |
that'd be usefull for hdpi screens too |
| 21:58 |
VanessaE |
Preload Item visuals --> Preload ite... |
| 21:59 |
VanessaE |
could be useful perhaps, yeah |
| 22:00 |
VanessaE |
though my screens' DPI is about average, 96 I guess. I just have bad eyes. |
| 22:00 |
VanessaE |
anyway, it all seems to work fine now. |
| 22:00 |
sapier |
well high dpi screens are another reason where that feature is usefull ... additional to bad eyes ;-) |
| 22:01 |
sapier |
it's rare but sometimes you can fix two issues with one feature ;-) |
| 22:08 |
VanessaE |
heh |
| 22:14 |
|
Megaf joined #minetest-dev |
| 22:22 |
|
grrk-bzzt joined #minetest-dev |
| 22:28 |
SoniEx2 |
RealBadAngel, 1. https://github.com/minetest/minetest/pull/1305#issuecomment-43383141 2. your code doesn't like t[x] and stuff |
| 22:35 |
|
stormchaser3000 joined #minetest-dev |
| 22:35 |
|
SmugLeaf joined #minetest-dev |
| 22:35 |
|
robmyers joined #minetest-dev |
| 22:35 |
|
VanessaE joined #minetest-dev |
| 22:35 |
|
book` joined #minetest-dev |
| 22:35 |
|
hax404 joined #minetest-dev |
| 22:38 |
|
ImQ009 joined #minetest-dev |
| 22:56 |
|
ImQ009 joined #minetest-dev |
| 22:58 |
cerulean256 |
I'm running into an issue... |
| 22:59 |
cerulean256 |
I just built minetest from latest git and started with a clean configuration. |
| 22:59 |
sapier |
what happened? |
| 23:00 |
cerulean256 |
when I enable the public server list in the clients tab... I get a lua error on line 911 |
| 23:00 |
sapier |
you did clone right now true? |
| 23:00 |
cerulean256 |
'attempt to get length of field 'favorites' (a nil value) |
| 23:00 |
cerulean256 |
yes |
| 23:01 |
sapier |
can you post the full message? |
| 23:01 |
cerulean256 |
0.4.9-228-g34d8726 |
| 23:01 |
cerulean256 |
sure. |
| 23:02 |
sapier |
wait |
| 23:02 |
cerulean256 |
ERROR: /usr/share/minetest/builtin/mainmenu/init.lus:911: attempt to get length of field 'favorites' (a nil value) |
| 23:02 |
sapier |
g34? ... you missed the latest commit which fixes that issue |
| 23:02 |
cerulean256 |
ah.. ok |
| 23:02 |
cerulean256 |
will refetch |
| 23:03 |
cerulean256 |
ok.. compiling ... again |
| 23:04 |
cerulean256 |
looking in that file it seems someone should make up their mind on how to spell the word favorite |
| 23:04 |
cerulean256 |
either british or american english... but mixing is bound to screw something up down the road ;) |
| 23:05 |
sapier |
well that's most likely me ... once english and us ppl agree to a single word us non native ones may get it right too ;-) |
| 23:06 |
cerulean256 |
I'd try to shoot for consistency... that's all |
| 23:06 |
cerulean256 |
I use american english... but if the original author used british.. I would adjust |
| 23:07 |
sapier |
hmm I don't know about any englishmen in here but a lot of us ppl but on the other hand usually you learn british english in other countrys |
| 23:07 |
cerulean256 |
sure, but you get my point right? |
| 23:07 |
sapier |
but for the main reason I agree to consistency should be the goal yes |
| 23:08 |
cerulean256 |
and this one can be corrected easily with a regex. |
| 23:08 |
sapier |
well last time I used a regex to replace a thing I got punched for making minetest to selfish ;-) |
| 23:09 |
sapier |
but I wont stop you from doing it and create a pull request for sure ;-) |
| 23:09 |
cerulean256 |
that's ridiculous... if it achieves consistency it's a service and not selfish at all. |
| 23:10 |
sapier |
well cerulean256 "selfish" isn't meant the way it's usually meant in this case ;-) |
| 23:10 |
sapier |
small explanation ... I had to replace this by self ... ;-) |
| 23:10 |
* cerulean256 |
thinks sapier has been punching too many nodes |
| 23:11 |
cerulean256 |
:) |
| 23:11 |
cerulean256 |
ok.. my build is done |
| 23:11 |
cerulean256 |
lets see if this is really fixed |
| 23:11 |
PenguinDad |
cerulean256: "self program is distributed" in the license for example |
| 23:12 |
sapier |
btw I hope someone did crosscheck I didn't miss to revert any of the accidental replacements ;-) |
| 23:14 |
cerulean256 |
another observation.... when I am in the server tab and typing in credentials to login to a server.. I usually press tab to go to the next field |
| 23:14 |
cerulean256 |
in this case... it exits minetest |
| 23:15 |
sapier |
hmm couldn't you have told this about an hour before? ;-) |
| 23:15 |
cerulean256 |
:p |
| 23:15 |
PenguinDad |
did anyone else notice that gui_scaling doesn't work in mainmenu anymore? |
| 23:15 |
sapier |
no because it's not supposed to work in mainmenu ;-P |
| 23:16 |
PenguinDad |
it worked earlier :/ |
| 23:16 |
|
blaise joined #minetest-dev |
| 23:16 |
sapier |
what exactly do you want to scale? |
| 23:17 |
VanessaE |
<PenguinDad> the whole menu? |
| 23:17 |
VanessaE |
;) |
| 23:18 |
sapier |
well define "earilier" ;-) because it doesn't scale for the pre cleanup version too |
| 23:19 |
cerulean256 |
well one thing is for sure.. the background stretching works... |
| 23:20 |
sapier |
that one is supposed to be stretched but not the menu itself |
| 23:20 |
cerulean256 |
i wonder if you can force it to maintain the aspect ratio. |
| 23:20 |
sapier |
not yet |
| 23:21 |
sapier |
buttons as well as text wouldn't scale only the empty room around ... that's not really usefull |
| 23:21 |
cerulean256 |
agreed |
| 23:22 |
cerulean256 |
besides..I'd rather see the background artwork of mods instead |
| 23:22 |
sapier |
of which mod? there could be plenty of mods |
| 23:23 |
cerulean256 |
I meant games |
| 23:23 |
sapier |
they are shown |
| 23:23 |
cerulean256 |
i know.. and I like that :) |
| 23:23 |
sapier |
ohhh :-) now I understand |
