Time  Nick           Message
01:31 kahrl          ShadowNinja: lua_pop(L, -1) should clear the stack if I read the implementation right
01:32 kahrl          but that's probably an implementation detail, the documented way to do that is lua_settop(L, 0)
02:24 Zeno`          hi kahrl, thanks for the comments
02:24 Zeno`          I don't understand the line comment for line 453 regarding the \\\\;
02:27 ShadowNinja    kahrl: That could actually be the source of the bug then, over-poping produces obscure segfaults in Lua.  Where'd you read that?  I've only seen it as "Effectively #define lua_pop(l, n) lua_settop(l, lua_gettop(l) - n)"
02:33 Zeno`          This is what I mentioned the other night
07:38 kahrl          Zeno`: that comment (for line 453) isn't relevant anymore
07:39 Zeno`          I took it away
07:39 kahrl          ShadowNinja: it's in lua.h: #define lua_pop(L,n) lua_settop(L, -(n)-1)
07:39 Zeno`          Can't remember but I think I meant to type unescape... not sure (just took it away)
07:39 kahrl          Zeno`: ah
07:40 kahrl          ShadowNinja: so with n==-1 that evaluates to lua_settop(L, 0)
12:18 Zeno`          hmm
16:59 kahrl          sfan5: weblate 2.0 was just released, should we try it out?
17:03 sfan5          kahrl: is it easier to install?
17:03 kahrl          dunno
17:04 kahrl          haven't tried it yet, hence why I asked :)
17:05 kahrl          but it supports django 1.7 out of the box so maybe some of the old bugs have vanished
17:10 Calinou        happy trolling, sfan5
17:10 Calinou        “Does it still remove the running kernel?”
17:10 Calinou        here's a question you can ask to the developers
17:11 sfan5          what?
17:17 MinetestForFun Hi everybody, somenone can inform me about the version of the protocol of minetest between 0.4.10-stable and the latest 0.4.10-dev release ? i need to know if the protocol have changed between this two version... thank you :)
17:52 OldCoder       Here is a clue to MT network problems:
17:52 OldCoder       https://forum.minetest.net/viewtopic.php?id=5026
17:52 OldCoder       This mod causes Kindle Fires, some, to time out every time
17:53 OldCoder       But not regular PCs
17:55 PenguinDad     OldCoder: maybe because of the models?
17:56 PenguinDad     *model
17:57 OldCoder       Unusual models have different effects, then?
17:58 OldCoder       Makes sense, I guess. But lockout occurs at media load time. Is this consistent?
17:58 PenguinDad     No I was thinking about the size of the model
17:58 OldCoder       Reviewing
17:58 OldCoder       I deleted it; how large is it?
18:00 PenguinDad     About 600 kb :/
18:03 Krock          kinda big
18:04 PenguinDad     using a binary model format might fix it
18:13 OldCoder       We don't especially need it. But apparently 0.5MB models can't be loaded even with large timeouts.
18:13 OldCoder       I set timeouts to 5 seconds as an experiment; no dice
18:13 OldCoder       Thank you for the comments. This would appear to be the explanation.
18:13 * OldCoder     rests
18:55 RiZom-91       Hi there! Is there an "easy" way to make entities flammable as node please ?
18:55 RiZom-91       I've tried groups = {flammable=2} but no success
18:56 sfan5          entities can't burn
18:56 sfan5          so, no
18:58 RiZom-91       Would it be merged if someone (me?) changes this in the core code ? or is it a chosen feature ?
19:01 sfan5          pull requests are always welcome
19:01 sfan5          especially if they add new useful features
19:02 RiZom-91       ok so I will try to do something for this
19:03 proller        welcome ignoring
19:03 proller        for
19:03 Calinou        proller, solved in freeminer?
19:03 proller        yes
19:06 RiZom-91       proller? you already have this (flammable entities) in freeminer?
19:06 proller        no
19:07 RiZom-91       so I don't understand your comment ?
19:08 RiZom-91       why would it be ignored ?
19:08 sol_invictus   because everything is ignored in minetest
19:11 RiZom-91       maybe. I'll see (if and when I'll have something good to pull...).
19:30 Krock          :/ latest irrlicht seems to be incompatible with MT or just broken
19:30 Krock          (r3526)
19:33 Krock          yeah. the stuff crashes without even starting irrlicht
21:05 Krock          ShadowNinja, I can reproduce the "bug" and uploaded an image. you wouldn't see more on an animated image. it just spawns and expires
21:07 ShadowNinja    Krock: Give me a few minutes and I'll confirm it.
21:07 Krock          okay.
21:14 ShadowNinja    Krock: Confirmed.
21:17 Krock          :)
21:32 ShadowNinja    Comments?  This is a modified version of donat-b's last_login patch: http://sprunge.us/JgjM?diff
21:47 kaeza          ShadowNinja, in line 39, you try to assert that last_login is a number, but I see nowhere where you actually ensure it is a number to begin with
21:48 ShadowNinja    kaeza: Good catch, thanks.
21:48 ShadowNinja    I thought I got all of those.
21:48 kaeza          only thing I see is last_login = tonumber(last_login), but at that point, the original last_login could be nil (due to not being in auth.txt), so tonumber() returns nil
21:49 kaeza          maybe last_login = tonumber(last_login) or 0 ?
21:51 kaeza          other than that, looks good
21:52 kaeza          `/last-login` could be `/last_login` or `/lastlogin` to be consistent (think `/setpassword`)
21:58 ShadowNinja    kaeza: Yes, both of those things were as it was before but 1. 0 is still a perfectly good date, nil is a better way to indicate invalidity.  2. It's two seperate words, so they should be seperated.  I just like '-' for commands because it's allowed.
21:58 ShadowNinja    The way I want to do this I have to change string.split though ('+' -> '*' in regex).
21:59 ShadowNinja    So a:b:c::d -> {'a','b','c','','d'}
21:59 ShadowNinja    The issue of the dozen or so different ways that a split function can act is discussed at length in a ua-users wiki article.