Time  Nick          Message
03:31 hmmmm         who the heck added ipv6 support
03:31 hmmmm         I don't get the purpose of IPv6AddressBytes
03:31 hmmmm         why couldn't you just use struct in6_addr?
03:32 hmmmm         there should be a socket class cleanup
05:55 hmmmm         https://github.com/kwolekr/minetest/commit/4195f613225c44140ac7380ae57ab169e044e306
05:55 hmmmm         guys check it out ^
05:55 hmmmm         I cleaned up test.cpp
06:08 Zeno`         is minetestserver still built?
06:08 Zeno`         my timestamp for it has not updated in ages
06:09 Zeno`         geez, no wonder my tests don't show stuff I expected
06:11 hmmmm         Zeno`, what error does checking "ipv6_enabled" in the socket unit test prevent?
06:11 Zeno`         wth
06:12 Zeno`         hmmmm, I don't know what the current thing it prevents is, but when I added it the situation was that if the user did not have ipv6 in their kernel then the unit test failed
06:12 hmmmm         no you didn't add that
06:12 hmmmm         nerzhul did, many months *after* you added your own fix
06:13 Zeno`         I added it originally but nzxrt changed it
06:13 Zeno`         yeah
06:13 hmmmm         so why are there two fixes
06:13 Zeno`         I dunno. I think the fixed it again for gerble
06:13 hmmmm         and one of the two fixes suck
06:13 Zeno`         (or whatever that build system is called)
06:13 hmmmm         you can guess which i like
06:13 Zeno`         the non-gerble one? lol
06:13 hmmmm         lol wth
06:14 hmmmm         you mean gerrit first off
06:14 Zeno`         jenkins
06:14 hmmmm         jernkins yeah
06:14 hmmmm         jenkins-ci
06:14 Zeno`         whatever the silly thing is called :)
06:14 hmmmm         it runs the unit tests in a jail - okay, fine
06:14 hmmmm         that reads a different setting though
06:14 hmmmm         bind_address
06:17 Zeno`         I remember arguing about it. He wanted to remove the unit tests for release builds entirely and I remember saying they were useful
06:17 Zeno`         beyond that I don't recall much
06:17 hmmmm         unit tests are not run by default anymore
06:18 Zeno`         yeah I approved *that* change
06:18 hmmmm         hmm i hope they get run in the CI
06:18 hmmmm         at the very least
06:18 Zeno`         yes they do
06:18 hmmmm         okay sweet
06:18 Zeno`         well, I was told they do
06:18 Zeno`         :/
06:18 hmmmm         so my patch changes the output quite a bit
06:18 hmmmm         it looks like this http://fpaste.org/215528/29131143/
06:19 hmmmm         i wonder if that will affect the CI functionality
06:20 Zeno`         So why is minetestserver not built?
06:20 hmmmm         you mean due to my patch?
06:20 Zeno`         No
06:20 hmmmm         man i have no idea
06:20 Zeno`         I mean since... since I dunno
06:20 Zeno`         it used to get built automatically
06:20 hmmmm         if anything it's probably caused by shadowninja's cmakelists cleanup
06:21 hmmmm         let's just say it hasn't been perfect
06:21 Zeno`         but now it's not built at all
06:21 hmmmm         but we're fixing the problems as they come along
06:21 Zeno`         so I've been testing on an ancient binary
06:21 hmmmm         oh it'll get built, you just have to explicitly check BUILD_SERVER
06:21 hmmmm         BUILD_CLIENT BUILD_SERVER
06:21 Zeno`         hmm
06:21 Zeno`         if neither are present shouldn't *both* be built like it used to?
06:21 hmmmm         actually I'm not sure if one of those options existed
06:22 hmmmm         well i'd hope so
06:22 hmmmm         but i guess not
06:22 Zeno`         I know that on my VPS I have to use SERVER_ONLY (or something)
06:22 Zeno`         which is fine
06:22 Zeno`         but on here I expected (and expect) both client and server to be built
06:24 hmmmm         humm https://github.com/minetest/minetest/commit/93fcab952b28f4db39d9326b83f14cbd86c0cf17#diff-af3b638bc2a3e6c650974192a53c7291L42
06:24 hmmmm         i'm not really sure i understand what the purpose of that logic is to begin with
06:27 Zeno`         Well, build server only is needed for VPS's that don't have irrlicht installed (just the headers)
06:27 Zeno`         i.e. the client cannot be built because there is no X
06:28 hmmmm         i think the server should be built by default on all platforms
06:28 hmmmm         not just non-win32-and-osx
06:28 Zeno`         well it used to be
06:28 Zeno`         :)
06:29 hmmmm         yeah now it's always disabled by default
06:29 hmmmm         like i said.  you can change it if you want, just change a word from false to true
06:31 Zeno`         seems that it got built on the 17th so something changed 9 days ago... is that when that diff is from?
06:31 hmmmm         it could be
06:32 hmmmm         the commit doesn't tell when it was pushed, only when it was 'authored'
06:32 hmmmm         well
06:32 Zeno`         yeah
06:32 Zeno`         I dunno
06:32 hmmmm         this has been part of the master branch for longer than 9 days
06:32 Zeno`         this is all very strange lol
06:33 Zeno`         Apr 17 might be when I last removed cmake cache
06:47 Etzos         hmmmm: I'm sure you noticed, but in case you didn't, your commit included files in src/unittest/CMakeFiles which I assume you didn't intend.
06:47 hmmmm         yeah I saw that
06:48 hmmmm         removing
06:51 Zeno`         I just got email about hmmmm's commit!
06:51 Zeno`         and he wins a prize!!!!!!!!
06:51 hmmmm         what?
06:51 Zeno`         for the 3000th unread email in my inbox
06:51 hmmmm         oh cool
06:51 Zeno`         what would you like? here's a, umm... a... medal!
06:51 hmmmm         so it's like one of those "you're the millionth visitor to this website, click here to claim your prize!" ads
06:51 Zeno`         yep
06:51 hmmmm         man, I sorta miss the internet of yesteryear
06:52 hmmmm         now there are no ads anywhere
06:52 hmmmm         no ads, no popups, nothing, just internet plain and simple
06:52 jin_xi        Zeno`, why?
06:52 hmmmm         because I made a pull request for it
06:53 jin_xi        why get emails? you can turn it off
06:53 hmmmm         i started making pull requests for things i don't want to immediately commit because more people seem to look at it posted there rather than in irc
06:53 hmmmm         and yeah, I do realize this is going to break shadowninja's log refactor PR
06:53 hmmmm         but it's easier to break his than the other way around since all he needs to do is remove a file
06:56 Zeno`         jin_xi, I know I can turn it off, but I like them
07:05 hmmmm         !!!!!!! https://travis-ci.org/minetest/minetest/jobs/60075127  see this shit?
07:05 hmmmm         this happens all the time
07:05 hmmmm         this is what I mean when i say that certain files get implicitly included by other files in the messy dependency web of some distros (and freebsd)
07:06 hmmmm         so it works for me, it works for debian, and then it fails to compile on windows and ubuntu or something like that
07:25 Zeno`         or android
07:49 hmmmm         lol :D
07:49 hmmmm         https://github.com/blokel/blokel/commits/master
07:49 hmmmm         a fork of minetest that does absolutely nothing aside from modifying README.txt and default settings
07:50 hmmmm         people are funny
07:50 Zeno`         He doesn't know C++
07:50 Zeno`         (well, apparently not based on the questions he's asked in #minetest)
07:50 hmmmm         why did he fork it then
07:50 Zeno`         *shrug*
07:50 hmmmm         why can't his thing just be a mod
07:51 hmmmm         sometimes, i just don't know.
07:51 Zeno`         maybe he thought he knew more than her does?
07:51 hmmmm         it's possible
07:52 hmmmm         is it my imagination, or was there a PR floating around somewhere that allowed colored console text?
07:53 Zeno`         wth
07:53 Zeno`         now my taskbar has gone lol
07:58 Zeno`         O.o
07:58 Zeno`         minetest crashed my 'puter!
07:59 hmmmm         s/minetest/graphics drivers/
07:59 Zeno`         s/graphics drivers/opencl
08:03 hmmmm         okay, so are people okay with  https://github.com/kwolekr/minetest/commit/f824216241c7f4d2f56aae25042d8be89bc0d50d  ??
10:48 celeron55_    f824216241c7f4d2f56aae25042d8be89bc0d50d seems reasonable to me
12:18 srifqi        ~tell est31 please check #2648
12:18 ShadowBot     srifqi: O.K.
13:29 Zeno`         will merge #2650
13:29 ShadowBot     https://github.com/minetest/minetest/issues/2650 -- Fix player pitch and yaw not being set properly by Etzos
13:30 Zeno`         and #2648
13:30 ShadowBot     https://github.com/minetest/minetest/issues/2648 -- Remove fly mode in simple main menu by srifqi
13:38 Zeno`         #2589 is next
13:38 ShadowBot     https://github.com/minetest/minetest/issues/2589 -- Fix fast leaves when texture_clean_transparent is enabled. by Warr1024
14:22 est31         #2620 is ready for review btw.
14:22 ShadowBot     https://github.com/minetest/minetest/issues/2620 -- SRP based login by est31
14:35 Krock         est31, is it fully tested?
14:45 est31         I have tested it, yes
15:05 est31         Trying to store id=21355 statically but block (-15,0,-151) already contains 49 objects. Fo
15:05 est31         rcing delete.
15:05 est31         arrrrggg!
15:10 Zeno`         I did fall asleep
15:10 Zeno`         heh
15:11 Zeno`         Unless we have lots more Chinese speaking people here I reckon just merge the new Chinese translation
15:12 est31         yea
15:12 Zeno`         There is no way to check this stuff... google translate of it looks ok (I guess)
15:12 est31         the issue with translations is that we core devs can't really review them
15:12 Zeno`         yep
15:12 Zeno`         Japanese the one guy does it a lot.. I just merge his changes as soon as I see them
15:13 est31         so two options: either let them rot, and then merge them, or merge them faster and trust ppl
15:13 Zeno`         for the others I leave them a week and if nothing comes up I figure merge
15:13 Zeno`         I think we should trust
15:13 est31         yea
15:13 Zeno`         (after Google translate to check for obvious stupidity)
15:13 est31         or third option: find a translation platform we can utilize
15:13 Zeno`         yeah... right :P
15:15 Zeno`         want me to merge it?
15:15 Zeno`         (I'd rather sleep but I'll do it first if you like, lol)
15:16 Zeno`         k... I will merge (doing now)
15:16 est31         ok
15:17 Zeno`         done :)
15:17 * Zeno`       idles
15:18 Zeno`         est31, 2602 is probably ok as well
15:18 Zeno`         but I will review more tomorrow if you really want me to
15:19 est31         20 or 02?
15:21 Zeno`         can't trick me that easily =D
15:21 Zeno`         02... but it probably should be re-reviewd
15:21 Zeno`         I'm too tired right now :(
15:24 est31         :/
15:32 ShadowNinja   sfan5: Zeno approved it, and it doesn't really matter since it's so trivial.
15:33 sfan5         ?
15:33 est31         what is "it"?
15:34 ShadowNinja   Zeno`: I disabled the server build by default because if you've got the full (client) binary it's unneeded and since it has to rebuild everything specifically for the server-only build it nearly doubles compile time.
15:34 Zeno`         ShadowNinja, I understand that... I just wish I'd known
15:34 Zeno`         what did I approve? :)
15:34 ShadowNinja   sfan5: The project name lowercasing optimization thing.
15:34 Zeno`         oh, is that all?
15:35 Zeno`         if the full minetest (not server only) is built, can there be a symlink created?
15:35 Zeno`         hmm... maybe that wouldn't work
15:36 Zeno`         ShadowNinja, server only build does have some differences though
15:36 Zeno`         So shouldn't the server only build be tested by the CI as well?
15:37 ShadowNinja   Symlink no, but probably a text file with only #!minetest --server or so.
15:37 Zeno`         a script might work, yeah (even if it's not +x and !#/path/to/shell) (I suppose)
15:38 ShadowNinja   Zeno`: Maybe, it's a subset of the full build though, so it should only fail if someone forgot to apply #ifndef SERVER somwhere.
15:38 Zeno`         ShadowNinja, hopefully they don't forget then :)
15:39 Zeno`         anyway, I'm just saying because I've been profiling minetestserver for a while now and didn't realise it was stale (LOL)
15:40 ShadowNinja   hmmmm: Re: colored text: https://github.com/minetest/minetest/pull/2411
15:40 hmmmm         no no colored CONSOLE text
15:41 est31         idk just know that freeminer has it
15:41 hmmmm         wel
15:41 hmmmm         I can add it easily, but i'd rather wait for shadowninja's logging cleanup first
15:42 est31         After that I can add the logging API.
15:43 hmmmm         I was thinking something like
15:43 ShadowNinja   hmmmm: Ah, I have a Python script that does that :-).  It's designed to manage multiple servers.
15:44 est31         you mean Zeno`?
15:44 ShadowNinja   (auto-reboot, backup, map, etc)
15:44 hmmmm         errorstream % COLOR(CLR_RED |CLR_INTENSITY) << "Some really important text here" % COLOR(CLR_RESET) << " less important text here" << std::endl;
15:45 est31         and what is COLOR?
15:45 ShadowNinja   hmmmm: Oh, I was just thinking errorstream gets red, verbosestream gets cyan, etc.
15:45 hmmmm         ShadowNinja:  That is doable but it might seriously get annoying
15:46 hmmmm         est31:  Hmm, I suppose it's not absolutely necessary if we're going to overload something different from <<
15:46 est31         also, it should recognize if the console we are piping to supports colors
15:46 Zeno`         colour
15:46 hmmmm         sure.  totally fine.
15:46 est31         so if its a > logfile, then we shouldn't send color
15:46 ShadowNinja   That's how LuaBot and my script handle it, and I like it.  We should probably add a --color option like coreutils have though.
15:46 ShadowNinja   hmmmm: What were you saying would break my cleanup?
15:47 hmmmm         doesn't your cleanup modify test.cpp?
15:47 est31         then people come along and demand "I want it as setting in minetest.conf"!
15:47 hmmmm         my patch deletes test.cpp
15:48 hmmmm         we'll also have to come up with rules on when you're able to use colored text in console messages
15:48 ShadowNinja   hmmmm: Oh, yes, more than one of them does.  Did you move the tests into tests/*.[ch]pp or similar?
15:48 Zeno`         can we have colour vs. color as a setting?
15:48 hmmmm         otherwise all the developers will think their messages are super important and minetest's console would turn into a rainbow
15:48 hmmmm         lol zeno
15:48 hmmmm         ShadowNinja, yeah
15:49 est31         how should that setting be called?
15:49 hmmmm         well i think it should be enabled by default
15:49 est31         colo(u)r_name = coluour
15:49 hmmmm         oh that
15:49 hmmmm         shrug
15:49 hmmmm         it's trivial to add support for setting aliases
15:50 hmmmm         I'd like to warn people about using aliases to things
15:50 hmmmm         it makes everything more complex because there's more to think about
15:50 hmmmm         in my mapgen lua api I have tons of aliases for things thinking it'd help people do things their style but really it just causes more complexity and confusion
15:51 ShadowNinja   hmmmm: Does my logging PR look good now?  The only issues I remember was fixing some things that aren't strictly related and a minor Irrlicht logging bug that existed before anyways.
15:51 hmmmm         yeah I thought it was fine from the start
15:52 hmmmm         aside from having multiple features in the same commit
15:52 ShadowNinja   Alright, I'll rebase and merge now then, and maybe split off some things.
15:52 est           It still has the WIP label, hasnt it?
15:54 ShadowNinja   est: It did.
15:58 ShadowNinja   Shouldn't the signal handler set the handler back to the system default on its first call?  Otherwise the second ^C just sets the default handler back without doing anything significant.
15:58 est           the second ^C is like a sigkill, isn't itß
15:58 est           ?
16:00 ShadowNinja   est: No, more like a SIGTERM.
16:01 ShadowNinja   It calls abort() or something like that.
16:02 ShadowNinja   Well, currently it just sets the signal handler back, but the third ^C will trigger that.
16:04 est           PlayerSAO::isStaticAllowed() {return false; }
16:04 est           this is impossible
16:05 ShadowNinja   print should really be separated from minetest.debug.
16:05 est           oh it is possible
16:05 est           playercao isnt used
16:05 est           is it?
16:05 est           sao*
16:06 ShadowNinja   hmmmm: Do you agree with those two points? ^
16:06 hmmmm         huh
16:06 hmmmm         i don't really have any strong opinions
16:06 ShadowNinja   print should go to stdout, not prefixed and sent to stderr and the log file.
16:07 ShadowNinja   minetest.debug should do that.
16:13 hmmmm         you mean minetest.debug should print with the prefix?
16:13 hmmmm         i agree that print shouldn't go to stderr but i think it should still have the prefix
16:13 hmmmm         everything has the prefix apart from debugging messages in fact
16:13 hmmmm         dstream is the only stream that doesn't have prefixes
16:17 ShadowNinja   hmmmm: dstream has a prefix now.  dstream was acutally completely seperate from all the other streams, I made it just another log level, except that it's always printed (higher priority than LL_ERROR) and it's printed without a name (as in 00:00:00: [ServerThread]: ...).
16:18 hmmmm         why
16:18 hmmmm         is it impossible to print without a prefix now?
16:19 ShadowNinja   Because that's much simpler, and it includes the timestamp and thread name where DTIME and get_current_thread_id were used before.
16:19 ShadowNinja   You can use cerr and the like directly, but yes, currently there's no unprefixed version.
16:19 hmmmm         well i don't agree with THAT behavior
16:20 ShadowNinja   The only time that you really want it to be unprefixed is if you're doing something like a migration, where you don't want it logged to a file.
16:20 est           migration?
16:20 ShadowNinja   est: SQLite -> LevelDB or similar.
16:21 hmmmm         well, I say there is another time
16:21 est           and why not have it there?
16:21 hmmmm         debugging events where you really don't need clutter that's irrelevant to you
16:22 hmmmm         i don't care that the event happened at exactly this date and time at 53:21 and 430 milliseconds
16:22 hmmmm         more clutter is bad
16:23 est           there is #2559
16:23 ShadowBot     https://github.com/minetest/minetest/issues/2559 -- Add options to control timestamps for logging by est31
16:24 hmmmm         i'm sorry
16:24 hmmmm         i just don't see any real purpose behind adding prefixes to dstream
16:24 hmmmm         it's not doing anything but removing options for the developer
16:26 est           Perhaps it can be done/discussed on another PR?
16:26 ShadowNinja   hmmmm: It removes other hacks like DTIME.  If you prefer I could add another unprefixed stream, but it doesn't seem like it would really be usefull to how have thet info.
16:26 ShadowNinja   not have that*
16:26 hmmmm         how does dstream have anything to do with dtime
16:27 ShadowNinja   hmmmm: DTIME was a macro to add the timestamp to messages logged with dstream.
16:27 ShadowNinja   As in dstream<<DTIME<<msg;
16:28 hmmmm         dstream needs to be out-of-band with the rest of the logging system
16:28 hmmmm         i already don't appreciate that the debugstream initialization comes as late as it does
16:28 hmmmm         and this makes it much worse
16:28 hmmmm         it's for DEBUGGING
16:29 hmmmm         DTIME could be made non-hacky, just turn it into a function getTimestamp() for the rare occasions where a developer would actually want the timestamp with their debug messages
16:29 hmmmm         or add dstream and dtstream (the latter has the timestamp prefix)
16:30 hmmmm         but jeez don't just change things for the sake of change.  that's not productive.
16:30 ShadowNinja   hmmmm: It's initialized within the first few lines of main(), or at least the stderr output (log file comes a bit later in init_log_steams in init_common).
16:31 hmmmm         i thought you said that dstream was added in with the rest of the normal logging levels
16:31 ShadowNinja   hmmmm: If it's just for debuging then it shouldn't be in the production code.
16:31 hmmmm         what about unit testing
16:32 hmmmm         come on shadowninja, stop
16:32 ShadowNinja   hmmmm: Use actionstream or infostream or so.
16:32 ShadowNinja   Or dstream.
16:33 hmmmm         but then it has a big ugly prefix
16:33 hmmmm         look all i'm asking for is that you stop taking away options
16:33 ShadowNinja   If it should always be printed then you should use dstream.  The prefix is usefull for tests.
16:34 ShadowNinja   I'll add direct_to_log_outputs_stream or so.
16:34 hmmmm         :|
16:34 hmmmm         no
16:34 hmmmm         then you're forcing a whole bunch of other code to change
16:34 hmmmm         what is the purpose of all this again?  how is this making minetest *better* than it was?
16:35 ShadowNinja   It adds threadnames and timestamps to dstream and cleans up the ugly debugstreams thing.
16:36 ShadowNinja   g_log_outputs_stream, forwarding to g_logger's log outputs for LL_NONE.
16:36 hmmmm         and then now to have no timestamps to dstream you're making everything else change
16:41 hmmmm         what do other people think about this
16:41 est           when I'm debugging, I'm logging to errorstream
16:44 ShadowNinja   ^ That's basically dstream with a prefix.
16:45 hmmmm         i like my current logging levels, and i need a stream that always prints no matter what the log level
16:45 hmmmm         for logging to a file, i can see timestamps as being useful
16:45 hmmmm         but sometimes i just need some things printed with no extra added BS
16:46 est           like?
16:46 hmmmm         if your solution fulfills these requirements, then i'm okay
16:47 ShadowNinja   hmmmm: dstream is still set to always print.
16:48 hmmmm         but the second is not
16:48 hmmmm         why remove options from the developer?
16:48 hmmmm         do you know better than everyone else and their use cases?
16:51 hmmmm         dstream with DTIME is only used in exactly 1 place: the signal handler in porting.cpp
16:51 hmmmm         I don't understand why you'd change EVERYTHING ELSE for such a rare use case
16:51 hmmmm         and the DTIME thing isn't a hack, it's just a shortened version of getTimestamp()
16:58 ShadowNinja   hmmmm: It's used in a few other places, although they might be mostly commented out.  I'm adding g_log_outputs_stream anyway.
17:00 hmmmm         and if you are, that means the behavior is still changed for every other instance of dstream
17:00 hmmmm         in any case a global search on the project only turned up those instances
17:01 hmmmm         i just don't get why timestamps need to be enabled by *default*
17:01 hmmmm         when it's clearly the more unpopular option
17:01 Calinou       in software, timestamps should be off by default
17:04 ShadowNinja   hmmmm: Me and est prefer the timestamps.  Calinou: Why?
17:06 Calinou       do you see any other game where timestamps are on by default?
17:06 Calinou       any system program?
17:06 Calinou       eg. GIMP
17:06 nrzkt         ShadowNinja, hmmmm : the PR linked is for colored chat, not colored console
17:06 hmmmm         you and est != the majority
17:07 ShadowNinja   nrzkt: Yes, I already heard so.
17:08 ShadowNinja   Calinou: Er, IDK, OpenSpades?  I don't look at the console a lot in games.
17:08 Calinou       probably a minority in the free game world
17:08 hmmmm         timestamps where they're uncalled for (i.e. debugging) do nothing but add extra bloat and noise to the output that's going to be ignored
17:08 Calinou       Xonotic, all Cube 2 games, Quake-based games don't
17:08 ShadowNinja   It's definitely useful in minetest though.  I've used it to check, eg, when a crash occured.
17:08 Calinou       then add them as an option
17:08 Calinou       timestamps = true in minetest.conf
17:08 Calinou       with timestamp_format optionally
17:09 est           so what about disabling the timestamps by default, and make them optional?
17:09 hmmmm         i think the way timestamps work right now are fine
17:09 hmmmm         they don't need to be modified
17:09 est           first as invocation parameters, then also if people want it as options in minetest.conf
17:09 est           but then its unified.
17:09 hmmmm         modifying them right now would just serve certain peoples' preferences and generally not add benefit
17:13 est           what would be a more apropriate time?
17:15 ShadowNinja   After I've merged this I think adding log_timestamp_format or similar would be good.
17:54 ShadowNinja   Ugh, I can't fix this dumb Irrlicht logging bug.
17:55 ShadowNinja   Some things, like "Irrlicht engine version... using OpenGL..." are passed to the MyEventReceiver, while others (Mainmenu "Loading texture: ..." and "Quit message received") aren't.
18:00 hmmmm         merging #2654 if nobody else has any problems with it...
18:00 ShadowBot     https://github.com/minetest/minetest/issues/2654 -- Tests: Modularize unit testing by kwolekr
18:00 est           wait
18:01 est           the order is determined by linker?
18:01 est           how that?
18:02 est           ah I see
18:02 est           ok then
18:04 hmmmm         because static initialization
18:04 est           yes I've seen it
18:05 hmmmm         adding unit test dependencies i think is a slippery slope
18:05 est           why
18:05 hmmmm         dependencies in general take additional consideration
18:05 hmmmm         detecting circular dependencies, etc.
18:06 est           I think hard coding the order would be enough
18:07 est           a method which populates the list of registered testbases
18:08 hmmmm         hard coding is eww
18:08 hmmmm         :/
18:08 hmmmm         besides, i'm getting rid of these dependencies
18:08 hmmmm         they're quite nasty tbh
18:08 est           yea
18:08 hmmmm         g_settings should not be used anywhere in unit tests
18:09 est           this is the problem when you use globals
18:09 hmmmm         the idea that unit test behavior is modified by the config file is bonkers
18:09 est           agree
18:11 ShadowNinja   hmmmm: TestMapblock has to be fixed, otherwise seems O.K. (other than the bug where log_set_lev_silence silences ALL log streams.
18:11 ShadowNinja   )
18:19 ShadowNinja   hmmmm: Latest changes (look for the g_log_outputs_stream-related ones) http://sprunge.us/EhSK?diff
18:21 * ShadowNinja changes the digprop_err level to deprecated.
18:23 ShadowNinja   hmmmm: Is that what you want?
18:24 ShadowNinja   hmmmm: For tests you probably want std::cerr, since it doesn't need to be logged and probably shouldn't be sent to the syslog (on android).
18:24 ShadowNinja   Logged to file that is.
18:34 hmmmm         disabling all log streams?
18:35 hmmmm         note those things have been outstanding problems
18:35 hmmmm         obviously unit testing needs work but it can't all be done in a single commit
18:36 hmmmm         do I really want to log to cerr and not stdout?
18:38 Vexyl         probably, so people can do 2>/file/path to save to file or 2>/dev/null to ignore and still see standard output
18:39 hmmmm         we can switch around logging things later
18:40 hmmmm         a pet peeve of mine is when people tell you to modify your pull request based on future tasks or to fix things that were already broken to begin with
18:40 hmmmm         you need to maintain focus
18:56 ShadowNinja   hmmmm: Well, it doesn't really matter, maybe std::cout is more appropriate. but you can easily redirect either if you need to.  I just suggested stderr because that's want the regular logger uses.
18:58 ShadowNinja   Also, sapier really likes doing things like mod_statistics[modname].stats.types[tp].min_us = 0; mod_statistics[modname].stats.types[tp].max_us = 0... when types[tp].min/max_us = 0 would do, which is really anoying because it's so much more inefficient.  He did this with dynamic_cast too.
19:04 hmmmm         I only used dynamic_cast *once*
19:04 hmmmm         *ONCE*
19:04 hmmmm         and that's to filter out non-schematic decorations from the decoration manager
19:04 hmmmm         in general i feel like you need a compelling reason to use RTTI
19:05 ShadowNinja   local r0, r1, r2, r3, r4, r5, r6, r7, r8, r9 = to_register(...) There's this syntax for loading parameter lists into a table...
19:05 hmmmm         in any case, the whole reason why I am cleaning up unit testing is because I want to encourage people to add more unit tests
19:05 hmmmm         I'm personally adding schematic unit tests, objdef unit tests, and maybe some mapgen ones
19:06 hmmmm         looking forward to unit tests for SRP and perhaps the entire player logon sequence
19:06 * ShadowNinja added threading tests in his PR :-)
19:06 hmmmm         well
19:06 hmmmm         add it in a different PR please
19:06 est           you wanted to add the RFC test vectors?
19:06 hmmmm         i want to keep PRs small and focused
19:06 hmmmm         est:  you already have them in there
19:06 est           ^
19:07 est           hmmmm, no
19:07 hmmmm         oh
19:07 hmmmm         well yeah add them in then
19:07 est           only in the csrp fork
19:07 est           not minetest
19:07 hmmmm         alrighty...
19:07 est           the issue is, we need some sha-1 implementation
19:07 hmmmm         pushing...
19:07 hmmmm         yea I realize
19:08 kahrl         sha1.cpp?
19:08 est           yes
19:08 est           but I have to adjust the srp.cpp
19:08 est           for that
19:08 est           perhaps when I have time.
19:09 est           when you have time, the PR is ready for review, only known bug is that empty passwords don't work
19:09 hmmmm         alrighty pushed
19:09 est           (yet)
19:09 hmmmm         what do you guys think about #2651?
19:09 ShadowBot     https://github.com/minetest/minetest/issues/2651 -- fix pathfinder to produce more useful paths. by obneq
19:09 Krock         est, any passwords, not just empty
19:10 est           darnit
19:10 est           you are right
19:10 est           seems to be a regression
19:10 est           they did work once
19:15 paramat       nore, sfan5, ShadowNinja perhaps sometime you could add your thoughts on game#496 to the thread?
19:15 ShadowBot     https://github.com/minetest/minetest_game/issues/496 -- Make flowers farmable, with seeds, instead of the slow uniform-density spread.
19:17 hmmmm         heh
19:17 hmmmm         https://github.com/est31/minetest/commit/3b34eb28f64d384d4bf67efd2d965f93fa32b025
19:17 hmmmm         we may possibly need to bring this exact stuff back when client side modding hits
19:17 sfan5         paramat: i agree with PilzAdam, making everything farmable makes the game more boring
19:17 hmmmm         i still don't know how this is going to work though
19:18 hmmmm         perhaps all handled through lua rpc
19:18 est           ^
19:18 est           I think we should do this with lua if we need it
19:18 hmmmm         all crafting handled through lua?
19:18 hmmmm         well
19:18 hmmmm         we'll figure it out when we get there heh
19:18 est           not all crafting
19:19 est           just the sending
19:19 hmmmm         yeah true.
19:19 est           I don't think that sending all recipes by default would be a good idea for example
19:22 est           but noted
19:22 hmmmm         that PR is totally done?
19:22 hmmmm         looked through and it seems good, let's merge it
19:22 est           yes. I wanted to keep it as non intrusive as possible
19:23 est           if you want we can keep the serialisation in
19:23 hmmmm         now you have me thinking
19:23 hmmmm         i think we can get rid of it, yeah.
19:23 est           ok then
19:26 est           Do you think I should squash it?
19:26 hmmmm         sure, they're closely related enough
19:26 est           ok
19:31 est           I'll keep the serialisation commit separate
19:31 est           if we still need it it can rolled back fast
19:31 hmmmm         good idea
19:33 est           hmmmm, you probably want to add src/unittest/CMakeFiles/ to the gitignore :)
19:33 hmmmm         yeah probably.  every time I do git add "src/*" it re-adds it
19:33 hmmmm         :(
19:34 est           pushed
19:38 hmmmm         sweet
19:40 hmmmm         2633 looks good to me
19:40 hmmmm         i'm pushing that now
19:40 hmmmm         #2633
19:40 ShadowBot     https://github.com/minetest/minetest/issues/2633 -- Fix minetest.clear_* breaking the corresponding minetest.registered_* tables by tomasbrod
19:41 est           you fix the style?
19:41 hmmmm         did the style need fixing??
19:41 est           yes
19:41 hmmmm         i thought he just used spaces instead of tabs and then fixed it
19:41 est           no
19:41 hmmmm         i'll check it out
19:43 hmmmm         oh he needs spaces in list[k]=nil
19:44 est           that 2
19:44 hmmmm         what's the other thing though
19:44 est           identation
19:44 hmmmm         ohh didn't see that
19:44 est           in the loop
19:55 est           ok, Krock's bug is fixed now
19:56 est           (was a order mistake because of a new param)
20:01 paramat       okay thanks for looking
20:02 hmmmm         alright
20:02 hmmmm         so what are we doing about #2641
20:02 ShadowBot     https://github.com/minetest/minetest/issues/2641 -- Revert the upper-case PROJECT_NAME nonsense that was part of #2402 by sfan5
20:03 hmmmm         it seems to be divisive
20:03 hmmmm         let's have a vote
20:03 hmmmm         here's my vote: i really don't give a turkey
20:03 est           ^mine too
20:05 hmmmm         let's make a strawpoll
20:06 hmmmm         http://strawpoll.me/4215136
20:12 celeron55_    i agree with sfan5 but i also agree with pilzadam's comment about the code
20:12 sfan5         I'm about to correct that
20:23 sfan5         *sigh*
20:23 sfan5         turns out "that" commit was committed
20:23 sfan5         more reverting work for me..
20:32 hmmmm         well
20:33 hmmmm         whatever you do, just do something 8)
20:34 sfan5         #2641 should be good to merge now
20:34 ShadowBot     https://github.com/minetest/minetest/issues/2641 -- Revert the upper-case PROJECT_NAME nonsense that was part of #2402 by sfan5
20:34 sfan5         now we wait for travis..
20:49 hmmmm         oh, looks like shadowninja's side made a comeback
20:49 hmmmm         it's now at 3/3
20:57 est           now its 4/3 for sfan5
20:57 sfan5         ok, the PR is not _really_ ready to merge
20:57 hmmmm         i wonder if either side is manipulating the votes
20:57 sfan5         now*
20:57 sfan5         also that's exactly why strawpoll isn't a good idea for this
20:58 hmmmm         haha
20:58 hmmmm         it gives a rough idea
20:58 hmmmm         i didn't think it'd be a very big deal honestly
21:02 sfan5         so.. do we merge it or not?
21:41 ShadowNinja   sfan5: No :-P
21:42 * ShadowNinja voted once, but could vote about 6 times with all the servers he has access to.
21:46 jin_xi        http://imgur.com/a/yQw50 pathfinder change example
21:53 ShadowNinja   hmmmm: Is logging O.K. to you with g_log_outputs_stream?
23:12 ShadowNinja   multi-socket rebased.  Any comments on it from anyone?  #2604
23:12 ShadowBot     https://github.com/minetest/minetest/issues/2604 -- Add support for multiple listen addesses by ShadowNinja