Time  Nick       Message
16:43 Sokomine   minetest/src/nodedef.cpp:800: void ContentFeatures::updateTextures(ITextureSource*, IShaderSource*, irr::scene::IMeshManipulator*, Client*, const TextureSettings&): Assertion `liquid_type == LIQUID_SOURCE' failed.   (when trying to connect to a particular server)
16:55 rubenwardy they have an invalid liquid definition
16:57 Sokomine   guessed that. it's the zombietest server that will shut down soon (not sure if i've built anything there - don't think so but wanted to check)
16:58 rubenwardy are you using a debug build? Assertions are only enabled for debug builds, so you shouldn't get an error in release
16:59 rubenwardy it's possible that's not a C assertion, though
16:59 Sokomine   no, no debug build on my side this time - just latest git compiled with cmake . -DRUN_IN_PLACE=1 -DENABLE_GETTEXT=1 -DENABLE_FREETYPE=1 -DENABLE_LEVELDB=1
17:00 Sokomine   the server is a bit old (and due to shutdown after all). just unsure if a crash is good or not
19:30 v-rob      Anyone know why formspec element IDs are always offset by 258? Just look at any FieldSpec definition
19:31 Krock      those prior are for hardcoded elements
19:31 Krock      though not really used
19:31 v-rob      Are there any hardcoded elements?
19:31 v-rob      Besides GUIFormSpecMenu
19:32 v-rob      I can't think of any
19:32 Krock      in GUIFormSpecMenu there are AFAIK none, but in other dialogues there are (password change, key change, ...)
19:33 v-rob      That shouldn't affect GUIFormSpecMenu since they are mutually exclusive
19:33 Krock      *shrug*  .. legacy code then
19:38 rubenwardy I think that's because of Irrlicht
19:39 v-rob      Might as well keep it instead of removing it without knowing what we're getting ourselves into.
19:39 v-rob      It still leaves 2147483389 IDs open :)
19:42 Krock      enough to play Doom
19:43 Krock      at a >4k resolution
19:43 v-rob      Heh
19:44 v-rob      Anyone else had the unique experience of trying to put a std::string in a union?
19:45 Krock      how's that supposed to work? use pointers
19:45 v-rob      It has a non-trivial con/destructor, so you have to define your own constructor, destructor, copy assignment, use placement new, forceably call the string's destructor (with the weirdest syntax), etc.
19:46 v-rob      And I thought C++ constructors were supposed to make life easier...
19:46 Krock      when unions don't help, class inheritance is your friend
19:47 v-rob      I'm trying to make something akin to StyleSpec with a key-value storage, but without parsing a string every time an element is requested.
19:47 Krock      see also: struct ClientEvent  (must be a struct)
19:48 v-rob      Parsing a string won't help if we get a JSON replacement
19:50 v-rob      ClientEvent does look like what I'm doing (struct with union and current type), but I don't use pointers to strings
19:51 v-rob      I probably should...
19:51 Krock      FWIW: my plans on redoing the HUD serialization. various types but same read/write function names: https://github.com/minetest/minetest/pull/10245/files
20:00 v-rob      I have this nasty habit of overcomplicating things when it comes to C++
20:00 v-rob      I wish we could use C instead
22:42 pgimeno    Sokomine: it's possible that cmake remembers old configuration flags; you can check all settings with: cmake . -LA (and look for CMAKE_BUILD_TYPE).
23:13 Sokomine   pgimeno: indeed! CMAKE_BUILD_TYPE:STRING=Debug  that might also explain why i got higher cpu usage than i'm used to. how can i reset it?
23:14 pgimeno    just set it to Release
23:14 pgimeno    i.e. cmake . -DCMAKE_BUILD_TYPE=Release
23:14 pgimeno    if you want to reset everything and start anew, I believe it's done by removing CMakeCache.txt
23:21 Sokomine   thank you! that ought to help a lot and speed the game up again