Time  Nick         Message
02:31 hmmmm        hey guys, what are your opinions on escaped strings in the config?
02:31 hmmmm        I am thinking of doing it backwards, like instead of a string having special chars escaped, all the special chars are escaped to begin with
02:32 hmmmm        adding multi-line string settings while i'm doing this other thing because why not
02:47 ShadowNinja  hmmmm: Um, I already did that...
02:48 hmmmm        what?
02:48 ShadowNinja  #1684
02:48 ShadowBot    https://github.com/minetest/minetest/issues/1684 -- Add multi-line settings and setting escapes by ShadowNinja
02:49 hmmmm        ah
02:49 hmmmm        this is implemented in a totally different manner though
02:49 hmmmm        in fact, what I wrote so far nukes everything in your commit
02:52 hmmmm        ShadowNinja, this is a lot simpler.  if """ is the value for a config setting all by itself, the next lines will be the multi-lined text until it sees the next """ on a line by itself
03:09 ShadowNinja  hmmmm: Ah.  You could also do the same with single quotes, since they'll probably not be at the start of a value.
03:09 ShadowNinja  Er, single double-quotes (").
03:11 hmmmm        i wanted to use three double-quotes because that's the multiline-text specifier for pytho
03:11 hmmmm        python
03:12 hmmmm        and it's pretty rare
03:12 hmmmm        i'm not bothering with escaping the text because that gets messy and it's not worth it in general
03:21 kahrl        hmmmm: you could use serializeJsonStringIfNeeded from inventory.cpp
03:21 hmmmm        absolutely disgusting
03:21 hmmmm        this is not quite json
03:22 hmmmm        http://fpaste.org/155226/41731774/
03:22 kahrl        that function leaves the string unchanged if there is nothing to escape
03:23 hmmmm        so basically the parser is super simple
03:24 hmmmm        parseConfigObject returns events now
03:24 hmmmm        could be a comment, a key-value pair, a group ({}), a multiline value, etc.
03:25 hmmmm        i also fixed updateConfigFile.  not that it didn't work, but holy crap...
03:27 hmmmm        settings are maps from strings to SettingsEntry, which is a struct with a string or a Settings *.  if the settings * is not-NULL, then it has child settings.  you can get settings from groups by either doing something like:  Settings *mapgen_config = g_settings->get("mapgen");  mapgen_config->get("seed");  or you could do g_settings->get("mapgen.seed");
03:27 hmmmm        yes? no?
03:28 hmmmm        as for setting defaults in settings sub-trees, i haven't figured that one out.  there's no elegant interface there.
03:30 kahrl        imo get() should return string and throw an error if the setting is actually a group
03:31 hmmmm        get() already returns a string
03:31 kahrl        have a getGroup() that returns a child settings object
03:31 hmmmm        oh, whoops
03:31 hmmmm        yeah, sorry.  there's getEntry() now.  get() just returns getEntry().value, and getChildren() returns getEntry().children
03:32 kahrl        sounds good
03:34 kahrl        how do you handle the case where one thread uses a child settings object and another thread deletes it?
03:35 hmmmm        i don't!  user beware, don't delete children Settings objects
03:36 hmmmm        at first I was going to have getChildren return a deep copy of a std::map<std::string, SettingsEntry> but that'd be a bit awkward to handle
03:39 kahrl        what if one thread calls g_settings->set("blah", "something") and setting blah was a group? that would delete the child settings object, right? does every caller have to check for that case?
03:40 hmmmm        that's one piece of behavior i haven't decided on yet
03:41 hmmmm        i think what i'm going to have it do is what it does right now, which is lets it succeed, but doesn't get written back to the config file
03:41 hmmmm        or maybe write two entries; blah = something \n blah = { ...child settings here... }
03:42 kahrl        I prefer the second version
03:42 hmmmm        same..
03:53 ShadowNinja  Groups would be great -- just make it accessible from Lua too.
03:53 ShadowNinja  And if \n and friends aren't escaped in keys throw some kind or error if you try to use them.
03:54 ShadowNinja  (That was a security issue in my security branch)
03:54 hmmmm        security issue?
03:55 ShadowNinja  Seems like only [a-zA-Z0-9_-.] should be allowed.
03:55 hmmmm        maybe from lua... i dunno...
03:55 ShadowNinja  hmmmm: minetest.setting_set("foo=true\nsecure.enable_security", "false")
03:56 hmmmm        multiline is for values, not keys
03:56 ShadowNinja  Causes a problem when reloaded.
03:56 hmmmm        ahhhh
03:56 hmmmm        i see what you're saying now
03:56 hmmmm        that's clever
03:57 hmmmm        though, isn't that true with the current Settings?
03:57 ShadowNinja  hmmmm: Yes, just while you're at it.
03:57 hmmmm        security in settings deserves a closer look
03:59 ShadowNinja  if (!string_allowed(name, ...) throw UnsupportedOperationException("Only alphanumeric characters, underscores, and dashes allowed in setting keys");
04:00 ShadowNinja  Just something like that in set((.
04:00 hmmmm        have to do the same thing for keys
04:00 hmmmm        erm values
04:00 ShadowNinja  Make foo.bar = true equivalent to foo = { bar = true } too
04:00 hmmmm        also need to check to make sure they're not inserting a multiline specifier
04:01 ShadowNinja  hmmmm: Yes, although I'd just make them escape properly.
04:01 ShadowNinja  (values that is)
04:02 hmmmm        agh the escaping
04:02 hmmmm        no escapes!  escaping strings opens a can of worms
04:03 kahrl        how so?
04:03 hmmmm        more potential security issues, more code complexity, more confusing to look at, etc.  I'd personally rather have my settings a little bit less flexible instead
04:03 hmmmm        if you can't write """ in a piece of multiline text... well then
04:04 ShadowNinja  hmmmm: Alright, sanitize it then.
04:04 kahrl        serializeJsonStringIfNeeded has been around for a long time without issues
04:04 hmmmm        actually
04:04 kahrl        it's also tested
04:04 hmmmm        you can still write """ in multiline text
04:04 hmmmm        it just can't be on a line all on its own
04:05 hmmmm        i'll check out the serializeJsonStringIfNeeded
04:05 kahrl        well, serializeJsonString is tested to be fully exact
04:05 ShadowNinja  hmmmm: Er, so are the leading and trailing newlines stripped?
04:06 hmmmm        not for multiline text
04:06 hmmmm        http://fpaste.org/155226/41731774/
04:07 hmmmm        actually what i pasted is a tiny bit inaccurate.. the closing """ would not be indented otherwise that line would be treated as another part of the multiline string
04:08 ShadowNinja  hmmmm: So that produces "\ntext begins here....\n"?
04:08 hmmmm        no beginning and trailing \ns
04:08 hmmmm        intermediate newlines are kept
04:08 hmmmm        so that text would be:
04:08 ShadowNinja  I'd prefer if leading indentation was stripped, but that would be trickier.
04:09 hmmmm        "text begins here\nwhitespace is not removed\n   in a multiline string"
04:09 hmmmm        see
04:09 ShadowNinja  hmmmm: Alright.
04:09 hmmmm        I considered stripping the leading indentation
04:09 hmmmm        sorta wonder if I could put an option for that in a config-meta-config
04:10 hmmmm        i also had another idea for config directives
04:10 hmmmm        @include "somefile.conf"
04:10 hmmmm        @multiline_indentation off
04:10 hmmmm        or something
04:10 ShadowNinja  hmmmm: Just strip the indentation in front of the setting declaration, if any.
04:10 kahrl        didn't you say that @include makes updateConfigFile almost impossible to write?
04:10 hmmmm        not impossible
04:10 hmmmm        just more difficult than it's worth
04:11 hmmmm        including other files is pretty useless
04:11 hmmmm        if I were REALLY bored I guess i'd put it in but i have so many better things to do
04:12 ShadowNinja  It's usefull if you have more than one server, but I can see how it would be much more complex.
04:13 hmmmm        you'd need to keep track of which settings originated from where
04:15 kahrl        you could probably get away with not overwriting the included files and appending changed settings to the end of the main config file
04:16 ShadowNinja  Just changed ones, yes.
04:17 hmmmm        i guess that's true
04:17 hmmmm        heh that's dirty but it works
04:20 ShadowNinja  Another thing to do to Settings:  Make them typed so you don't have to stoi/strtof/is_yes/etc on every get.  I'd convert all settings that weren't obviously numbers or booleans to quoted strings to remove ambiguity (1337/"1337"/true)
04:21 ShadowNinja  Possible issue: some people use 1 instead of true/yes/on.
04:22 Zeno`        Sounds pretty complicated
04:22 ShadowNinja  It would add a fair amount of complexity though.
04:23 hmmmm        ShadowNinja, that's way outside the scope of this
04:23 ShadowNinja  Zeno`: You said as as I typed ^.
04:23 hmmmm        I originally wanted to do that but everybody told me it was worthless
04:23 ShadowNinja  hmmmm: Yes, just saying.
04:24 ShadowNinja  hmmmm: IMO you should use getGroup instead if getChildren, it sounds better.
04:24 ShadowNinja  (If I understood you)
04:25 hmmmm        I was debating between getSubtree, getChildren, getBranch, and getGroup
04:26 ShadowNinja  subtree -> no, sounds complicated | branch -> not descriptive of what it is | children -> meh, just doesn't sound right to me
04:28 ShadowNinja  Also, Settings isn't necessarily a RB tree, it could be a hashmap, or even a vector, that's an implementation detail.
04:29 hmmmm        it becomes a tree in concept with this new feature though
04:30 ShadowNinja  I see, but it still sounds complicated.  Use simple words where possible.
04:31 ShadowNinja  But I'll be off to sleep now.  o/
06:46 Zeno`        Can anybody explain what https://github.com/minetest/minetest/commit/d0be27415eb9386b2367ab798268da6fbe8b7a68 actually does?
06:47 Zeno`        It seems to be the source of https://forum.minetest.net/viewtopic.php?f=3&t=10655&p=163167#p163167
06:47 Zeno`        s/source/cause
06:51 Zeno`        https://gist.github.com/Zeno-/a6b5191d2190d053f219
07:06 Calinou      https://github.com/minetest/minetest_game/pull/350
07:10 Calinou      https://github.com/minetest/minetest_game/pull/351
07:16 Calinou      https://github.com/minetest/minetest_game/pull/352
07:19 Calinou      https://github.com/minetest/minetest_game/pull/353
07:19 Calinou      they can all be tagged as low-priority enhancement
07:19 Calinou      except the tools one, which can probably be enhancement
08:12 Calinou      https://github.com/minetest/minetest_game/pull/354 – 3D signs
08:38 Krock        Hello world.
09:45 jack_desktop Howdy, I'm currently wondering if there is a wiki or something that goes through the steps of developing what is now Minetest? I assume what I'm asking may not exist though, just thought I would ask, would love to know how Minetest was created from very basics to what it is now.
09:46 jack_desktop I know there is the dev.mintest.net wiki, does that fit the question, currently reading through bits of it now actually.
09:49 sfan5        i don't think we have a wiki entry for that
09:50 jack_desktop Ahh okay, thanks.
10:29 sapier       any final thoughts on https://github.com/minetest/minetest/pull/1561 ? I'm gonna merge it in a few hours
10:31 Zeno`        sapier, regarding #1885, my latest commit exposes the g_core_settings pointer as const, so I'm not sure how things can get out of sync. This also eliminates have to "sync back" to g_settings when the class is destroyed (i.e. all sets must be done through g_settings())
10:31 ShadowBot    https://github.com/minetest/minetest/issues/1885 -- Create CoreSettings class for time-critical sections to use by Zeno-
10:32 sapier       Zeno what happens if someone changes the setting via directly accessing that member
10:32 Zeno`        Which member?
10:32 sapier       +	bool enable_shaders;                                      +	bool enable_fog;                                      +	bool doubletap_jump;                                      +	bool node_highlighting;
10:32 Zeno`        they can't
10:32 sapier       for example
10:32 sapier       why not?
10:32 Zeno`        it's const
10:33 Zeno`        the whole external reference (pointer) to the core_settings object is now const
10:33 sapier       ok so they couldn't without using const_cast
10:34 sapier       so only the unmaintainable shadowcopy remains
10:34 Zeno`        well, yes, but if people are going to do that... *shrug*
10:34 sapier       ppl are doing that
10:34 Zeno`        why?
10:35 sapier       because they don't see how they're supposed to change the value and will just thing you've ben over-cautious when making it const
10:35 sapier       having two central copys of variables feels quite bad and I'm almost 100% sure this will cause errors in future
10:36 Zeno`        At the moment there are more that two copies of this critical variables
10:36 sapier       any "protection" you add can be easyly removed by someone not knowing the consequences
10:36 Zeno`        every class makes their own copy
10:36 Zeno`        of course protection can be removed :)
10:36 sapier       not here ain't more then one central copy
10:37 Zeno`        I can even access private members of a class, but I think if something went wrong it'd be my own fault
10:37 sapier       100 local copys which are controled by that local instance are better then two global stores to be kept in sync
10:37 sapier       everyone usually looks for the first way to do something and wont continue to look for another one once this seemed to work
10:38 Zeno`        ?
10:38 Zeno`        this first way to do something is cast away const?
10:38 sapier       It's useless to discuss who's fault it would be as it's allways gonna be someone else who has to fix it ;-)
10:38 Zeno`        Anyway, I guess I just can't think of a better way :/
10:39 sapier       what's wrong with using the callback mechanism I suggested?
10:39 Zeno`        I'm not sure how it's different to what I've implemented
10:39 sapier       the difference is the lack of second central store
10:40 Zeno`        So every function that want to keep copies of these values would register a callback?
10:40 sapier       we're in c++ so it's more every class but that's a semantic detail
10:40 Zeno`        err, yes, that's what I meant
10:41 sapier       basically yes
10:41 Zeno`        well, that's easily done now using my callback class
10:41 Zeno`        I guess I could modify it
10:42 sapier       As I said your callback mechanism is quite good, the minor change I'd like would be a way to pass a parameter to a callback
10:42 jin_xi       hey Zeno, did you read the logs? there was some discussion regarding settings whey you were off
10:42 sapier       I'll push the font_engine today there's a more simple callback mechanism for settings. I'd like to replace this by your one in next step
10:42 Zeno`        I really don't see much of a difference though ;) An advantage, though, of the current method is the actual place where "perform the callbacks" can be isolated to somewhere it's safe(r)
10:43 Zeno`        jin_xi, do I dare read it?
10:43 Zeno`        (s/the current method/this patch)
10:44 jin_xi       http://irc.minetest.ru/minetest-dev/2014-11-29#i_4043326
10:44 Zeno`        All the callback does is set a dirty flag... the actual updates are done in a place where syncronisity shouldn't be an issue (as far as I can see)
10:44 Zeno`        The font patch looks good
10:44 Zeno`        celeron55, has misunderstood
10:45 Zeno`        there is no necessity to add code in 10 different places :/
10:45 jin_xi       read on
10:45 Zeno`        I have no idea where he got that idea (nor kahrl)
10:45 sapier       looks like another variant of callbacks to me
10:45 Zeno`        templates?
10:46 * Zeno`      sighs
10:48 sapier       :-) well I don't like templates to much too
10:49 Zeno`        I don't think celeron55 actually read the code based on his comments :/
10:49 Zeno`        meh
10:49 Zeno`        sapier, I haven't tried the font thing for about a month, but I tried it when you first made the PR
10:50 sapier       I added Zeframs commit about a week ago so there might be a major difference
10:50 Zeno`        When are you merging it?
10:50 sapier       my part is only the basics, zefram fixed the formspecs
10:50 sapier       in about 5-6 hours
10:51 Zeno`        I'll test before then
10:51 sapier       thanks
10:51 Zeno`        If I don't find anything it will be because I found no issues and/or fell asleep :)
10:51 Zeno`        If I don't say anything it will be*
10:52 sapier       well there will be changes, quite reasonable as now font scaling is added :-)
10:56 nore         ./minetest --address digitalaudioconcepts.com --port 30001 --name nore --password <password> --go --config minetest.conf <-- if "video_driver = null" in minetest.conf, then I get floating point exception right after connecting
10:57 sfan5        nore: fixed in freeminer
10:57 sfan5        </troll>
10:57 nore         lol
10:57 sfan5        nore: suggestion: gdb it, fix it and pull request
11:00 nore         SIGFPE in CNodeDefManager::fillTileAttribs
11:07 Zeno`        sapier, where is doCallbacks() called?
11:17 Zeno`        I'm not sure if people are going to add settings if they have to register a callback for every setting they're interested in :p
11:18 Zeno`        sapier, are the valgrind leaks false positives? (in #1561)
11:18 ShadowBot    https://github.com/minetest/minetest/issues/1561 -- Implement proper font handling by sapier
11:50 sapier       Zeno`: for what I remember doCallbacks is called on setting a setting
11:51 sapier       valgrind leaks? I'll check
12:02 sapier       Nope no false positives. font is grabed once to often so it's not deleted on shutdown, fixed
12:03 sapier       It'd be a shutdown leak only but better to have no leak at all
12:03 Zeno`        That's the only issue I experienced. The only other (very minor) one was that on the pause screen my path was chopped off at the bottom
12:04 sapier       is this a new issue? I remember this beeing told multiple times by now? maybe we need to implement some sort of size check for it
12:06 Zeno`        I don't know if it's new... I just happened to notice it
12:07 Zeno`        I can't check if it's a current issue or not because I'm fixing merge conflicts in another one of my branches :(
12:08 sapier       no problem if it's new I'm gonna address it later guess that's nothing mission critical
12:10 Zeno`        yep, very minor as I said (just cosmetic)
12:16 PenguinDad   the path being cut off isn't a new issue for me
12:18 Zeno`        It's new for me (just tested), but *shrug*
12:26 kilbith      https://github.com/minetest/minetest_game/pull/327
12:26 kilbith      https://github.com/minetest/minetest_game/pull/338
12:26 kilbith      nore, sfan5 ^
12:27 kilbith      then we'll done with the textures for 0.4.11
12:30 sapier       What about a feature freeze as of about 10th december to be able to do a release for christmas?
12:31 nore         seems good
12:32 sapier       but means that everyone who wants things in for this release has to get it polished till next weekend
12:32 nore         together with the addition of snow biomes & other snow related things to minetest_game
12:32 nore         yep
12:34 sapier       I'm gonna check the android pulls and issues, hopefully most of them can be merged. My first check was quite positive but without real test I can't be sure
12:34 sapier       And I don't know how much was broken through other commits since last android build ;-)
12:35 sapier       -through + due to
12:45 rubenwardy   https://github.com/minetest/minetest/pull/1603
12:45 rubenwardy   "Use frame instead of words 'no screenshot available"
12:45 rubenwardy   The build doesn't fail, Travis is lying
12:45 rubenwardy   (I only changed an image)
12:46 PenguinDad   rubenwardy: the build failing isn't caused by your change
12:46 rubenwardy   Yeah. It is something to do with wrong version of irrlicht (no matching call to irr:: blah blah)
12:47 PenguinDad   you just forked at a point where the build was failing
12:48 sapier       rubenwardy: just rebase to latest version
12:53 rubenwardy   Done
12:53 sapier       thanks
13:03 rubenwardy   Travis passed
13:14 Zeno`        sapier, what if I close my PR and just work on the callback class
13:14 Zeno`        would that be more digestible?
13:14 Zeno`        and celeron55*
13:14 sapier       good Idea, if you want to merge it I suggest waiting till font_engine is merged as you can replace that callback mechanism by yours too
13:15 sapier       -merge +split
13:16 Zeno`        Yeah, I think that's the way to go because it doesn't cause arguments and it can be used with whatever method is finally agreed upong
13:16 sapier       yes
13:18 Zeno`        If you want to cherry pick those two files (callback.cpp|.h) before I get around to it, feel free :p It's almost midnight here
13:19 Zeno`        Also, the data type passed... std::string is good I think (to pass the setting name) but should there be user data as well?
13:19 Zeno`        Actually it doesn't matter
13:19 Zeno`        std::string is fine
13:20 sapier       well we'll have to think about how to pass parameters in a generic mechanism. that's an open issue true
13:20 Zeno`        yes
13:20 sapier       I'd prefere to avoid passing a void* pointer ;-)
13:20 Zeno`        same
13:21 celeron55    one way to do that kind of thing is to make an abstract base class for callback parameters and dynamically casting it when receiving it
13:22 celeron55    then it's a matter of preference whether to add template functions for getting and passing it directly as the final type or leaving that to the caller and callback
13:22 celeron55    the class can be completely empty, C++ will store the type of it
13:22 Zeno`        quite possibly. But depending on what happens the name of the setting might be enough. They (the settings) don't change that much after startup, though. But that's an idea worth exploring
13:23 Zeno`        I'm not sure about templates (only because I find them ugly... I am biased)
13:23 celeron55    for a simple purpose like that they don't look bad
13:24 Zeno`        quite possibly yeah
13:24 celeron55    but they will cause weird errors when used wrong in any complexity
13:26 sapier       well the first version (font_engine) one has built in callbacks. I think we should design the generic mechanism in a way it's most likely gonna handle all situations where we may need callbacks in future. Let's better take some time then rush into something we have to fix soon again
13:27 Zeno`        I think callbacks are the way to go. I know there is that thing called, umm, main_game_callback or something, but it's not very useful atm (it might be client only as well)
13:28 sapier       and of course once we have a solution we should replace ALL callback style things in mt by this one ;-)
13:29 Zeno`        yeah
13:37 Zeno`        #1889 ... feel free to modify/adapt/whatever. I need to rest (spent half the afternoon bisecting stuff heh)
13:37 ShadowBot    https://github.com/minetest/minetest/issues/1889 -- Proposed callback manager. by Zeno-
13:37 sapier       good night zeno
13:37 Zeno`        I'll hang around, but night ;)
13:37 Zeno`        night celeron55
16:18 sol_invictus I get a lot of messages like: Got packet command: 60 for peer id 642 but client isn't active yet. Dropping packet
16:18 sol_invictus What could be the reason?
16:19 sapier       most likely some old client connects
16:19 sapier       maybe a andoid fork
16:20 sapier       there's been a lot of issues with client joining about a year ago sadly we couldn't fix them backwards for those old clients
16:21 sfan5        nore: did you already agree to #327?
16:21 ShadowBot    https://github.com/minetest/minetest/issues/327 -- Localplayer check position by MirceaKitsune
16:22 sapier       how often do you see those messages and on which loglevel sol_invictus, maybe we need to lower the level where that messages are shown
16:23 sol_invictus I see this in server console with --quiet
16:24 sol_invictus it's ERROR[ServerThread]
16:24 sapier       did anyone check how much cpu load 327 gnerates?
16:24 sfan5        sapier: talking about minetest_game
16:24 sapier       sol_invictus: hmmm guess we really should lower the loglevel for that entry
16:24 sapier       ahhhh wrong link again :-)
16:25 sapier       I wonder if we have that code in ...
16:25 sapier       well I don't wonder enough to check it
16:28 nore         sfan5, yes
16:32 sfan5        nore: I'l merge it then
16:32 nore         ok
16:37 CraigyDavi   Anything stopping https://github.com/minetest/minetest_game/pull/345 from getting merged?
16:38 sfan5        nore: ^
16:39 nore         no, nothing preventing merg
16:39 nore         e
16:39 nore         sfan5, your thoughts?
16:39 Zeno`        well, I don't see why de-indenting the block mentioned can't be done now
16:39 Zeno`        why make it a separate PR?
16:40 sfan5        nore: the PR looks good, but Zeno` is right
16:40 sfan5        CraigyDavi: can you do what Zeno` suggested?
16:40 CraigyDavi   Ok I'll de-indent it
16:40 Zeno`        I'm ok with it apart from that
16:40 nore         sfan5, btw: your thoughts on the pine-* pull?
16:41 sfan5        nore: link?
16:41 nore         https://github.com/minetest/minetest_game/pull/336
16:41 sfan5        nore: pine PR is ok
16:42 nore         ok, and should we do it all at once, or add tree models & mapgen at the same time?
16:44 sfan5        nore: tree models & mapgen later
16:44 sfan5        nore: can i merge 336? (pine-pr)
16:44 nore         yep, merge it
16:45 Zeno`        yeah
16:45 Zeno`        nore, but I thought you wanted it for a Christmas release :P~
16:46 CraigyDavi   #336 doesn't have any sapling growth code though?
16:46 ShadowBot    https://github.com/minetest/minetest/issues/336 -- Settings checkboxes not saved if you quit without first leaving the Settings tab
16:46 nore         no, not yet
16:46 sfan5        Zeno`: later doesn't imply after christmas
16:46 sfan5        CraigyDavi: correct
16:46 nore         but it's easy to import the code from mg
16:47 CraigyDavi   Updated minetest_game/pull/345 with de-indentation
16:47 Zeno`        sfan5, my bad. I thought he said he wanted it as part of the Christmas release. I sometimes imagine things (and I was joking anyway) :)
16:47 nore         sfan5, btw: what about adding code to get snow biomes in mgv6? (so that would be lua mapgen)
16:47 sfan5        nore: can i merge the new 345?
16:48 nore         sure
16:49 sfan5        ok
16:50 sfan5        CraigyDavi: error: mods/bucket/init.lua: patch does not apply
16:50 CraigyDavi   Oh that's due to https://github.com/minetest/minetest_game/commit/e707ba3
16:50 nore         kaeza fixed a few globals too IIRC
16:51 CraigyDavi   Well it was in my PR before kaeza...
16:51 sfan5        CraigyDavi: can you rebase it?
16:51 CraigyDavi   Yeah I'll remove the bucket part
16:52 sapier       I'm preparing https://github.com/minetest/minetest/pull/1561 for merge now so last chance to mention issues prior merge
16:56 sfan5        sapier: why XML?
16:56 sapier       ask irrlicht guys that's the way they do fonts
16:57 sapier       our old font was legacy only
17:00 sfan5        sapier: 1) see github comments 2) who agreed to that pull?
17:01 sapier       all the comments have been handled
17:01 sapier       unless there's been a new one within the last 10 min ;-)
17:02 sfan5        i made some new ones
17:02 CraigyDavi   sfan5, rebased
17:03 sapier       its desktop only because android for example provides valid dpi information automaticaly
17:03 sapier       would you agree to replace "PC/MAC" to "Desktop"
17:03 Zeno`        irrlicht probably uses XML because of fontconfig
17:04 sfan5        Zeno`: no
17:04 Zeno`        no?
17:04 sfan5        the irrlicht xml files contain which chat is where in the png
17:05 sapier       I don't know why irrlicht uses those fonts, and to be honest I don't care because I can't change it
17:05 sfan5        that has nothing to do with fontconfig
17:05 sapier       sfan5 is "Desktop" ok?
17:05 sfan5        yes
17:08 sapier       sfan5 only those 3 comments or don't I see the other ones yet?
17:09 sfan5        sapier: no, thats the only one
17:10 sapier       ok in this case I'm gonna merge it. hopefully our formspec scaling issues are gone after this
17:11 kahrl        ugh
17:11 kahrl        I started downloading the patch when you said you're going to merge it to try it and it's not even downloaded yet
17:11 kahrl        that's how bloated those XML files are
17:12 kahrl        why not just drop non-freetype support?
17:13 kahrl        for comparison, the current minetest working tree is about 44K, while 1561.patch is 5.08M
17:14 kahrl        ok, download is finished now
17:15 kahrl        still, requiring everyone who wants to git clone minetest to download this stuff (which they even probably won't use, as they'll enable freetype) is bad
17:16 kahrl        especially since the download rate at github.com is so low
17:18 kahrl        17.5KB/s according to wget
17:18 Calinou      <+kahrl> why not just drop non-freetype support?
17:19 Calinou      I agree, it is worth it
17:19 Calinou      (we could do the same for gettext.)
17:19 Calinou      LANG=C forces english.
17:21 kahrl        well supporting USE_GETTEXT=0 is quite trivial
17:21 Calinou      gettext should be enabled by default
17:21 Calinou      I don't like having to use cmake . -DENABLE_GETTEXT=1 -DENABLE_FREETYPE=1 -DENABLE_LEVELDB=1 -DENABLE_REDIS=1
17:21 kahrl        just do #define gettext(String) String  (as is done in gettext.h)
17:22 Calinou      (LevelDB and Redis should be used if found on system, that's all)
17:22 Calinou      (and could be explicitly disabled if found)
17:22 PenguinDad   Calinou: Redis is enabled if it's found
17:22 Calinou      but why not LevelDB?
17:24 kahrl        how often do you actually run cmake? I never have to do it
17:24 Calinou      a few times a month since I make builds I publish
17:24 Calinou      (and I strip them of useless files by hand…)
17:25 kahrl        sounds like a job for a script
17:27 Zeno`        I run cmake maybe 12 times a week :3
17:28 PenguinDad   I run cmake 4 times a week :D
17:29 kahrl        I feel bad for neglecting my cmake now ;)
17:29 sfan5        Calinou: xyz wanted me to enable redis by default
17:31 Calinou      both should be enabled if found
17:31 Calinou      trying out my script… (I'll publish it when done)
17:32 kahrl        is it as simple as changing OPTION(ENABLE_LEVELDB "Enable LevelDB backend") to OPTION(... 1)?
17:32 sfan5        kahrl: i think it is
17:32 kahrl        if so then I support it
17:34 Zeno`        Sort of related, why are debug builds built using -O1?
17:35 kahrl        they would be way too slow otherwise
17:35 kahrl        unplayable for most
17:35 PenguinDad   Zeno`: because -Og is pretty new?
17:35 Zeno`        PenguinDad, what about -O0
17:36 Zeno`        kahrl, with no optimisation (compared to -O1) I find very little difference
17:36 Calinou      what is your hardware?
17:37 Zeno`        Calinou, not really relevant because there is very little difference
17:38 Zeno`        But if you must know, Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz
17:39 Zeno`        I run debug builds with no optimisation on my laptop as well without -O1, though
17:40 Zeno`        If there is a huge difference between -O0 and -O1 there is a big problem (IMO)
17:42 sapier       kahrl you know that everyone not wanting to use freetype has to download those fonts too ;-) well true ttf files are only 5 mb compared to 10mb for the other ones. but I'm quite sure the xml files are heavyly compressed for transmission
17:42 sapier       if you really wanna debug you have to set it to -O0 anyway, -O1 is quite useless for gdb debugging
17:45 sapier       kahrl yes we could drop non freetype support. I'd not do this for two reasons: 1) Freetype code isn't anything different except it creates those "files" otf ... and second the code doing this is a nightmare, it's buggy as hell misses all checks and we don't have anyone who understands it
17:46 sapier       if irrlicht would have native freetype support I'll be first to agree dropping non freetype support
17:46 kahrl        what are those bugs?
17:47 sapier       try setting your font file to something invalid ;)
17:48 kahrl        it falls back to the builtin irrlicht font
17:48 sapier       If I remember correct zefram mentioned some font parameters aren't exactly correct too, He did do special handling for this for his formspec fixes
17:48 sapier       something invalid not something not existing ;-)
17:49 kahrl        same thing
17:50 kahrl        "works for me" (tm)
17:50 sapier       well I don't know what you're doing different but I got various out of bounds memory accesses because the code did interpret a png file as ttf
17:52 sapier       And as I don't have any knowledge about fonts so I can't fix it
17:55 kahrl        maybe older versions of freetype didn't reject invalid files?
17:55 kahrl        dunno
17:55 sapier       is there any additional concern left except the size issues? I don't say they don't exist but that's the price to pay if we wanna have full scaling support
17:55 Calinou      font size could be slightly larger by default
17:55 sapier       it's crashing outside of freetype withing our own code
17:55 Calinou      also make shadow alpha 255 instead of 128
17:56 Calinou      makes font easier to read
17:57 sapier       I don't have any preferences about the alpha does anyone agree to changing this?
17:58 sapier       current size and alpha was checked by various ppl so I'd only change it if there's more then one person saying same
17:59 Calinou      do these people play the game on a regular basis?
17:59 sapier       I think so
18:04 sapier       But as I said I don't really have a preference
18:26 sapier       ok I'm pushing it now, we can change those parameters later if there's a majority for it
18:28 Calinou      thanks
18:29 sapier       thanks? I didn't change them by now
18:30 sapier       but if you want to change them I wont stop you from doing it ;-)
18:44 kilbith      sfan5: I still need your vote for the new header : https://github.com/minetest/minetest_game/pull/341
18:44 kilbith      thanks again
18:54 shadowzone   kilbith: I like the last one
19:01 Megaf        sapier: take a look https://github.com/minetest/minetest/issues/1890
19:01 Megaf        #1890
19:01 ShadowBot    https://github.com/minetest/minetest/issues/1890 -- Copying text from gnome terminal to minetest causes both to crash.
19:02 Megaf        Professional look you say. kahrl
19:02 Megaf        :)
19:07 sapier       is this new or as of font scaling?
19:07 kahrl        I think it's a duplicate of #373
19:07 ShadowBot    https://github.com/minetest/minetest/issues/373 -- Copy-paste doesnt work on GNU/Linux, from another program to Minetest
19:08 kahrl        it's mentioned in the comments there
19:13 sapier       hmm on kde there ain't a crash nor does copy n paste work
19:14 PenguinDad   Megaf: I already noticed this ~6 months ago but thought it's not worth reporting :/
19:14 Calinou      whoa, why were so many files added to repository?
19:14 Calinou      quite heavy ones
19:14 sapier       fonts
19:15 Calinou      that sucks
19:15 Calinou      bitmap fonts should be buried
19:15 sapier       well once irrlicht supports freetype we can do this
19:15 Calinou      we have Freetype support?
19:16 sapier       but irrlicht doesn't have
19:16 sapier       and our freetype support is quite ugly and buggy code
19:16 Calinou      works well enough
19:16 Calinou      handles some Latin characters, better than nothing
19:16 sapier       unless you try your own fonts
19:16 Calinou      also we could have coloured chat :(
19:16 Calinou      I've tried the Droid Sans fonts without issues
19:16 sapier       once you do that youd recognize how broken our code is
19:17 Calinou      I can't type stuff like é in chat, only in console
19:17 sapier       I've read multiple times irrlicht guys want to support freetype maybe there's light at the end of the tunnel
19:17 Calinou      yes, but when will we ship it? it takes years until all distributions package it
19:18 sapier       true
19:18 sapier       so we don't really have options
19:21 kahrl        Calinou: I complained about the huge files (see log) but I was the only one, so I figured no one else cared
19:21 sapier       well none of you seems to have mentioned that the biggest file (with huge difference) is a truetype font
19:22 Calinou      it looks cleaner when you don't ship a lot of random files
19:22 Calinou      but anyway the damage is done, you commited the files, they are in history :P
19:23 sapier       still I can't merge truetypes font generator ;-)
19:23 sapier       -truetype + irrlicht
19:26 sapier       btw according to git size of all those files is 3.8m compressed, less then that single ttf fonts size
19:26 Megaf        we should really migrate to other engine
19:26 Megaf        irrlight is our biggest limitation...
19:26 Calinou      no, just no
19:26 Calinou      we can sure do a lot of things using it
19:26 Megaf        no copy and paste, no proper fonts, lots of bugs
19:26 Megaf        slow development
19:27 Megaf        and so on
19:27 sapier       no too ... any other engine will just have other limitations ... limitations we don't know yet
19:27 Calinou      it would take too much time to change engine
19:27 Megaf        and it's not that lightweight and simple
19:27 kahrl        sapier: couldn't you have compressed that ttf file then?
19:27 kahrl        I mean removed unused glyphs
19:27 Megaf        Calinou: more time than fixing all bugs and implementing all resources?
19:27 sapier       how to know which ones are unused? we do have i18n support?
19:28 kahrl        do we really need the whole kanji set?
19:28 Calinou      yes, way more
19:28 Calinou      the game seems to support Latin and Russian
19:28 Calinou      but remember about Japanese users…
19:31 kahrl        even buildat, at its early stage of development, already had to make changes to its engine (urho3d) to work properly
19:31 Calinou      Megaf, as for copy-paste, there are other doable chat improvements, like history in chat window, chat colours, …
19:32 Calinou      shortcuts (Ctrl + U erases whole line to the left, Ctrl + K erases whole line to the right)
19:32 kahrl        use f10 and those shortcuts work :)
19:35 sapier       btw colored chat could be implemented if someone would care ;-)
19:37 nore         sapier, I thought it was backported...
19:38 kilbith      there's coloured chat in FM
19:38 sapier       did someone check the license yet nore?
19:39 nore         no, but it was said to have been backported from before the license change
19:39 nore         I did not check that though
19:39 sapier       well we have to be sure unless we wanna remove it again ;-)
19:39 sapier       backporting is from fm is usually a bad idea better just look at what something does and implement yourself
19:41 nore         https://github.com/freeminer/freeminer/commits/master/src/chat.cpp <-- chat colors were implemented before license change
19:41 sapier       nice :-)
19:42 nore         and not modified since then
19:42 nore         (at least, chat.cpp did not have a change that affected colors it seems)
19:42 kahrl        while talking about fm: does anyone agree that we should replace the server flags in the multiplayer tab by icons before the next release?
19:43 hmmmm        I agree
19:43 sapier       me too
19:43 nore         me too
19:44 kahrl        ok great, I'll work on it then
19:44 nore         btw, about fm: their "COPYING" file says GPL but the "LICENSE" file says LGPL
19:44 sapier       as there are so many dev online right now what do you think about feature freeze as of about 10th december to get a release till christmas?
19:44 kahrl        sapier: sounds good to me
19:46 nore         already told it looked good, that leaves about 1 week to complete a few features for mt_game too
19:56 nore         sfan5, about what to do in minetest_game before the 0.4.11 release: make pine trees generate and grow, add snow biomes, maybe pine/jungle sticks?
19:56 nore         any other thing I might have missed?
19:56 sfan5        i don't think you missed anything
19:58 nore         should we add an issue somewhere for the TODO, or in the wiki?
20:00 kahrl        I think something in #1561 is droppped too early
20:00 ShadowBot    https://github.com/minetest/minetest/issues/1561 -- Implement proper font handling by sapier
20:00 kahrl        if I switch tabs, I get a segfault on shutdown
20:00 kahrl        s/droppped/dropped
20:00 sapier       are you sure? it's been dropped to late before :-(
20:01 kahrl        I saw that in the log
20:01 sapier       hmm maybe order of deletion is wrong
20:01 kahrl        gdb backtrace is not very helpful (it crashes in ~CGUIEnvironment())
20:03 sapier       can you try moving the fontenginne deletion behind device dropping in main.cpp l1572?
20:03 kahrl        will try
20:05 kahrl        http://sprunge.us/QFdI
20:07 sapier       oh the other way round someone drops the font fontenginge is supposed to handle .. great
20:07 kahrl        oh, forgot to mention this is a build without freetype
20:08 sapier       was this backtrace prior or after modification?
20:08 kahrl        after
20:08 kahrl        I think when you do m_env->getFont() in initSimpleFont you have to grab() the font
20:10 sapier       great ... I just removed that assuming it to be same memory leak as for freetype mode :-(
20:10 kahrl        the irrlicht convention is: if a function name starts with create, you have to drop() the resulting object yourself, otherwise you must not
20:11 kahrl        so drop the result of createTTFont but not the result of getFont
20:11 sapier       hmm my fault, I always mix it up
20:17 sol_invictus what happened to freetype support, sapier?
20:18 jin_xi       sol_invictus: use -DENABLE_FREETYPE=1
20:19 sol_invictus it's always on in my builds
20:19 sol_invictus I can't use custom fonts anymore with HEAD
20:19 Krock        <3 costom fons
20:20 Krock        -o+u+t
20:21 sol_invictus this was taken with droid sans set in config https://i.imgur.com/oBIECfL.png
20:21 sapier       freetype should still be on
20:22 sapier       there's no change in default settings just non freetype mode looks way better then before
20:24 sol_invictus then why doesn't it work anymore?
20:24 sapier       can you give me your settings?
20:24 sol_invictus I reverted your commits and now it's fine https://i.imgur.com/oBIECfL.png
20:25 sapier       still can you give me your settings ;-)
20:25 sol_invictus http://sprunge.us/gYVA
20:26 sapier       do you have a DroidSans.ttf file there?
20:26 sol_invictus of course!
20:27 sapier       can you post debug log of startup somewhere?
20:27 sol_invictus no, won't do this now
20:28 sapier       btw both of your screenshots are same file
20:29 sol_invictus https://i.imgur.com/cD0MGfX.png *
20:30 sapier       doesn't look to me like not working at all but just some slightly different font size
20:33 sol_invictus slightly different?!
20:33 sol_invictus what are smoking? it's tiny!
20:34 sapier       well as mentioned before our freetype code doesn't really provide correct font sizes so zephram had to use some reference which most likely was 800x600
20:34 sapier       you can verify my assumption about the size by setting it to a bigger value
20:34 celeron55    you should aim to preserve font size settings in a maximized full hd window though
20:34 celeron55    that's what people use anyway
20:35 sapier       then next one will complain about not beeing identical in 800x600
20:35 celeron55    but who uses 800x600 for actual gameplay?
20:35 celeron55    sounds insane
20:36 sapier       well we can tune the default size
20:36 celeron55    this isn't a large issue at all but this just seems like the wrong assumption anyway
20:36 sapier       right now for the first time freetype and non freetype sizes are equal (within certain range)
20:36 celeron55    feel free to do nothing, i guess people can just adjust their settings
20:37 sapier       don't get me wrong, I don't have any problem with adjusting the default size, as long as I don't have to change it every day due to someone complaining ;)
20:38 sapier       prior someone mentions this as bug, the strange gui scaling due to y size of window was removed too
20:39 sapier       celeron don't make me lookup the chat log where you did agree to this ;-)
20:40 celeron55    also, to make myself clear, i don't want default font sizes to be preserved; they have been too small for years
20:40 celeron55    (i was only referring to settings that users may have set by themselves)
20:41 sapier       I agree that this is suboptimal but we'd get only one thing correct either freetype or non freetype as both haven't been consistent by now
20:42 sapier       by definition you can't make two things consistent which haven't been without making one different from befor
20:47 sapier       well I'm gonna wait for a little bit more feedback prior fixing the issues with those font changes. I'm quite sure there will be more
20:49 hmmmm        alright done
20:51 hmmmm        agh god damn
20:51 jin_xi       lol
20:53 hmmmm        merge conflict with the proper font handling patch
20:54 sapier       :-) not very surprising
21:00 hmmmm        usually it's not too bad but this needs a lot of manual intervention
21:05 hmmmm        sapier... question
21:05 sapier       yes?
21:05 hmmmm        where are the settings.cpp modifications actually used
21:05 hmmmm        I mean where is doCallback called
21:06 sapier       I'll have a look, hope it's not been lost due to merge error
21:07 sapier       oops
21:08 hmmmm        don't fix it yet!
21:08 hmmmm        woah those are big fonts
21:08 hmmmm        heh it compiles and passes the unit tests
21:08 hmmmm        ok good enough for me
21:09 sapier       it's supposed to be called in the set functions, obviously it ain't
21:09 hmmmm        ironically this callback mechanism was what zeno- was going to do
21:10 hmmmm        hrmmm it's taking a long time to push to github for some reason
21:10 hmmmm        https://github.com/kwolekr/minetest/commit/bc4fbc8567d9a35d463484e128a9b3fa40cf5132
21:10 hmmmm        goot i just realized i forgot something
21:10 sapier       well that mechanism was implemented quite some time before. And that part of zenos fix wasn't denied what was questioned was the shadowcopy of settings
21:11 hmmmm        btw
21:12 hmmmm        from now on, people should turn on their code editor's remove-trailing-whitespace feature no matter how messy it makes diffs
21:12 hmmmm        I'm really getting tired of the trailing whitespace stuff
21:12 sapier       I'm with you ... even I'll make sure it's on again ;-)
21:13 sapier       tell me when I can fix the docallbacks bug
21:13 hmmmm        well
21:13 hmmmm        any concerns about my patch?
21:14 sapier       what's it supposed to do?
21:14 hmmmm        add settings groups and multiline settings entries
21:14 hmmmm        see the test.cpp diff for a sypnopsis
21:16 sapier       I'm not sure about the groups do we really need it? isn't it just a optical change?
21:16 hmmmm        it's part of a bigger change
21:16 hmmmm        i.e. noise parameters
21:16 hmmmm        and map_meta.txt
21:17 sapier       well I guess you wouldn't add it if you didn't need it so if it works merge it
21:17 hmmmm        this way, map_meta.txt contents == g_settings->getGroup("mapgen");
21:18 hmmmm        and also now i'll be able to add noise params as i'd like
21:18 sapier       hmm I wonder how groups do fit the callback mechanism
21:18 hmmmm        okay, fetch from upstream, rebase, and then fix the set doCallbacks thing
21:19 hmmmm        have them callback as well, why not
21:19 sapier       how to call the callback for a parameter within a group ... or can't those be changed on their own?
21:20 sapier       e.g. group1->param1 vs group2->param1
21:20 hmmmm        oh..
21:20 hmmmm        your callback should pass the Settings object it was called frmo
21:20 hmmmm        do you *assume* your callback was from g_settings?  that's not a good idea
21:21 sapier       well by that time there wasn't any other settings object ;-)
21:21 hmmmm        sure there was
21:21 sapier       Zenos callback mechanism would solve that issue
21:21 hmmmm        just not global
21:21 sapier       there was? where?
21:26 hmmmm        well all the *_meta.txt files for one
21:26 hmmmm        mapgenparams is a Settings
21:26 hmmmm        i'm sure there are more
21:27 hmmmm        now we have structured settings!
21:27 sapier       well I don't see a immediate problem as a callback is registred to that specific settings instance.
21:28 sapier       yet I see that right now you can't use same callback to handle different settings instances
21:28 hmmmm        wonder if it would've been a better idea to move to an XML-based config file
21:49 sapier       wow .... months ago mentioning xml would've caused an immediate flamewar
21:50 jin_xi       resignation
21:50 hmmmm        i'm being sarcastic :)
21:51 sapier       hmm could someone write a world -> stl converter? I'd like to try printing a world :-)
21:51 sapier       hmm a small one :-)
21:52 celeron55    just look in the forums, there have been topics about that
21:52 sapier       really? :)
21:52 celeron55    probably better use google because forum search sucks
21:53 sapier       wow there's really a mod to do this :-)
21:53 hmmmm        why would anybody want to waste real-life resources on perlin puke
21:54 sapier       well I'd guess "just for fun" ;-)
21:54 celeron55    looks like these guys did it succesfully https://forum.minetest.net/viewtopic.php?f=9&t=9949
21:54 celeron55    or... this guy
21:54 celeron55    it's only one talking there 8)
21:54 shadowzone   http://picpaste.com/pics/Screenshot-rCli6W1i.1417384474.png  <-- is it suppose to look like this?
21:55 sapier       but I'm missing a photo of a actual print
21:55 * shadowzone waves at celeron55
21:55 sapier       I think so yes
21:55 sapier       I'd prefer it a little bit smaller but that's my personal taste
21:56 kahrl        that's what she said
21:56 * kahrl      hides
21:57 sapier       I wonder how long the size finetuning will take this time ;)
22:20 kahrl        sapier: shall I push the segfault fix?
22:20 sapier       the non grabbed font?
22:20 sapier       yes
22:20 kahrl        ok
23:49 paramat      nore > "sfan5, about what to do in minetest_game before the 0.4.11 release: make pine trees generate and grow ...", yes pine sapling growing code and the pine schematic will be done for 0.4.11, thanks for merging the pine nodes