Minetest logo

IRC log for #minetest-dev, 2015-10-28

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

All times shown according to UTC.

Time Nick Message
00:00 VanessaE well just saying, ncurses can behave weirdly under screen :)
00:08 zat joined #minetest-dev
00:10 luizrpgluiz I hope you have approval because it would be too bad if you have not ncurses :9
00:10 luizrpgluiz :(
00:11 JohnnyComeL8ly Yeah.
00:11 zat joined #minetest-dev
00:18 luizrpgluiz est made a good upgrade to minetest, hope it can work very well on Linux and Windows
00:20 hmmmm PilzAdam:  can't you just use 1 for number of emerge threads?
00:21 hmmmm I realize there is a bit more serverthread lag this way but it's unavoidable
00:21 hmmmm this isn't fixable until the entire game's locking is changed
00:22 Miner_48er joined #minetest-dev
00:34 Supertanker2 joined #minetest-dev
01:05 est okay ShadowNinja updated my pr to address your remarks
01:24 luizrpgluiz left #minetest-dev
01:27 paramat joined #minetest-dev
01:29 paramat will merge game#681 game#600 soon
01:29 ShadowBot https://github.com/minetes​t/minetest_game/issues/681 -- Fire: Add 'permanent flame' node by paramat
01:29 ShadowBot https://github.com/minetes​t/minetest_game/issues/600 -- Remove unused TNT & torch textures by CraigyDavi
01:53 Taoki_1 joined #minetest-dev
01:53 est dangit, I had enough of colorless gcc errors
01:54 est I update after I've finished my commit
01:54 hmmmm est, what's your opinion on the "none" logging level?
01:54 hmmmm do you think that's an okay name or would you rather it changed?
01:55 est irrlicht has the same, its ok for me.
01:55 hmmmm right now to the developer's perspective it doesn't make sense (whereas it would to the end user because the meaning is inverted for debug_log_levels)
02:05 hmmmm can I push this already?  https://github.com/kwolekr/minetest/commit​/c56d7fe0eba7905b0a63c4a1cfe909988653c23d
02:06 hmmmm I added a usage comment as requested
02:06 est LGTM
02:07 JohnnyComeL8ly What does that mean?
02:15 JohnnyComeL8ly I had a brain fart and I looked it up.. sorry I bothered. :-}
02:15 JohnnyComeL8ly *and so I looked it up....
02:28 hmmmm https://github.com/kwolekr/minetest/commit​/688556a5d1910ead2cf06a87af6908110fa9d035  PTAL
02:29 est well, shrug, okay for me.
02:31 Lunatrius joined #minetest-dev
03:06 paramat now merging game 681 game 600
03:10 paramat done
03:28 est hmmmm, somehow casting doesn't work for me
03:28 hmmmm ??
03:28 est this commit breaks https://github.com/est31/minetest/commit/​d3bcc02558117cab2ba09873dfde9012eebaa9e7
03:28 est it creates a crash if you try to chat
03:29 est and the backtrace includes handleAdminChat
03:30 hmmmm oh
03:30 est right in std::string name = evt.nick;
03:30 hmmmm that's because you're downcasting ChatEventChat/ChatEventNick/etc. to a ChatEvent
03:30 hmmmm when you push it to the queue
03:30 hmmmm aren't you getting any kind of warning for that?
03:31 est I don't think so
03:31 hmmmm C++ isn't Java or C# - you cannot do what you are doing there
03:31 hmmmm you need to make it a pointer instead
03:33 est well, in Java everything is a pointer, no?
03:34 hmmmm MutexedQueue<ChatEvent *> command_queue; ...
03:34 hmmmm command_queue.push_back(new ChatEventChat(blah, bla, blah));
03:34 est will try
03:34 est and delete when emptying the queue
03:34 hmmmm but then that means you need to manage memory
03:35 est and that it lives on the bad bad heap
03:35 hmmmm whether or not that's better than simply having all possible fields is up to you
03:35 hmmmm well ermm
03:35 hmmmm it is already on the heap lol
03:36 hmmmm for what it's worth, it might not be that bad to have all the different parameters in the same struct
03:36 hmmmm a std::string on my platform is only 40 bytes
03:36 hmmmm and i don't think it allocates memory unless you actually set it to a non-blank string
03:37 hmmmm the other obvious suggestion, to use a heap, isn't doable here since you have std::string and std::wstring which are non-POD types
03:37 hmmmm errm
03:37 hmmmm to use a union* is what I meant to say
03:54 OldCoder joined #minetest-dev
04:09 OldCoder joined #minetest-dev
04:23 OldCoder joined #minetest-dev
04:50 paramat left #minetest-dev
05:10 est okay it works now
05:31 est31 joined #minetest-dev
05:36 est31 okay hmmmm updated #3292, it should address most of your points,
05:36 ShadowBot https://github.com/minetest/minetest/issues/3292 -- Add server side ncurses terminal by est31
05:36 est31 almost all of them
05:38 hmmmm unrelated question
05:38 hmmmm why is settingtypes.txt in builtin?
05:39 hmmmm also, do I need to update this from now on instead of minetest.conf.example or vice versa?
05:39 est31 update settingstypes.txt
05:39 est31 and ask pilzadam to update .conf.example
05:39 hmmmm ugh
05:39 est31 its automated
05:39 hmmmm isn't there some kind of python utility to do this
05:39 hmmmm there should be
05:39 est31 no python is ugly
05:40 est31 there is a lua utility though :)
05:40 hmmmm err yeah
05:40 hmmmm lua :) okay
05:40 hmmmm is it a part of util/?
05:40 hmmmm annnd the answer is no
05:41 est31 its in the mainmenu file that handles the setting tab
05:41 hmmmm we need to get PilzAdam to commit the update script
05:41 est31 a method
05:41 hmmmm why...
05:41 hmmmm ugh
05:41 est31 well, I guess it has all the parsing
05:41 hmmmm well whatever
05:42 hmmmm so let me understand the behavior of getch()
05:42 hmmmm this blocks for 100 milliseconds, if there is no input available in the buffer
05:42 hmmmm if there is, it'd get the next char and return it immediately
05:42 est31 correct
05:42 hmmmm so it's like poll() i guess
05:42 est31 idk about poll
05:42 hmmmm alright, sure, I guess that's fine then
05:43 est31 but I have confirmed both by behaviour
05:43 hmmmm lol init_of_curses
05:43 hmmmm void init_of_curses() {
05:44 est31 heh
05:44 hmmmm for (size_t i = 0; i != ARRLEN(list_of_curses); i++) m_curses.push_back(list_of_curses[i]);
05:44 hmmmm const char *list_of_curses[] = {"FUCK", "SHIT", "PISS", "ASS", "AHHHHHHHHH DAMMIT"}
05:45 hmmmm and then write a gdb plugin
05:45 hmmmm on segfault print m_curses[myrand(0, m_curses.size() - 1)]
05:45 est31 lol
05:46 hmmmm "IT'S TIME TO START DROPPIN' SOME F-BOMBS!"
05:47 hmmmm https://www.youtube.com/watch?v=Ol89jfwoPBY
05:48 est31 dont say fuck too fast or bombs spawn too close and explode right in your face when they collide
06:11 OldCoder joined #minetest-dev
06:13 celeron55 umm... here's an email i got: "I have recently come across a strange problem of which i don't know if its just me or if its a bug. When I make a new world and I want it flat, I create the world and go into the map_meta.txt and get rid of all the flags except light and noflat and change noflat to flat. When I load the world, it generates as if i never changed anything but occasionally it will generate flat but with trees and caves (of ...
06:13 celeron55 ... which I supposedly disabled). I have tried doing it in different ways such as loading the world only partially and closing the window before it loads (at 'initializing nodes...') and then changing the map gen but or loading it all the way and doing it (neither of which work). At first i thought that it was the moretrees mod I have but I disabled all the mods and it still did this. Either I'm doing something wrong or this is a ...
06:13 celeron55 ... bug. If there is something else I need to do or something different, please tell me what that is."
06:13 celeron55 is this PEBCAK?
06:13 hmmmm yes
06:13 celeron55 good
06:14 celeron55 i'm too lazy to answer it then
06:14 hmmmm this guy should not be modifying map_meta.txt directly anyway
06:14 celeron55 does there exist a guide anywhere for how those flags are actually supposed to be used?
06:16 celeron55 i guess this http://wiki.minetest.net/Map_generator/settings
06:18 OldCoder joined #minetest-dev
06:18 OldCoder joined #minetest-dev
06:25 hmmmm hmm
06:26 hmmmm I don't think I ever did write any official documentation on mapgen parameters in general because I consider them advanced options that you'd really have to understand the algorithm used to generate map to do anything useful with them
06:26 hmmmm obviously this isn't the case with the simpler things like flat, caves, trees, etc.  features
06:27 hmmmm it's almost the year 2016 and there STILL is no actual "flat" mapgen
06:27 hmmmm i should remove flat as a flag and make it into its own mapgen perhaps
06:38 Hunterz joined #minetest-dev
06:43 est31 okay, fixed a regression and added a note
06:43 julienrat joined #minetest-dev
06:44 julienrat left #minetest-dev
06:51 est31 hmmmm, anything to be done on the ncurses pr before merging? Or do you want to have a look at it first
06:52 hmmmm uhh looking
06:52 hmmmm what was the regression btw?
06:52 est31 I have renamed the variable as you asked and as its meaning was negated I have negated it everywhere but one place
06:52 est31 so the fix was to negate it there too
06:53 hmmmm oh
06:54 hmmmm answer_to_sender needs to be a pointer...
06:54 hmmmm see the rule about non-const reference parameters
06:54 est31 whats the point about it
06:54 hmmmm we went over all this already
06:55 hmmmm the syntax for a parameter being modified vs. constant is ambiguous
06:55 est31 it isnt ambiguous
06:55 est31 if its constant you pass constant reference
06:56 hmmmm this is already a coding rule
06:56 OldCoder joined #minetest-dev
06:56 est31 well its point is that you shouldn't avoid writing const if you can
06:57 est31 its not to force people to use pointers
06:57 hmmmm read this:  https://google-styleguide.googlecode.com/s​vn/trunk/cppguide.html#Reference_Arguments
07:01 est31 so the only thing that isn't "at google we do it and we are smarter than you" or actually lists an advantage of using references is a single sentence
07:01 est31 "References can be confusing, as they have value syntax but pointer semantics."
07:02 hmmmm so your own argument is that "I just decided I don't like this rule, I'm not gonna do it"
07:02 est31 no it has a point
07:02 hmmmm the point for the rule is that it's a convention
07:02 est31 passing non const references as input values is dumb
07:02 hmmmm the convention is, const ref parameters for inputs, non-const pointers for outputs
07:02 est31 hmmmm, you often ignore rules yourself, and when I ask you you ignore me.
07:03 est31 or you change the rule
07:03 hmmmm like where
07:03 est31 for example the line limit one
07:03 est31 I asked you please make the line shorter its too long
07:03 hmmmm which one
07:03 est31 you just adjusted the rule to take 4 tabs
07:03 est31 dont know anymore, look at the logs
07:03 hmmmm I don't remember anything like that..
07:04 hmmmm the rule isn't a hard requirement
07:04 est31 well this one is neither
07:04 hmmmm sigh.....
07:04 est31 there is no point, it would just need additional code
07:06 est31 I still dont get why references aren't banned completely
07:06 est31 I mean  the argument applies to every use of non-const references
07:08 hmmmm if I have a std::string parameter and a const std::string & parameter, both of them have the same semantics to the caller
07:09 hmmmm you have a strong guarantee that the string you pass along is not going to be modified, because it's impossible in fact
07:09 hmmmm so when I have:  std::string foobar, and I pass it along to a function like do_thing(foobar, blah blah ... ); I know that, reference or not, foobar isn't going to change somehow
07:09 hmmmm however if I type &foobar, hmmm
07:10 est31 well you can cast it to const pointers everywhere
07:10 est31 err const references
07:10 hmmmm huh?  I'm not following
07:12 est31 do_thing((const std::string &)do_not_change_string,change_this_string)
07:12 est31 and this way its more secure
07:12 hmmmm why would you do that...
07:12 hmmmm that's ridiculous
07:13 hmmmm whenever you actually want to modify a parameter, it's being used as an output, a second return value
07:13 hmmmm this is the exceptional case
07:13 hmmmm it's much more common that the parameter being passed won't be modified
07:14 est31 well okay you convinced me
07:14 est31 I'll remove the non const reference
07:14 hmmmm again, look at this from a detached point of view, not your own
07:14 hmmmm pretend you've never seen this code before
07:14 hmmmm you have this function you want to use, and you pass these two strings along to it
07:15 hmmmm all of a sudden one gets modified because you didn't inspect every single side effect of every single function you use first
07:15 hmmmm and your code now has possibly a subtle bug in it
07:15 hmmmm bool foobar(const std::string &foobar, std::string &foobaz) is downright dangerous I'd argue
07:17 hmmmm for what it's worth, I did not just decide on this rule.  I proposed it and Zeno` along with a couple others (can't remember) strongly agreed
07:19 est31 I've removed it  see last commit on the branch
07:21 hmmmm ermm wtf
07:23 hmmmm you realize you completely broke it right?
07:23 est31 yea Im fixing
07:26 est31 pushed fixes
07:26 hmmmm wtf
07:26 hmmmm come on stop screwing around with me
07:29 est31 I know, the GNU people wouldn#t like me returning other stuff than an int
07:29 est31 but why return a bool and have a pointer in the params, that is only written to if the return value is true?
07:30 hmmmm because you're leaking memory
07:30 hmmmm rather blatantly
07:30 est31 how
07:30 est31 I do delete it in any case
07:30 hmmmm where?
07:30 est31 right after the if
07:30 est31 I could move it in if you want
07:31 est31 there is no benefit in delete NULL
07:31 hmmmm oh I see
07:31 hmmmm so do you really want to have the ability to send blank lines?  just curious
07:34 est31 I'll make it return a string
07:37 est31 okay pushed
07:37 hmmmm great
07:38 est31 In fact we never would have returned an empty string
07:39 est31 if the string from the script hook were empty, we'd have returned NULL
07:39 hmmmm one more thing
07:39 hmmmm if the console thread throws an exception, the logging outputs won't get readded
07:40 hmmmm so the error output would never be seen
07:41 est31 no finally :(
07:42 hmmmm i think you should add a finally for that macro and put it there
07:42 hmmmm but not in this same commit
07:45 hmmmm now what about the macro parameters with no parentheses?
07:46 est31 what about https://github.com/minetest/minetest/commi​t/f2c45dd142020f56f71252ddea06880c802cb83d
07:47 est31 ah you mean parentheses internally, not externally
07:50 hmmmm yes, that's kind of important...
07:50 hmmmm consider what would happen if somebody decided to use your macro and they did:
07:51 hmmmm IS_ASCII_PRINTABLE_CHAR(some_condition ? 'a' : '~');
07:51 est31 yeah
07:52 est31 first I had them then I removed them again because zeno would appear and complain about too many parentheses
07:52 hmmmm no, parentheses all the time for macro parameters, you don't screw around with macros..
08:01 hmmmm you need to return L""
08:05 Darcidride joined #minetest-dev
08:06 hmmmm okay, looks good to me.
08:12 est31 okay, build fixed
08:13 est31 hmmmm, is this method to achieve what you asked ok, or should I somehow move the class declaration, e.g. into the header or so? https://github.com/minetest/minetest/pull/3292/​files#diff-42a115173bd1d0c211ba295a43271207R89
08:13 hmmmm ewww....
08:14 hmmmm do you have to declare it right in the function?
08:14 hmmmm that does work though, yes.
08:16 nrzkt joined #minetest-dev
08:21 est31 okay moved it to header
08:24 hmmmm LGTM +1
08:24 est31 nice, thanks!
08:24 est31 do you think it can be merged?
08:24 hmmmm wait hold on, this is pretty big, do you have anybody else who wants to take a look at it?
08:25 est31 well, blockmen has +1ed it, but he has made no comments on the implementation, so I don't think he has really audited the code.
08:26 hmmmm yeah... a very large part of a code review is to look for errors and bad implementations
08:26 hmmmm i'm sure if we asked everybody here they'd universally agree that a server console is a good idea
08:27 hmmmm so a +1 based on the feature concept alone is pretty useless
08:27 hmmmm I bet I missed some actual bugs here so it's really a good idea to have somebody else take a real look at it too
08:28 est31 yeah.
08:43 nrzkt est31 let me looking for it
08:46 est31 ok
08:46 CraigyDavi joined #minetest-dev
08:50 nrzkt some first comments added
08:50 julienrat joined #minetest-dev
08:50 julienrat left #minetest-dev
08:51 est31 does really everything have to be const?
08:51 est31 I add alot of methods, if they all got const that would be a lot of work
08:53 nrzkt i looked at many methods and i note those things to be const, no method needs to be const because you modify class content in every method if i have correctly read
08:53 nrzkt i juste noticed some arguments should be const because this is their essence to be const.
08:53 nrzkt just*
08:54 est31 yea I mean that
08:54 est31 I add a large number of methods
08:54 est31 or touch their aruments
08:55 est31 if I made them all const, it would mean a lot of work
08:55 nrzkt there are not many things to change :) and if you use a good IDE it's very fast to change this :)
08:55 nrzkt you don't need to make anything const
08:55 nrzkt just come function arguments i notice
08:55 nrzkt don't be evil :p
08:55 nrzkt some*
09:01 werwerwer joined #minetest-dev
09:01 PilzAdam joined #minetest-dev
09:03 est31 PilzAdam, you want to look at the pr as well?
09:04 PilzAdam ncurses?
09:05 est31 yup
09:07 PilzAdam well, I guess the people who already looked that the code know more about whether it's good or not
09:08 PilzAdam I like the idea, so I can give you a +1/2
09:08 Calinou joined #minetest-dev
09:08 sfan5 i like the idea too
09:08 sfan5 haven't looked at the code though
09:09 PilzAdam I'd say merge it now and see if it breaks for someone
09:09 PilzAdam master isn't stable
09:10 PilzAdam just be around to fix stuff
09:22 Megaf_ joined #minetest-dev
09:36 nrzkt PilzAdam, no
09:36 nrzkt is this the new idea to break master to see it someone complains ?
09:37 PilzAdam heh, no
09:37 PilzAdam that was exaggerated
09:37 PilzAdam if a pull request sits on it's own for a too long time then it doesn't help anybody
09:38 PilzAdam if it's stable enough then it should be merged into master so we can get proper testing on it
09:39 PilzAdam just waiting for bugs to magically appear while the PR is not merged yet is wasted time
09:41 nrzkt PilzAdam, i agree, but where are the users complaining because they are crying master is not stable and they are using it in prod ?
09:41 nrzkt like VanessaE two month ago :s
09:42 PilzAdam if people use the dev branch in production then it's their fault
09:42 PilzAdam release builds are stable; we don't promise anything else
09:42 est31 it is good to have as many people as possible to test the dev branch
09:42 PilzAdam also, it's not like the dev branch regularly breaks
09:43 nrzkt PilzAdam, i agree with you, but some core devs spit on me months ago when i said master is a dev branch not the production....
09:43 PilzAdam who?
09:44 nrzkt hmmmm for example
09:45 nrzkt but i don't care about master stability, for me master is a dev branch, but i only notice somes cry because of master stability two month ago and now it seems we can do anything on it. It's not logical. Maybe it's time to have a rule for master
09:48 PilzAdam Totally official rule for master branch stability from now on: "The master branch is a development branch. It is intended and should be used for testing. We try to keep it stable enough, so people can test it."
09:50 PilzAdam btw, any plans for the next release date?
09:50 PilzAdam it seems there is no milestone on github yet
09:51 est31 I don't think it should be made issue based
09:51 est31 it should rather be made time based
09:51 est31 e.g. we say we release twice a year
09:52 Calinou twice a year is probably not enough, we're a game, not a GNU/Linux distribution
09:52 est31 perhaps we can substract one month, thats in average the time a release gets stalled
09:52 PilzAdam the original plan was to release every month
09:52 Calinou quarterly releases sound much better
09:52 est31 Calinou, releases have ~4 weeks of stability freezes
09:52 est31 that would totally stop development
09:52 PilzAdam release discussions are always fun, because everyone has great ideas and in the end nothing changes anyway
09:52 est31 I mean you could commit nothing to master during that time
09:53 PilzAdam feature freezes are supposed to be only 1 week
09:53 Calinou games feel dead if they don't have regular releases
09:53 est31 haha
09:53 Calinou also, the freeze could be 2 weeks only
09:53 Calinou again, this is just a game
09:53 Calinou nobody really makes money on it
09:53 est31 still its taken very seriously
09:54 Calinou PilzAdam, still better than those "Let's rename Minetest to Minetest Game!" ideas
09:54 PilzAdam I'd say, release every 2 to 3 months with a feature freeze of 1 week
09:55 PilzAdam but actually provide rc builds in the feature freeze
09:55 PilzAdam I mean real official rc builds with posts in the news forums
09:55 Calinou Minecraft was released once per month in its Beta period
09:55 Calinou it felt extremely active, and players loved it
09:55 Calinou it was nearly rolling release
09:55 est31 we are not minecraft
09:55 est31 we release code the moment we develop
09:55 est31 its no over the wall development
09:57 est31 but yeah I do agree with PilzAdam even changing the smallest thing about releases will terribly upset a small group of people, so therefore everybody does simply the same than before
10:01 cib0 joined #minetest-dev
10:09 nrzkt est31, PilzAdam , Calinou i propose 3 months release per year. 2 months + 1 week of dev: 3 weeks for stability
10:10 PilzAdam 3 weeks feature freeze is way too long
10:10 PilzAdam I'd say make it 1 week, with the option to make it longer if serious problems arise
10:10 nrzkt PilzAdam, 1 week isn't because every dev is not aware in one week. 2 is sufficient
10:10 est31 man, feature freeze doesnt go with "weeks"
10:10 est31 I have tried precisely that
10:10 est31 but hmmm reverted my version commit
10:10 est31 hours after it was commited to the repo
10:11 est31 I have announced the release date one month before
10:11 est31 I have waited one month
10:11 est31 but it got reverted
10:11 est31 even though we were already multiple weeks in feature freeze
10:11 nrzkt est31, maybe you should have support from other core devs ? :)
10:12 nrzkt and the feature freeze should be more relevant, with less absence ? :)
10:12 est31 then hmmm started flagging all kinds of issues as "blocker"
10:12 est31 and even got issues like that "multi second lag" one
10:12 nrzkt blocker issues for a release should be done at the FF start
10:13 est31 PilzAdam, you realize that that issue wont be fixed within a week?
10:13 est31 I mean you cant even specify when it is resolved
10:14 PilzAdam there shouldn't be open "blocker" issues when we enter the feature freeze
10:14 est31 man, it
10:14 PilzAdam it should be all release-ready then
10:14 nrzkt PilzAdam, the blocker issues are there for beeing fixed for release no ?
10:15 est31 but the main problem of this whole "lets set a date" or "2 weeks for this" or "all X months" stuff is that you can't do it if on the other hand you guarantee that these issues get fixed before release
10:19 est31 You know, I have asked paramat once whether I can release in one month
10:19 est31 he said it wasn't enough he'd rather have two months
10:19 est31 so I've said ok, lets have two months
10:19 est31 a bit later he said that now he has to do everything in a haste
10:20 est31 because the release was so soon
10:20 est31 and his feedback was the _only_ feedback I've received about a release date
10:21 est31 even if it was failed eventually
10:22 est31 releasing is no fun
10:23 est31 if anybody wants to do it, ok, but I don't want to be release manager again
10:23 est31 PilzAdam, you want to be the release manager for 0.4.14?
10:24 PilzAdam I would do it, but I can't promise that I have enough time
10:24 PilzAdam IRL and stuff
10:43 technics joined #minetest-dev
11:07 proller joined #minetest-dev
12:02 cib joined #minetest-dev
12:25 DFeniks joined #minetest-dev
12:52 crazyR joined #minetest-dev
13:01 AnotherBrick joined #minetest-dev
13:21 zat joined #minetest-dev
13:32 luizrpgluiz joined #minetest-dev
13:35 Darcidride joined #minetest-dev
13:39 Samson1 joined #minetest-dev
14:57 julienrat joined #minetest-dev
14:57 julienrat left #minetest-dev
14:59 rubenwardy joined #minetest-dev
15:05 hmmmm joined #minetest-dev
15:12 cib0 joined #minetest-dev
15:36 Amaz joined #minetest-dev
16:54 Krock joined #minetest-dev
16:58 luizrpgluiz left #minetest-dev
16:59 Darcidride_ joined #minetest-dev
17:13 Gael-de-Sailly joined #minetest-dev
17:18 kilbith joined #minetest-dev
17:19 cib0 joined #minetest-dev
17:20 kilbith is there a way to hide those spammy errors ? https://lut.im/78KsAzXRCA/3SzQYcwdFvBTEJU7.png
17:20 kilbith 1) it happens frequently on every servers whenever there's an entity on sight
17:20 kilbith 2) there only was one boat around on this screenshot
17:20 kilbith 3) the error flooding is endless
17:24 hmmmm whose server are you on?
17:24 kilbith VoidPixel, Raw Land, and another i don't remenber
17:25 kilbith the last was VoidPixel
17:25 VanessaE users on my servers occasionally report the same errors.
17:26 hmmmm VanessaE, after destroying the pipeworks that had thousands of items in it, did you see that error anymore?
17:27 VanessaE hmmmm: I personally have not seen it since, but I recall at least one report of that error on my "Basic" server.  the only entities there are item frames, pedestals, signs (all three of which are immobile), and carbone_mobs.
17:27 hmmmm yeah I saw it on OldCoder's server too
17:27 hmmmm and now there's this VoidPixel one that seems to be having it often
17:27 hmmmm I'm gonna take a closer look at these servers soon
17:28 kilbith thanks
17:34 kilbith 811, 4, -560 at VoidPixel is a good place for catching them
17:35 hmmmm I can't do any diagnosis with a client alone
17:36 hmmmm are you the owner of that server?
17:36 kilbith nope
17:39 Player2 joined #minetest-dev
17:40 hmmmm hmm
17:40 hmmmm well
17:41 hmmmm I need to work with the owner of that server or OldCoder since he has the same problem quite frequently on his own server
17:41 hmmmm the key difference between you two and VanessaE is that your problem is readily replicatable
17:41 kilbith i'll put the admin in contact with you
17:42 hmmmm ok
17:42 hmmmm is the admin of that server technically inclined, do you know?
17:43 Robert_Zenz joined #minetest-dev
17:43 hmmmm like if I tell him to modify the source would he be able to do that and compile it
17:44 kilbith no idea of his technical skill but he can compile
17:44 hmmmm okay then
17:44 luizrpgluiz joined #minetest-dev
17:47 nrzkt joined #minetest-dev
18:04 Darcidride joined #minetest-dev
18:40 asl97 joined #minetest-dev
18:48 younishd joined #minetest-dev
19:04 ElectronLibre joined #minetest-dev
19:19 cib_ joined #minetest-dev
19:24 celeron55 joined #minetest-dev
19:28 luizrpgluiz left #minetest-dev
19:44 kilbith left #minetest-dev
19:44 H-H-H joined #minetest-dev
19:46 est31 joined #minetest-dev
19:58 paramat joined #minetest-dev
20:04 paramat hmmmm i can start work on a flat mapgen, mgv5/v7 don't support 'flat' and would be messy if they did. a dedicated mapgen can use the biome API, can have a simple parameter for terrain level, and spflags for optional biomes and decorations
20:06 paramat the 'trees' flag does not affect mgv5/v7, perhaps it can toggle all decorations in those
20:07 est31 nrzkt, are these all places where I should put const? or is it needed somewhere else too?
20:12 est31 I just don't want to put const to every param of every method, because that's what it seems you want
20:33 Puma_rc left #minetest-dev
20:38 celeron55 uh what
20:39 celeron55 we do not have a code style that requires const for any parameters except for references
20:39 celeron55 the fact that C++ has the wrong default constness isn't a reason to spam all code with so much const that it's unreadable
20:41 celeron55 so in this case whatever nrzkt is saying is most definitely wrong
20:42 CraigyDavi joined #minetest-dev
20:51 hmmmm paramat:  yeah not only terrain level but what about cave level and so on
20:51 paramat ok
20:52 hmmmm the "flat" parameter came from a time where there was only v6 and v7, and it made more sense to have one with the new and the old biome systems
20:59 crazyR joined #minetest-dev
21:03 nrzkt celeron55, const parameters are not anoying if they are correctly used. Maybe some should changes their glasses
21:04 celeron55 hmmmm: what do you say about this
21:05 hmmmm adding const qualifiers to parameters where they are obviously not needed at all is kind of... dumb
21:07 celeron55 yeah; the thing is, const is actually *needed* on parameters that are not copied; otherwise it's not *needed*
21:08 celeron55 i.e. it will not worsen the interface
21:09 hmmmm it's bad because it's weird
21:09 celeron55 as i said, everything should be const by default, but isn't because C++ is based on C which is based on not enough real programming experience
21:09 hmmmm developers already have too much to think about, they don't need more mental load thanks to constructs that are non-standard and have no utility whatsoever
21:10 hmmmm i take it you've jumped onto the Rust train
21:10 est31 and even rust allows modifications
21:11 est31 by default
21:11 hmmmm personally I won't even look at a language or whatever until it's 10 years old
21:11 est31 I dont even know whether it has such a concept
21:11 celeron55 hmmmm: this doesn't have anything to do with rust
21:11 est31 well it has read only references
21:11 est31 (which are the default)
21:12 hmmmm I thought you had to explicitly specify "mutable"
21:12 est31 if you create a reference that you want to edit later on you have to say "mut"
21:12 hmmmm right, that's only for references though
21:12 est31 but the built in types like i32 or so they all get copied
21:12 hmmmm for parameters passed by value, why would you ever care about its constness
21:12 est31 and are modifyable, at least I think
21:13 est31 I don't think its a good thing to expose the fact whether you edit a passed value in the internal implementation of your method
21:13 celeron55 true; it doesn't belong to the interface
21:15 celeron55 not that C++ allows you to make clean interfaces in the general case but in this case it kind of does
21:16 celeron55 hmmmm: i think nrzkt's idea is that you don't then accidentally modify the value that you wanted to stay constant during the execution of the function
21:16 nrzkt this is one idea yes, make the code robust
21:17 nrzkt what users want ? robust code or readable for newbie code ?
21:17 hmmmm so you're saying it's pro if you spam "const" everywhere in totally inappropriate places?
21:17 hmmmm wtf man
21:17 celeron55 i have to say that i have probably almost never stumbled upon a bug caused by a parameter being modified mid-way in a function
21:17 nrzkt hmmmm, another time, like every day you are saying something that i don't said
21:18 hmmmm for what it's worth
21:18 hmmmm I am a professional working in the industry, and I don't make parameters const that are not references or pointers
21:18 hmmmm I guess my code is newbie though
21:18 hmmmm :|
21:19 celeron55 as if others in here weren't professionals :P
21:19 nrzkt i don't care you work in industry. Linus Torvalds and some OpenBSD developers are not in the industry and they are better than you to make good code.
21:19 hmmmm freaking interns screwing everything else
21:19 nrzkt this is a non argument
21:19 nrzkt (and i don't said those mentioned developpers said we should make all const :) )
21:19 hmmmm nrzkt:  so why don't you contribute to the linux kernel by adding const to all the parameters
21:20 nrzkt i don't contribute to linux kernel
21:20 hmmmm I'm sure they'll be happy with your contribution because you're a pro and they are newbs
21:20 nrzkt the only patch for kernel i provide is for FreeBSD kernel a little sysctl added to the nfs driver to make something customizable :p
21:20 celeron55 staying on the topic, i would like to leave the choice in this matter to the author of each function; you can try to persuade anyone you want with any reasons you want, but they don't have to comply with anyone
21:21 nrzkt celeron55, no problem for me, i only point the constness is a good thing to do if you have time :)
21:22 hmmmm are you going to tell people to use yoda comparisons next? *shivers*
21:22 nrzkt i think you are waiting SW VII too much :p
21:22 PilzAdam how to dev: 1) take an keyword from the programming language of your choice 2) start fighting over it's usage 3) start insulting people
21:24 nrzkt PilzAdam, i suggest to use java instead of C++ for MT because every industry use it and there are many bloated jars to use :p
21:25 celeron55 PilzAdam: i'm just using this channel as a warm-up to actually developing one of my smaller projects!
21:26 est31 okay, seems I were wrong, rust does have all function params immutable by default
21:26 PilzAdam celeron55, if you read any ideas that you have for that project here then you should reconsider that idea
21:26 est31 and you have to make them mutable with "mut"
21:34 paramat left #minetest-dev
21:54 nrzkt https://lut.im/CPAC0Ddd6e/vAKKEuo3f7wDYjMH.png
21:54 nrzkt is that normal ?
21:56 est31 no
21:56 est31 I think however that bug should have been recently fixed by hmmmm
21:56 nrzkt this night ?
21:56 est31 some u32 prng stuff
21:56 nrzkt it's the seed problem ?
21:56 hmmmm sounds to me like it's a corrupted set of mapgen params.
21:57 hmmmm told you guys to not tough map_meta.txt
21:57 nrzkt i didn't modified it
21:58 nrzkt i removed the whole world and recreate it
21:58 hmmmm touch* i meant
21:58 nrzkt and i show you this pic :p
21:58 hmmmm well what are your map_meta.txt params then?
21:58 hmmmm because they're obviously corrupt
21:58 nrzkt rm -Rf ~/.minetest/worlds/mapgenv6
21:58 nrzkt and started a new map :)
21:59 hmmmm you might've messed up your mapgen param overrides in your config file then
21:59 nrzkt i will look at this, else there are the newly created world things: http://pastie.org/10514543
22:01 hmmmm maybe one of your mods are messing it up
22:01 nrzkt it's the default game
22:04 kilbith joined #minetest-dev
22:04 kilbith i got the same issue two weeks ago (default game, untouched map_meta) : http://irc.minetest.ru/minet​est-dev/2015-10-17#i_4426712
22:05 hmmmm dunno I'm trying out the same params and it seems fine to me
22:05 kilbith unless if it has been touched behind my back
22:06 hmmmm well I can't reproduce
22:11 nrzkt hmmmm, can "LuaPerlinNoiseMap::LuaPerlinNoiseMap(NoiseParams *params, int seed, v3s16 size)" be the problem because seed is a int ?
22:11 hmmmm what does that have to do with the mapgen
22:12 hmmmm that's for mods only
22:12 nrzkt don't know i'm not mapgen aware
22:12 kilbith i put a dollar on the new setting tab as suspect
22:12 nrzkt kilbith, can be possible, i merged them in my local tree today
22:13 nrzkt but it seems i have the same problem with dedicated server
22:13 nrzkt and dedicated server didn't use the setting tab
22:15 crazyR joined #minetest-dev
22:17 rubenwardy I find it quite hard to find things in the new settings tree
22:20 Player2 joined #minetest-dev
22:28 nrzkt very strange, it seems my mapgenv7 on the new world is a fractal mapgen...
22:34 proller joined #minetest-dev
23:06 paramat joined #minetest-dev
23:09 paramat hmmmm i'd like to merge this later #3302 , is tested. i have plans for creating patches of bare stone in deserts by setting 'depth top' to 0 and using filler noise to reduce 'filler depth' to 0 in places
23:09 ShadowBot https://github.com/minetest/minetest/issues/3302 -- Mgfractal: Add filler depth noise by paramat
23:11 paramat meh jenkins checks are failing, ignoring
23:33 AnotherBrick joined #minetest-dev
23:33 Fixer joined #minetest-dev
23:51 AnotherBrick joined #minetest-dev

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