Time  Nick      Message
10:15 ANAND     Any of the null_command_handler entries in the *commandTable arrays can be used for new packet types, right?
15:23 ANAND     Both PlayerSAO and RemotePlayer classes provide getPeerId(). Is there a reason one should be chosen over the other?
15:25 Krock     depends on context. it's the same number anyway in-game
15:26 ANAND     It will be used in script/lua_api/l_object.cpp
15:28 Krock     PlayerSAO. No need to do the way over RemotePlayer there
15:28 ANAND     Alright, thanks
15:36 ANAND     Any of the null_command_handler entries in the *commandTable arrays can be used for new packet types, right?
15:37 ANAND     can be replaced by new packet types*
15:39 Krock     yes
15:39 ANAND     Right, thanks again :)
15:39 Krock     null_command_handler discards the packet, so keep that in mind when providing backwards compatbility
15:40 Krock     but well, that's currently not an issue due to what 5.0.0-dev is
15:41 ANAND     Yes, I was about to clarify that...
15:53 Krock     lhofhansl: #7188
15:53 ShadowBot https://github.com/minetest/minetest/issues/7188 -- Do not crash if the world is read only by HybridDog
15:53 lhofhansl Just found that one. :)
15:54 lhofhansl That one avoids any client only changes. What I have in mind is that a game can proceed, nodes added/removed/etc.
15:54 lhofhansl Perhaps I should play around the rollback.
16:03 lhofhansl I'll file a PR and we can have a discussion there.
16:05 Krock     sounds good :)
16:09 lhofhansl #7544
16:09 ShadowBot https://github.com/minetest/minetest/issues/7544 -- Allow an optional readonly base database by lhofhansl
16:10 lhofhansl Just 10 lines or so of changes.
16:11 lhofhansl Any comments on #7542?
16:11 ShadowBot https://github.com/minetest/minetest/issues/7542 -- Reduce block load glitches by lhofhansl
16:11 lhofhansl Mostly interested in whether this should be configurable or not.
16:12 lhofhansl Options are nice, but MT has sooo many options already.
16:13 sfan5     regarding 7542: IMO we don't need an option for this
16:13 lhofhansl I tend to agree. I will say that this could potentially load a lot of more blocks from the server (especially in huge underground caverns).
16:14 lhofhansl To which I will add that huge underground cavern currently suck big time (because not enough blocks are loaded)
16:14 lhofhansl :)
16:17 lhofhansl So yeah, this is currently broken and we should make it the default and not configurable.
16:38 lhofhansl Can anybody explain why we still need the max_simultaneous_block_sends_per_client setting?
16:38 lhofhansl It defaults to 10, which is way to low, especially when there's latency between the server and client.
16:39 lhofhansl It seems we should just remove this and let the network be the natural limiter. (I tried that by setting this to 1000, and everything is loading faster with bursts up to 100).
23:32 ANAND     {server|client}CommandFactory table has a "channel" field of u8 data-type. What is it and what value should be used for new packets?
23:33 sfan5     pick whatever all other packets have
23:34 ANAND     Alright, and how about reliable (bool)? Should be true?
23:34 sfan5     depends on the packet
23:35 sfan5     but you probably want to set it to true
23:35 ANAND     Ok thanks