Minetest logo

IRC log for #minetest-dev, 2013-06-12

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

All times shown according to UTC.

Time Nick Message
01:14 kahrl joined #minetest-dev
01:40 NakedFury joined #minetest-dev
02:06 ssieb joined #minetest-dev
04:40 neko259 joined #minetest-dev
04:45 kahrl Exio: I tested it with a variety of keyboard layouts in gentoo linux and it seems to work (I only seem to have layouts in the basic multilingual plane though)
04:46 kahrl of course some scripts can't be displayed because the font is missing those characters
04:46 kahrl I made it print the numeric character code to dstream to test those, they work
04:47 kahrl How about BSD (for example), has it been tested there?
05:01 ch98 joined #minetest-dev
05:07 * khonkhortisan shows up early
05:53 Anchakor joined #minetest-dev
06:18 darkrose joined #minetest-dev
07:00 NakedFury joined #minetest-dev
07:04 dexter0 joined #minetest-dev
07:20 VanessaE joined #minetest-dev
07:27 celeron55 joined #minetest-dev
07:27 Calinou joined #minetest-dev
07:39 celeron55 joined #minetest-dev
07:53 celeron55 joined #minetest-dev
08:12 VanessaE joined #minetest-dev
08:22 celeron55 joined #minetest-dev
08:31 thexyz Exio: sadly, this is ucs-2, not utf-8
08:31 Calinou joined #minetest-dev
08:57 darkrose joined #minetest-dev
08:59 Anchakor_ joined #minetest-dev
09:06 kahrl thexyz: what exactly is ucs-2?
09:08 thexyz kahrl: it's fixed 16-bit encoding
09:09 thexyz which obviously cannot provide the full set of unicode symbols
09:09 kahrl oops, I know what ucs-2 is
09:09 kahrl I was trying to ask what you meant by "this"
09:09 thexyz well, he said >what's up with adding the UTF8 support to the console on linux?
09:10 thexyz which is incorrect
09:10 thexyz anyway, that patch works for me as well
09:10 kahrl wouldn't it be ucs-4?
09:10 kahrl or utf-32 since those are the same
09:13 thexyz wouldn't what be ucs-4?
09:13 thexyz wchar_t is 2 bytes on windows afaik
09:13 kahrl oh, well wchar_t is 4 bytes on linux (where that patch has any effect)
09:14 kahrl but wstrings are serialized into ucs-2, you're right
09:16 kahrl so you can probably type and see symbols outside the basic multilingual plane, but they will come out as junk after you submit them
09:17 kahrl you won't actually see them because the font doesn't have them afaik
09:27 kahrl <rant>wchar_t being 2 bytes on windows is another one of them things that make unicode programming horrible and not fun at all
09:27 kahrl unless you write your own wstring class, that basically forces you to treat wstrings as either utf-16 or ucs-16 or add platform specific code
09:27 kahrl ucs-2*
09:28 kahrl and utf-16 has the disadvantages of both utf-8 and utf-32 combined and none of the advantages
09:28 kahrl and of course you still have to treat wstring as utf-32 on linux, adding platform specific code anyway
09:28 kahrl </rant>
09:45 zweipktfkt joined #minetest-dev
10:18 Zeg9 joined #minetest-dev
10:36 Exio kahrl: it works (without patch or "workaround") for hmmmm
10:36 Exio and, aw, didn't know
10:36 Exio i just said utf8 because that same workaround is used in the "patched-for-freetype" file :P
10:43 Exio but after all, the only thing what it does is get the same behavior that the actual "chat windows" has in the console
10:50 Exio also, shouldn't #763 be a "bugfix"?
10:57 PilzAdam joined #minetest-dev
10:57 Exio http://irrlicht.sourceforge.net/docu/structirr_1_1_s_irrlicht_creation_parameters.html#ac790f1359a357f705bc2a5b24a6cc55d
10:58 Exio "only relevant for DirectX"
11:12 khonkhortisan joined #minetest-dev
11:36 iqualfragile joined #minetest-dev
12:34 EdB joined #minetest-dev
13:03 salamanderrake joined #minetest-dev
13:50 Anchakor_ joined #minetest-dev
14:08 smoke_fumus joined #minetest-dev
14:10 serengeor joined #minetest-dev
15:07 Jordach joined #minetest-dev
15:07 Jordach joined #minetest-dev
15:31 rubenwardy joined #minetest-dev
15:48 BlockMen joined #minetest-dev
15:48 hmmmm joined #minetest-dev
15:51 NakedFury joined #minetest-dev
16:15 Calinou joined #minetest-dev
16:21 sapier joined #minetest-dev
16:26 Jordach joined #minetest-dev
16:43 OWNSyouAll_DESKT joined #minetest-dev
16:49 ImQ009 joined #minetest-dev
16:52 Anchakor joined #minetest-dev
16:52 neko259 joined #minetest-dev
17:40 rubenwardy joined #minetest-dev
18:00 Taoki Think I found a solution to hardware lighting... in theory at least
18:00 Taoki http://forum.minetest.net/viewtopic.php?pid=94508#p94508
18:01 Calinou Minetest not MineTest
18:01 Taoki Ahh, I did that again :/
18:12 Taoki Anyway, from my understanding the code should be pretty easy, and the shader easy enough. For someone who knows coding enough that is. Thought maybe I can get something done, if someone can at least do the C++ part
18:13 Taoki Namely create a MineTest preferences that enables a new shader, sends the param1's of all nodes to it, sends light sources, and disables the existing vertex color lighting.
18:13 PilzAdam Taoki, speak to RBA before doing anything, he is working on shaders and hardware lighting too; just to make sure that your things dont conflict
18:14 Taoki PilzAdam: Sure thing. Hard to think I'd get anything done by myself as it is :P And great to hear he is, I'll speak to him about it :)
18:17 sapier did anyone test the current formspec menu?
19:02 sapier guys the menu needs testing its already a 15k changes commit and it's not like scriptapi changes only copy n paste
19:04 thexyz okay, what to test?
19:16 sapier everything ;-)
19:17 sapier 1) does it doe everything exactly as old one
19:17 sapier 2) does modmanager work in lokal mode (install mod via zip
19:17 sapier 3) does mmdb integration work
19:20 thexyz sapier: https://gist.github.com/xyzz/755a7d8a739b3823f66f
19:20 sapier what os?
19:21 thexyz gentoo x86_64
19:22 sapier what?
19:23 thexyz hm?
19:23 sapier I'm on wheezy x86_64
19:23 sapier ok much older gcc I suppose what does gentoo use atm?
19:23 thexyz https://gist.github.com/xyzz/18fd18da42dfd9ece63e
19:24 thexyz 4.6.3 in +stable, 4.7.3 in ~unstable
19:24 sapier that's no explanation at all debian uses 4.7.2
19:24 sapier did you run cmake before?
19:25 thexyz yes, I deleted CMakeCache.txt and rerun cmake
19:26 sapier I wonder whats so special neither travis nor my windows env did show any problem like that
19:27 thexyz http://sourceforge.net/p/trrntzip/bugs/3/ probably related
19:28 sapier seems to be the bug I wonder why this only happens on gentoo ;-)
19:29 kahrl https://bugs.gentoo.org/show_bug.cgi?id=383179
19:30 sapier ok so actualy a bug was fixed ;-) a wonder how often compatibility macros need to be added  ;-)
19:32 sapier ok pushed the fix
19:37 thexyz okay, it compiled
19:38 thexyz it segfaults after clicking on pretty much anything
19:39 kahrl the dynamic_cast problem?
19:40 thexyz yes
19:40 kahrl Ok, so there is a correlation to running it under gentoo
19:41 thexyz ?
19:41 kahrl I had (have) the same segfaults and I'm using gentoo too
19:42 kahrl Haven't had time to investigate yet
19:42 thexyz oh
19:43 sapier great ... I guess my oppinion about gentoo beeing alpha software isnt only  a prejudice
19:44 sapier works on ubuntu 12.04/debian wheezy/ winxp
19:47 sapier I'd ask for your help to find out what's wrong on gentoo
19:56 MJaoune joined #minetest-dev
20:44 iqualfragile_ joined #minetest-dev
20:54 kahrl ok so about the mesh update thread thing
20:54 kahrl anyone have any ideas?
20:54 kahrl the only place the mesh update thread is started is in afterContentReceived
20:55 kahrl and then after that a TOCLIENT_MEDIA message is received and processed
20:55 sapier what's wrong there?
20:55 kahrl on login, after receiving media: http://pastebin.com/4ctZUC2v
20:56 kahrl note that the line numbers are a bit outdated
20:59 sapier client is started prior data is completely received so wherever this exception happens that code should be able to handle that
21:02 kahrl so the server somehow sends more media than the client requested, or does something else happen?
21:02 kahrl I mean the client should not crash when the server does that, but fixing it would just hide the symptoms
21:05 sapier does this happen on first login too or only on relogin?
21:07 kahrl it happened to me on a first login to VanessaE's server
21:07 kahrl wasn't able to reproduce it after that, even after clearing the cache
21:07 sapier is game and mesh tread at same prio level?
21:07 kahrl that shouldn't matter at all
21:07 VanessaE that happens to *everyone*
21:08 PilzAdam I had that on VanessaE's server with a cache that only contained minetest_game media, too
21:09 kahrl actually I'm not sure it was a first login
21:09 kahrl I might have logged on there months before that... don't remember
21:10 kahrl of course the selection of mods changed in between these logins if there were multiple
21:10 sapier there's a problem in jthread
21:10 VanessaE kahrl: indeed, the server'
21:10 VanessaE server's set of mods has evolved considerably over the past months
21:11 sapier is this bug recreatable?
21:11 kahrl sapier, not really
21:11 kahrl but it happens often enough
21:11 VanessaE not reliably no
21:12 sapier maybe we should add traces that tell when a thread is created
21:12 PilzAdam sapier, it happened to me when adding a new sound, but when retrying with the same cache it worked
21:13 sapier jthread.cpp line 132 is wrong
21:13 sapier but i guess thats unrelated
21:14 kahrl what about client.cpp:846
21:14 kahrl it sends a media request to the server but doesn't increase m_media_count
21:15 kahrl so when the server sends the last file m_media_received_count is already equal to m_media_count or larger
21:15 kahrl which means texturesReceived() (<-- that should be called mediaReceived()) might have already returned true
21:16 kahrl which tells game.cpp to stop the load loop and call client->afterContentReceived()
21:16 sapier do we check if everything is complete by comparing numbers?
21:16 kahrl sapier, yes
21:16 kahrl and afterContentReceived starts the mesh update thread (possibly too early in a situation like this)
21:16 sapier great ... no surprise this is wrong
21:19 kahrl do we have a debug.txt of this bug with at least --info level messages?
21:19 kahrl I want to grep "Failed to remote-fetch" it
21:20 kahrl hmm, not increment m_media_count there seems to be correct
21:20 kahrl incrementing*
21:21 kahrl since that variable counts both CURL fetched stuff as well as traditional media fetches
21:22 thexyz are you sure this didn't happen before that cURL patch of mine?
21:22 kahrl I'm not
21:22 kahrl VanessaE: any idea?
21:23 PilzAdam the only thing I can remember is that it occured first time after 3d model support was added
21:24 sapier I guess we shouldn't value "memories" to much they're often wrong
21:24 kahrl if this was reproducable we could bisect
21:25 kahrl although bisecting with a client+server+large amounts of media would be a pain
21:25 VanessaE no idea there, except that while I do have cURL enabled in my builds, I don't have a media server set up.
21:25 sapier bisecting doesn't show you what the erorr is ... and if its a bug that old I guess searching the old way is much more easy
21:26 sapier I haven't ever seen it and I don't use curl .... might point to curl ... or not
21:26 kahrl as long as each commit is decently small bisecting can help narrow the bug down a lot
21:27 sapier ok I don't know how much commits like formspec mainmenu are in there ;-)
21:27 kahrl it's more likely to be in the ones of that size ;)
21:27 VanessaE another new bug I'm starting to see is water flickering between opaque and translucent, even in places where there's nothing around but stone and such
21:27 VanessaE (for later)
21:28 sapier I've never seen it so I can't help sorry
21:28 kahrl VanessaE: I've seen that on kaeza's server, yeah
21:28 VanessaE ok so it ain't just me :)
21:29 Exio VanessaE: I had that bug about water flickering between opaque and translucent when trying to swim in it
21:29 Exio when i'm underwater or outside, "translucent", in the middle => opaque
21:30 VanessaE not to change the subject, I just wanted to mention it before I forgot
21:31 kahrl it never happens in my singleplayer worlds, so maybe it is some mod that changes water?
21:32 kahrl though I don't get why it would flicker
21:32 * VanessaE shrugs, don't think so
21:33 Exio i always get it in my SP world
21:33 Exio maybe we are talking about different bugs :P
21:49 NakedFury joined #minetest-dev
22:17 ch98 joined #minetest-dev
23:06 serengeor joined #minetest-dev
23:07 Taoki Is it possible to issue a Lua command from the console like any script would? Wanna test a Lua function I'll be adding and it would be much easier that way
23:08 Exio make an "eval" mod :P
23:08 VanessaE I think worldedit has that feature, Taoki
23:08 VanessaE //lua <code here>
23:08 VanessaE something like that
23:08 Taoki Ok, so the mod needs the feature?
23:08 Taoki Thought the console would allow you to simulate any script line
23:09 Taoki Ah, or maybe not. Cuz scripts are re-compiled at startup
23:09 kahrl Lua can be dangerous to expose to remote clients
23:09 Exio s/can //
23:09 kahrl worldedit restricts it to users with the server privilege
23:09 Exio it is a scripting language.. that can be used for removing files, running other programs and so on
23:09 Exio :P
23:11 Taoki Ah, yes. Surely that would only be safe to allow in singleplayer
23:11 Taoki Else if you run a Minetest server any player could even delete your drive
23:12 Taoki From the Minetest world :)
23:12 kahrl Worldedit allows it on an actual server too. (though you need the server privilege as I said)
23:13 kahrl of course there could be vulnerabilities that allow one to gain the server privilege
23:15 Taoki Ouch. That does sound risky
23:16 kahrl yeah; I would probably take that out if I ran a public server
23:16 Taoki My Structures mod (an alternative to Worldedit from what I can tell) doesn't use console commands, and only a node  with a formspec, which I assume is safer.
23:16 Taoki I added a privilege requirement for that too though
23:24 kahrl Does somebody know a bit about the media code?
23:24 kahrl When can it happen that m_media_count == 0 when the Client receives a TOCLIENT_MEDIA message?
23:25 kahrl client.cpp:1739
23:48 VanessaE kahrl: tempting fate? :)
23:49 VanessaE (note to self, install the crash_kahrl_randomly mod)
23:50 kahrl I thought that was already installed :P
23:50 VanessaE kahrl: as long as you're on, please pay special attention to your fps around the spawn, particularly how widely it varies over time.
23:51 kahrl still downloading... bunch 218/446
23:58 kahrl ok well I didn't get a mesh update thread error this time
23:58 kahrl or the error message I might have replaced it with
23:59 VanessaE how's your fps here just out of curiosity?
23:59 kahrl if I look up, a steady 25 FPS
23:59 VanessaE we've noticed something about the way the server and client communicate - the server *can* drag the client's fps down, and hard too
23:59 kahrl otherwise between 15-20 FPS, rarely going outside that range

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