Minetest logo

IRC log for #minetest-dev, 2013-11-11

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

All times shown according to UTC.

Time Nick Message
01:06 Taoki joined #minetest-dev
01:12 SpeedProg joined #minetest-dev
01:30 hmmmm SegFault22, please take it to #minetest or post on the "feature requests" section of the forums
01:30 hmmmm this channel is for development discussion only
01:31 SegFault22 I tried taking it to #minetest, and one of your moderators suggested to take it here.
01:31 hmmmm welp
01:31 SegFault22 There is no feature requests section, by the way.
01:32 hmmmm perhaps the feature Discussion page then
01:33 SegFault22 That has been tried once before. But seeing the number of people who responded, I now understand that on the forums, I have to do something against the rules/bad for anyone to recognize it...
01:33 SegFault22 So here is the best, and seemingly only place, for contacting the developers.
01:40 SegFault22 left #minetest-dev
01:57 mrtux joined #minetest-dev
02:36 mrtux joined #minetest-dev
02:37 ShadowNinja About LuaJIT packaging: I would rather have no libs packaged in Minetest and use dynamic libs. I agree with segfault, but you shouldn't have to register types, just store it as a std::string rather than a enumeration.
02:38 ShadowNinja I actually tried using the builtin method and was surprised that craft types were hardcoded.
02:38 ShadowNinja (For a technic machine)
03:41 SegFault22 joined #minetest-dev
03:46 SegFault22 left #minetest-dev
04:03 OldCoder joined #minetest-dev
04:04 OldCoder This is not yet a bug report. It won't be unless I see a definite pattern. But thought I'd mention that multiple worlds of mine, running git close to current, hang sometimes on Item Definitions load.
04:16 deltib joined #minetest-dev
04:45 Sokomine question: why is a map where all the chunks the server sent so much smaller than the map on the server? even though it seems to encompass all the buildings?
04:46 Sokomine part of it may be underground (with mines), explored but unbuilt territory, air?
04:46 VanessaE underground.
04:46 Sokomine it can't be all underground
04:46 * Sokomine looks frightened
04:46 VanessaE wanna bet? :)
04:46 * Sokomine takes picks away from players! :-)
04:49 OldCoder Sokomine, the other world is online and very nice
04:49 OldCoder See me in PM or other chanels
04:49 OldCoder * channels
06:18 nore joined #minetest-dev
07:14 proller joined #minetest-dev
08:11 darkrose joined #minetest-dev
09:30 proller joined #minetest-dev
09:49 nore I found a bug: inv:set_size("space crash", 1) for an inventory crashes the server
09:49 nore with segfault
09:49 nore (the problem is the space inside the list name)
09:49 nore any thoughts on this?
09:56 nore debug_log_level = 4 gives nothing more that the traditionnal:
09:57 nore 10:53:50: ACTION[ServerThread]: singleplayer places node test:crafting_table at (-19,6,23)
09:57 nore 10:53:50: ACTION[ServerThread]: facedir: 2
09:57 nore and after that, server crash with segfault
09:57 nore so, does anyone knows where it comes from?
10:10 nore ah, and another bug: a negative list size does _not_ crash the server immediately, but when it goes to saving the map, it uses 100% CPU power
10:11 nore and it seems it uses a lot of RAM too since it was very hard to stop it and after having stopped it, there was still lag
10:14 Jordach joined #minetest-dev
10:14 iqualfragile joined #minetest-dev
10:22 BlockMen joined #minetest-dev
10:23 nore BlockMen, https://github.com/minetest/minetest/issues/996
10:24 nore ^ what do you think should be done?
10:31 BlockMen does it crash in singleplayer for you?
10:32 nore yes
10:32 nore the very moment I place the node, I got a segfault
10:33 nore what OS do you use? it may not be the same as me...
10:34 nore I guess I should write that in the issue too
10:34 BlockMen thats strange...first times it worked for me, now it crashes with "bad allocation"
10:34 BlockMen and win7 64bit
10:34 nore for me it is Debian... I will write that
10:35 BlockMen but i had not last git, so i will need to compile first and then test again
10:36 nore I have found another bug too: a negative list size value will _not_ crash the server (I mean, when there is no space in the name), but at map save, the server eats 100% CPU and RAM
10:36 nore it was very hard to stop it
10:38 Jordach joined #minetest-dev
10:40 BlockMen for me it throws also "bad allocation" but freezes then
10:41 nore the negative stack count?
10:42 nore does it freeze your computer? (I had a lot of difficulties to stop it the first time)
10:43 nore the second time, I did something that killed minetest after 30 seconds (and it was enough to make it eat a lot of RAM)
10:44 BlockMen yes, the negative list size. it freezes minetest and windows kills the task then
10:44 nore I guess it is killed because it eats too much RAM or something like that
10:45 BlockMen i dont think so because it freezes nearly instantly
10:45 nore anyway, those two bugs need to be fixed (at least a better error message for the first one and a clean crash with error message for the second one)
10:46 BlockMen yeah, will take a look at it later, need to do other things now
10:46 BlockMen you made a report for the second one too?
10:47 nore no, not yet
10:47 BlockMen *on git
10:47 nore will do that
11:05 ImQ009 joined #minetest-dev
11:14 Calinou joined #minetest-dev
11:41 Jordach joined #minetest-dev
11:53 Taoki joined #minetest-dev
11:53 Jordach joined #minetest-dev
11:59 Jordach joined #minetest-dev
12:03 Taoki joined #minetest-dev
12:12 Jordach joined #minetest-dev
12:45 EvergreenTree joined #minetest-dev
12:51 zat joined #minetest-dev
13:02 jojoa1997 joined #minetest-dev
13:27 hmmmm joined #minetest-dev
13:54 nore hmmmm, what do you think of that: https://github.com/minetest/minetest/issues/996 https://github.com/minetest/minetest/issues/997
13:54 nore (what should be done?)
14:33 salamanderrake joined #minetest-dev
14:50 Ritchie joined #minetest-dev
14:56 Zeitgeist_ joined #minetest-dev
14:56 Zeitgeist_ joined #minetest-dev
15:24 djdduty joined #minetest-dev
15:33 RealBadAngel joined #minetest-dev
15:43 celeron55 hmm, i have had this thing stubbornly floating in my head for enough time that i must do something to it
15:43 OldCoder What thing?
15:47 nore celeron55, what?
15:50 celeron55 is anyone whose thoughts i generally tend to approve interested in attempting a from-scratch redesign of the minetest engine, privately with me, with the goal of getting a good understanding of what kind of approaches are possible, what their pros and cons are and how the development of each could possibly work in the future (and publishing the end result)?
15:50 celeron55 i'm mostly pointing at kahrl, but i understand that kahrl is terribly busy
15:51 nore I don't understand the engine nor C++ enough, and currently have almost no time, so it can't be me...
15:52 nore but when you say from-scratch, that would change the whole API/etc too?
15:52 Jordach celeron55, ask darkrose, she'd help
15:52 celeron55 it's one of the many questions, and i don't want to start a discussion here because they almost always de-evolve into some kind of shouting competitions
15:54 OldCoder Here?
15:54 OldCoder This channel seems quiet
15:55 OldCoder One thing, celeron55... would such a rewrite be completely backwards compatible?
15:55 celeron55 it's one of the many questions
15:55 OldCoder Yes; if it was *not* the project might fail
15:55 celeron55 well that's one of the many questions too
15:55 OldCoder People will wish their old worlds to continue
15:55 jojoa1997 joined #minetest-dev
15:55 kaeza or maybe not...
15:55 OldCoder But, then, a rewrite will let folks learn from past mistakes
15:56 celeron55 yes, one of the many questions!
15:56 OldCoder Heh. Do not rush. But imagine the possibilities. I assume that it would still be C++
15:56 OldCoder Oh, well... Life is not perfect
15:57 celeron55 there are like infinite possibilities but the question is whether any of them makes any sense
15:57 OldCoder Perhaps you could design a new language with C as the back-end. The new language could be used to write the core engine. Possibly layers on top of mods as well.
15:58 Akien joined #minetest-dev
15:58 nore OldCoder, about the old world thing: it would not be very difficult (compared to the other things to do) to make a map converter...
15:59 kaeza suddenly: Minetest written in Python 8)
15:59 proller php !
16:00 celeron55 don't get hyped though; the end result likely is that nothing makes sense
16:00 proller celeron55, also make it really threaded, for using all cores
16:00 OldCoder nore If that was taken seriously, yes. But I am upgrading worlds right now inside of a consistent framework and everything breaks. Backwards compatibility requires thought.
16:01 OldCoder Multi-Threaded. celeron55 there is an idea most will agree on.
16:01 celeron55 proller: this kind of discussion is completely counterproductive; i wish people would just stick to the original question and some thoughtful person would volunteer
16:02 OldCoder celeron55, You must allow people to dream and to joke if you wish buy-in to the concept
16:02 OldCoder PHP is pushing it though :-)
16:03 OldCoder celeron55, Besides this *is* productive...
16:03 OldCoder You have identified backwards compatible and multi-threaded as concepts. And keep my "new language with C as back end" thought on the table; it might open up new possibilities.
16:04 celeron55 i already had those in mind and the whoever i would like to thoroughly dig this through would too have had them in mind; it's bringing nothing to the table except hiding my request
16:07 celeron55 (except that "new language" is nonsense)
16:09 proller celeron55, can you define short goal? like "threads, physics, speed, ...."
16:09 OldCoder As you wish, celeron55. Good luck.
16:09 celeron55 proller: no, because that is the question
16:10 proller + sizeof(int) world size
16:13 nore proller, that is not a good idea because it is architecture-dependent
16:17 proller okay,  2^32
16:19 proller but it not very big problem make with current code
16:24 OldCoder Why limit to 32 bits?
16:24 Calinou joined #minetest-dev
16:24 nore proller, there are problems (float precision, etc.) with Irrlicht when you go too far away
16:24 VanessaE nore: that's more a problem with how minetest uses irrlicht I guess
16:25 proller where problems started? if near 2^31 - it not a problem
16:28 general3214 joined #minetest-dev
16:29 general3214 Anyone online?
16:30 VanessaE no.
16:34 general3214 Oh okay
16:34 general3214 I'll just not bother VanessaE since VanessaE is not online.
16:34 VanessaE :P
16:35 general3214 :D
16:36 proller celeron55, and maybe something like ogre or... as renderer
16:36 general3214 What's celeron like?
16:37 general3214 Wait
16:37 thexyz this is not the channel for offtopic
16:37 general3214 Wrong channel
16:37 thexyz please don't shitpost here
16:37 general3214 omg these people so sensitive
16:37 VanessaE general3214: this channel has a fairly strict subject limit.
16:38 thexyz please stop
16:38 general3214 I know. I ain't dumb
16:38 general3214 general3214> Wrong channel
16:42 Calinou <nore> proller, there are problems (float precision, etc.) with Irrlicht when you go too far away
16:42 Calinou fix: move mesh around camera instead of moving camera around mesh
16:42 Calinou (iirc)
16:43 Calinou floating point precision errors always happe,n
16:43 EvergreenTree joined #minetest-dev
16:44 nore Calinou, how do you store object pos then?
16:45 Calinou don't know that :P
16:45 Calinou fix is very hard to do
16:47 nore we would then need fixed point... (and rewrite a lot of things)
17:02 sapier joined #minetest-dev
17:04 general3214 Is this channel ONLY about core development, or development in general?
17:04 thexyz read the topic
17:05 general3214|AFK I take that as a yes
17:05 thexyz and be aware that while I personally don't give a shit some people are easily irritated by afk nicknames and the most crazy ones can even kick you for that, I think
17:06 OWNSyouAll joined #minetest-dev
17:09 sapier can someone explain to my why recently everyone wants to start from scratch?
17:09 thexyz read the log
17:10 sapier I did that's why I ask
17:10 thexyz it doesn't seem to me that everyone wants to
17:10 thexyz people are just showing interest
17:10 sapier is there any major issue that would be fixed by starting from scratch I missed by now
17:11 thexyz it seems announcing it was not the best idea
17:11 nore perhaps entity duplication (dunno though)
17:11 ShadowNinja Speed, perhaps using a better graphics library, better network stack, etc.
17:11 sapier entity mechanics could be rewritten without rewriting whole minetest ;-)
17:12 sapier same for most other things ;-)
17:12 sapier the only benefit a complete rewrite would have is "yes that feature will be added later" ;-)
17:13 ShadowNinja Anyone have experience with SFML? I've also heard of using just plain OpenGL.
17:14 sapier dropping irrlicht means rewriting/replacing whole gui too
17:14 sapier both ingame as well as menu
17:15 sapier and there is no guarantee our own engine will be faster or more portable than irrlicht my personal bet is  it wont be
17:15 nore on the other hand, there would be only one to rewrite now we have formspec menu...
17:15 ShadowNinja Yes, and changing v3f and the like too.
17:15 sapier I guess formspec will be dropped too
17:16 sapier and our main problem won't be fixed ... lack of parallelism within main game engine core
17:16 nore perhaps something good to do would be to change all v3s16 types to a POS type with a #define somewhere, so it can easily be changed in the future
17:16 nore and float pos too, etc
17:16 sapier it's not a matter of beeing good or not it's just a matter of time to be spent to fix it
17:17 ShadowNinja nore: There are also constants used in places like position hashing.
17:17 rubenwardy joined #minetest-dev
17:17 sapier https://github.com/minetest/minetest/pull/971 what about this one I wont add things that ugly without clear permission of at least 2 core devs
17:17 ShadowNinja 32-bit positions would be wholy incompatible with 16-bit ones.
17:18 nore ShadowNinja, there should be some world conversion probably then
17:18 sapier we already have performance issues do you really think increasing complexity will do any better?
17:18 ShadowNinja nore: Yes, it would probably be mostly runtime incompatabilities though.
17:19 ShadowNinja sapier: No need for 32-bit positions now. Just discussing it.
17:19 sapier is that bug already fixed segfault22 discovered?
17:19 ShadowNinja As for #971 I echo what kwolekr said.
17:20 nore that's why I suggest a #define POS_TYPE v3s16 ...
17:20 nore sapier, what bug?
17:20 ShadowNinja Or typedef.
17:20 sapier echo what?
17:20 nore ShadowNinja, what's typedef?
17:21 ShadowNinja nore: It adds another name for a type, eg size_t is typedefed to something like unsigned int.
17:21 sapier nore the crash with invalid list sizes
17:23 sapier ShadowNinja as kwolekr didn't say yes or no what's your oppinion to 971
17:26 ShadowNinja sapier: It's awully ugly, and I would really rather have a better option. But if there is no other way...
17:28 sapier YES or NO ... a maybe is useless ... I'm gonna continue repair my car now if noone has courage to decide by time I'm finished I wont ask again
17:29 thexyz sapier: unrelated, why do you put /***************************************​***************************************/ everywhere? I don't remember seeing stuff like that in our code guidelines
17:29 thexyz it's yes for sure
17:29 thexyz too bad that's what happens with most of pulls
17:29 thexyz people who didn't research the subject complain about stuff they contain and can't decide whether to merge it or not
17:29 thexyz so people who did the research and made pull request eventually burn out
17:30 proller i'm ready to join all my pulls in one
17:30 proller now i cant make many things
17:31 OWNSyouAll joined #minetest-dev
17:31 thexyz proller: did you see https://github.com/minetest/minete​st/pull/895#issuecomment-28129708
17:32 proller no problem with abm timers, they increase chance if some calls was missed
17:33 proller but some every second on_step will be skipped
17:34 proller but anyway it better than now in master
17:42 Akien joined #minetest-dev
17:44 proller idea: improve commiting rules: if pull waits more than [1-2] weeks and have no objections - every core dev can commit it (even pull creator)
17:44 OWNSyouAll joined #minetest-dev
17:44 proller no future in current state
18:02 PilzAdam joined #minetest-dev
18:14 OWNSyouAll joined #minetest-dev
18:21 sapier proller not a good idea as any crap can be merged if just enough core devs didn't have time to review
18:23 VanessaE sapier: to be fair, "just any crap" could be merged anyway, regardless of such a rule.
18:26 sapier thexyz I added the /**/ because I regularly do this in all of my cpp files to have a chance to find functions ... there's no other function separator defined in coding style so /***/ is as good as any other. I still believe it's way more readyble but if this is the only reason for not merging it I'm gonna remove it
18:27 thexyz it's not, it's just something I find strange
18:27 thexyz the pull is OK
18:27 sapier Ok let me be more precise VanessaE any crap might be merged even more easy ;-)
18:28 sapier especially if there are many small functions code with /***/ in there is more readable ... at least to me
18:28 VanessaE actually, there IS a defined separator, unofficially.
18:28 PilzAdam nore, https://github.com/minetest/minetes​t/issues/996#issuecomment-28225332
18:28 VanessaE it's just that it isn't really written down and it's only been used in a few places - a line of -------
18:29 VanessaE (see some of the files in builtin)
18:29 sapier that's MY separator too for lua code ;-)
18:30 sapier /***/ variant is ancestor of ----- as comments are -- in lua :-)
18:30 thexyz_ joined #minetest-dev
18:30 sapier "best.gir.never.wins..."  ?? :-)
18:30 thexyz_ joined #minetest-dev
18:32 sapier ok as I now have at least one clear yes vote for 971 I'm gonna count my own vote as second yes and merge in half an hour ... so last chance to veto
18:32 PilzAdam celeron55, I am very interested in such a rewrite because it would give me the ability to understand the basic concepts of such and engine better
18:32 sapier ShadowNinja did you merge your lua error handler fixes yet?
18:33 PilzAdam celeron55, although I dont know if you would benefit from that
18:33 sapier I suggest forking rewrite to minetest2
18:35 proller joined #minetest-dev
18:35 troller joined #minetest-dev
18:38 sapier another suggestion for speeding up merges what about assigning pull requests to core devs to review? method of assignment has to be defined. maybe most simple variant credit core devs with lines changed reviews
18:43 sapier REQUESTING FEATURE FREEZE FOR 0.4.8 RELEASE  at end of November anyone agree?
18:44 * sfan5 agree
18:44 sfan5 s
18:45 sapier maybe we could create a integration branch to not stall development while freeze is active
18:46 sfan5 wouldn't that be messy?
18:46 sapier no merges during feature freeze should be minor bugfixes onla
18:46 sapier only
18:47 troller sapier, something like dev branch - https://forum.minetest.net/viewtopic.php?id=7033
18:47 sapier similar yes but with same merge rules as current master
18:48 Akien joined #minetest-dev
18:49 sapier I'm not talking about a permanent thing only while feature freeze is active once release is done this branch should be merged to master ... this way we could review merges for next version and don't have to check again after 0.4.8 release
18:49 troller current rules like for atomic station software
18:49 troller and result is frozen development
18:50 sapier imho we'd need a stable as well as development branch ... but that's another issue
18:53 PilzAdam we have a stable branch
18:53 VanessaE sapier: agreed, with the proviso that the release be delayed beyond the end of November if some bugs just can't be fixed in time.
18:53 sapier we do? when was it used last time?
18:54 celeron55 it has been updated on each release
18:54 PilzAdam https://github.com/minetes​t/minetest/tree/stable-0.4
18:54 PilzAdam master is our dev branch
18:54 sapier of course VanessaE End of november would be only "target date"
18:55 sapier wow didn't even know about that one this is 0.4.7 not 0.4.0?
19:17 BlockMen i noticed a problem with lvm. after doing a few more calculations i get an "not enough memory" error, even by letting grow a lot saplings. http://pastebin.com/FFVcFyjV
19:17 BlockMen is there a problem with not released memory?
19:18 sapier did you check memory usage on your system? asking to be sure it's a out of memory not some "in lua" limitation
19:23 sapier ok pushing i18n pull request now
19:24 sapier plz test it I checked linux mingw32 and vs2012 32 bit only hope this is enough
19:27 BlockMen its stable around ~330mb (minetest) until it uses lvm. then it increases more and more until mem usage 80%+
19:27 sapier what are you doing in there?
19:28 PilzAdam sapier, "/home/adam/Minetest/minetest/​src/guiChatConsole.cpp:560:78: warning: ignoring return value of 'int mbtowc(wchar_t*, const char*, size_t)', declared with attribute warn_unused_result [-Wunused-result]" IIRC this was there before your patch too?
19:28 BlockMen growing 200 saplings
19:29 sapier I don't remember changeing anything in chatconsole
19:30 sapier nope changed almost all other gui files but not this one ;-)
19:30 PilzAdam this seems to be the last warning left (for gcc, that is)
19:31 sapier sadly I didn't see it I'd have fixed it
19:31 sapier fixed some other warnings within that patch too
19:34 sapier is saplings using l-tree?
19:35 nore sapier, no
19:36 sapier strange would be more likely to eat up all memory on recursive algorithms
19:37 nore the strange thing is that I have already used LVM for quite intensive tasks (i.e., reading ~100 4x150x4 areas every 2 seconds)
19:37 nore and I never had this problem (except when I tried to read a 240x150x140 area)
19:38 sapier maybe it's a special function triggering the bug, blockmen can you provide code to trigger it?
19:38 sapier lol I'd not consider that a problem nore just buy a bigger machine if you want to do things like that
19:39 nore I don't consider it a problem either... anyway, that was badly-written code, I optimized it much after...
19:40 nore gtg
19:40 BlockMen sapier, i uses the code from default, so https://github.com/minetest/minetest_game/b​lob/master/mods/default/functions.lua#L138
19:41 BlockMen but i also noticed when using my mines mods, etc. everything that uses lvm and used for a lot of operations/time
19:41 sapier yes but how do you grow 200 saplings? :-) manually?
19:41 BlockMen ya, sure. 200 is plant fast (<2min) ;P
19:41 BlockMen s/plant/plated
19:42 sapier lol ... I'm gonna try once ;-)
19:43 BlockMen even if it wont crash...the mem usage increases from ~330mb to ~1,5gb just because of lvm
19:47 sapier maybe a garbage collection issue?
19:49 sapier I just planted 250 saplings no noticable memory increase
19:54 BlockMen well, if it just happen for me then we have no problem :)
19:55 sapier we have a problem we just need more information and as you're the only one to notice by now it's up to you to find out how to reproduce ;-)
20:00 sapier https://github.com/minetest/minetest/pull/993 any comment?
20:01 PilzAdam valgrind anyone?
20:02 PilzAdam (re that sapling problem)
20:03 sapier I can't reproduce so blockmen has to do
20:03 troller sapier, copypaste 8(
20:04 sapier copypaste?
20:06 thexyz copypasta indeed
20:07 thexyz sapier: he's complaining about some blocks of code looking too similar
20:07 thexyz like, they're exactly same
20:08 sapier it's json parser what do you expect?
20:08 sapier yet that's already in master as well as the bugs ;-)
20:09 thexyz I'd expect maybe a function to which you pass the string?
20:09 thexyz and this is exactly what happens when you copypasta
20:09 thexyz you have to fix bugs in multiple places
20:09 thexyz but meh, I guess you know all of that
20:10 sapier I do yet someone said to create a better json parser
20:10 sapier if I remember correctly it was proller
20:11 troller yes, but... ;)
20:12 sapier but what? *g*
20:12 sapier *g* --> :-)
20:12 werwerwer_ joined #minetest-dev
20:13 troller in my queue was more important (for me) things
20:14 sapier I won't write it either so we'll have to stick with current suboptimal version
20:25 general3214 joined #minetest-dev
20:30 Miner_48er joined #minetest-dev
20:35 Akien joined #minetest-dev
20:38 werwerwer joined #minetest-dev
20:41 NakedFury joined #minetest-dev
20:52 bas080 joined #minetest-dev
21:28 Akien_ joined #minetest-dev
22:06 zork_ joined #minetest-dev
22:09 hmmmm celeron, a from-the-ground approach is not a very good one
22:11 hmmmm the codebase we have today is not just a huge block of code, it's days, weeks, months, and years of labor and knowledge gained
22:11 hmmmm you say you want to use the knowledge gained from minetest to make the right decisions a second time, but how do you know you're not going to make more bad decisions on other parts?
22:12 hmmmm also what if much of the knowledge gained only pertains to designs only made because of some root design choices
22:13 sapier https://github.com/minetest/minetest/pull/993 hmmmm do you agree to this one?
22:13 hmmmm if you go with something fundamentally different, say for example, executing things on events instead of polling a queue every step, what you have learned might not be very helpful at all
22:14 hmmmm and then there are things like the modding api that really are decent for what it is and we can't do much better due to the nature of scripting
22:15 hmmmm don't really understand the point of changing *clients_raw != 0 to clients_raw != "" when you kept the thing to the right of it
22:15 sapier asString() creates a temporary object
22:15 sapier thus storing a pointer to a temporary object is quite dangerous ;-)
22:16 hmmmm oh oh nevermind, you changed that from c_str()
22:16 hmmmm makes sense
22:16 hmmmm yeah, what you did looks fine
22:16 sapier yes done it at other place recently didn't know it's hidden here too
22:16 hmmmm yes, can't you push things on your own anymore?
22:16 sapier I can yet I need agreement ;-)
22:17 hmmmm not so much for a bugfix that you know is going to solve the problem
22:17 sapier I'm still the new one so I stick to the rules ;-)
22:17 hmmmm more controversial things like restarting the process under a different locale, that could take agreement I'd say
22:18 sapier actually I didn't merge it without agreement
22:18 hmmmm i know you didn't, I was just using that as an example of something that anybody would need a consensus on to merge
22:18 sapier btw it's only done for silly msvc build ... I strongly discourage building mt with it
22:18 hmmmm whereas a memory leak fix like above is fine
22:19 sapier it's more a possible access violation ... most likely it wont cause any harm but the other ones did cause harm on windows
22:20 hmmmm oh, access violation?  hmr
22:20 sapier actually it's more likely you read crap
22:21 hmmmm oh, sorry
22:21 hmmmm I'm not paying enough attention to what I'm reading
22:21 sapier that pointer is just random space ... ok not exactly random as it's data most likely is still there, but it may have already been reused
22:21 hmmmm just got back from a day of work
22:22 sapier oh I see :-) btw I do think same at rewrite thematic ... partial rewrite of components yes but not a complete rewrite
22:22 sapier and partial can be done within normal development (if there's someone who really cares to do)
22:22 hmmmm right
22:23 hmmmm I'd rather work on specific things that need work rather than the clean slate rewrite, e.g. formspec and mainmenu
22:23 sapier what about the real issues? map/environment locking? *smile*
22:23 hmmmm right
22:23 hmmmm we need to really work on that
22:24 sapier formspec and mainmenu may be suboptimal but missing map/environment locking stops any attempts to gain performance by parallelization
22:25 troller and dont save not changed new blocks
22:25 sapier pushing temp object fixes now
22:25 troller my map 2.6G already 8(
22:25 hmmmm troller, that would screw up minetestmapper
22:25 troller minetestmapper dont work on my map too ;)
22:26 troller then make it adjustable
22:27 sapier https://github.com/minetest/minetest/issues/997 I consider this to be a blocker issue for 0.4.8 what do you think?
23:04 sapier https://github.com/minetest/minetest/pull/998 should fix 996 as well as 997
23:09 ShadowNinja Actually the mapper thing isn't that important because you usually don't need to see what is in unmodified blocks.
23:10 sapier ahh Shadow :-) what about your lua errorhandler fixes?
23:14 ShadowNinja sapier: I don't have agreement for them yet. And there is a small issue that I tried to fix locally...
23:15 sapier what issue?
23:15 ShadowNinja sapier: 'throw LuaError("Something")' from a API function causes "ServerError: LuaError: C++ exception." I think LuaJIT is at fault for this.
23:16 ShadowNinja And Regular LuaError's din't have a traceback anymore. (I added that locally)
23:16 ShadowNinja s/R/r/
23:16 ShadowNinja don't*
23:17 ShadowNinja I don't know how to compile with non-JIT Lua though...
23:17 sapier ok they should have tracebacks for sure
23:17 sapier didn't we have issues with exceptions within lua called  functions before?
23:17 ShadowNinja Put adding the traceback is pointless when you just get "C++ exception"
23:17 ShadowNinja Yes.
23:17 sapier of course
23:18 sapier worst thing is enclosing all lua functions within try catch .... quite a lot of work and no guarantee to work
23:23 ShadowNinja sapier: This is what I have locally, does it look good? https://gist.github.com/ShadowNinja/7422436
23:26 Mallot1 joined #minetest-dev
23:27 sapier sorry to late to review today I recommend updating your pull request if you think it's better than the current one.
23:27 sapier left #minetest-dev
23:31 Taoki joined #minetest-dev
23:46 zat1 joined #minetest-dev

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