Time  Nick      Message
06:27 sfan5     merging #11232, #11159 in 10m
06:27 ShadowBot https://github.com/minetest/minetest/issues/11232 -- Add helpful error messages if Irrlicht library or include dir are set incorrectly by sfan5
06:27 ShadowBot https://github.com/minetest/minetest/issues/11159 -- Fix rotation for falling mesh degrotate nodes by Wuzzy2
06:46 nerzhul   sfan5, what do i do for irrlicht#33 then ? restore ? keep it ? :)
06:47 sfan5     i'll take a look at it again
07:09 sfan5     nerzhul: restore the extra methods (e.g. toUTF16_s), delete inside of `#ifndef USTRING_CPP0X_NEWLITERALS` and keep functions outside
07:16 nerzhul   ok will do
07:18 sfan5     pushing http://sprunge.us/RQUSM9?diff in 5m
07:33 nerzhul   sfan5: restored functions :)
07:34 nerzhul   mt compilation is fine
07:51 nerzhul   i'm thinking also on client performance improvement, especially if we add more logic on it. I know i introduced the RenderingEngine singleton, but it as more a easy win path to refactor our original 0.4 code. it's time to make it over i think
07:51 nerzhul   i will start to produce a MR about it
07:58 nerzhul   easy win PR #11234 to review :)
07:58 ShadowBot https://github.com/minetest/minetest/issues/11234 -- fix: drop old irrlicht <1.8 compat on Client::loadMedia by nerzhul
07:59 sfan5     okay looks like toUTF16_s doesn't compile, then we can throw it away
08:03 sfan5     nvm that's easy to fix
08:13 nerzhul   strange it compile on my side
08:13 nerzhul   what is your issue ?
08:14 sfan5     core::string<>.c_str() returns a const T* and that cannot be written to
08:15 sfan5     similar to that
08:15 nerzhul   i compiled with latest gcc and i have no issue :o
08:15 sfan5     clang here
08:15 nerzhul   haha, this time it's the opposite :p
08:19 nerzhul   to continue on the removal on the RenderingEngine singleton, the removal will permit to define who can access to the rendering engine and permit to ensure caller are on the right thread. *
08:19 nerzhul   I'm thinking about decoupling the packet handler part from the rendering part, using event based model to perform irrlicht actions. This will drop frame drop when loading many mapblocks, or having many in memory which are processed by our events directly on packet receive
08:20 nerzhul   i will modify the previous PR if not already merged ('as it's pretty trivial to add a protection on the filesystem pointer to prevent permitting access to it from everywhere in the code first, and will continue commit by commit to protect other parts
08:20 nerzhul   i should find during the refactor some functions which should be moved, or some code parts, to better owners
14:06 nerzhul   i enhanced #11234 refacto PR. We now have better separation and less global calls to the rendering engine. Some few are remaining, and pretty complex to think about currently. Fucntionnally, no change on this PR, but far more proper code, and rendering engine protection
14:06 ShadowBot https://github.com/minetest/minetest/issues/11234 -- refacto: don't permit to access to RenderingEngine::filesystem widely by nerzhul
14:06 nerzhul   i want to enhance our code to isolate it more and more from all our legacy fat Client, ClientLauncher, Game and other objects
14:35 nerzhul   i hope someone can take a look before the weekend, i really want to do some code cleanups on client :)