Time  Nick        Message
08:59 Wuzzy       #10693
08:59 ShadowBot   https://github.com/minetest/minetest/issues/10693 -- Builtin translate (2nd attempt) by Wuzzy2
12:50 Zughy[m]1   can I complaing about how the 2 new core devs did basically nothing? Last Pyrollo MT activity was December 5th, last review November 23th. V-rob has been focusing on his own things, great, but he reviewed 3 PRs in 3 months. MT has a huge problems with PRs and issues, and they're definitely ignoring it. So, for instance, why Pyrollo is a core dev in the first place? I've never seen him here
12:51 Zughy[m]1   * can I complain about how the 2 new core devs did basically nothing? Last Pyrollo MT activity was December 5th, last review November 23th. V-rob has been focusing on his own things, great, but he reviewed 3 PRs in 3 months. MT has a huge problems with PRs and issues, and they're definitely ignoring it. So, for instance, why Pyrollo is a core dev in the first place? I've never seen him here
12:51 Zughy[m]1   * can I complain about how the 2 new core devs did basically nothing? Last Pyrollo MT activity was December 5th, last review November 23th. V-rob has been focusing on his own things, great, but he reviewed 3 PRs in 3 months. MT has a huge problem with PRs and issues, and they're definitely ignoring it. So, for instance, why Pyrollo is a core dev in the first place? I've never seen him here
12:52 sfan5       FYI matrix sends the whole message again when you edit so it just looks like spam from this side
12:53 Zughy[m]1   oh, I'm sorry for that. I'll leave the mistakes next time
12:58 celeron55   eh what, are you saying it would be better to not have v-rob and pyrollo as core devs than to have them doing whatever they are doing?
13:00 Zughy[m]1   I'm saying it doesn't seem Pyrollo is doing anything when he's fresh meat
13:01 Zughy[m]1   and that v-rob could review more things
13:01 sfan5       to the point: aside from "basically nothing" being way too harsh, there is no required minimum level of work you have to put into to be considered an useful addition to the team of coredevs. Also you could say the same about some existing coredevs.
13:03 sfan5       There are bigger problems than PRs and issues (for example that the proposal from months ago that could help here has not been properly discussed and put into action) but either way you can't blame this on two newcomers.
13:06 rubenwardy  if you start requiring minimum hours, then no one will want to become a dev
13:09 celeron55   instead of complaining, you should try to help (or continue trying to help) core devs with the issue and PR problem, and/or maybe aim for becoming a core dev yourself
13:09 MTDiscord   <w​war> ?
13:10 rubenwardy  there's value in pointing out issues, but only if systematic rather than individual lack of time
13:21 Zughy[m]1   <sfan5 "There are bigger problems than P"> wait, which one?
13:22 Zughy[m]1   <rubenwardy "if you start requiring minimum h"> yeah, for sure, this is not a company
13:25 sfan5       can't find the discussion now but the proposal was roughly that upon a PR being opened the devs have 1 week to reject or accept the concept
13:25 Zughy[m]1   <celeron55 "instead of complaining, you shou"> but you see, a normal contributor can't help in full, because it's still up to core devs to accept or reject a PR (and I won't ever be a core dev, I don't have an inch of your knowledge). I could make 10 PR in one day, but if no one checks them out (it's just an example) they'll lie there forever
13:27 Zughy[m]1   <sfan5 "can't find the discussion now bu"> it sounds pretty suicidal
13:27 rubenwardy  <sfan5> can't find the discussion now but the proposal was roughly that upon a PR being opened the devs have 1 week to reject or accept the concept
13:27 rubenwardy  This was part of the roadmap proposal
13:28 sfan5       Zughy[m]1: why is that?
13:29 sfan5       rubenwardy: these warnings are new https://0x0.st/-inl.txt
13:30 Zughy[m]1   if there's a period where only one core dev is present, that guy has to judge every new PR coming by. I saw weeks where there was at least one new PR every day. That could stress the guy and bring to rushed choices
13:30 rubenwardy  that'll be because the file has changed, I haven't touched that method
13:30 rubenwardy  but it should be fixed anyway
13:30 rubenwardy  unless having override in any method in a class enables that warning
13:31 sfan5       well there has to be some kind of consensus, if there's only one coredev available for the entire week a PR cannot be accepted or rejected
13:31 rubenwardy  there are some projects that close PRs if they're not linked to an accepted issue
13:32 Zughy[m]1   imho, it'd be better to say: "roadmap -> no -> one week time, where if no one says nothing it's automatically a no"
13:33 rubenwardy  #10536
13:33 ShadowBot   https://github.com/minetest/minetest/issues/10536 -- Add roadmap by rubenwardy
13:33 Zughy[m]1   so that if a core dev _really_ wants that feature, they can say "fine, let's do that". The difference is that instead of starting from "who knows" starts from "no"
13:35 rubenwardy  So:
13:35 rubenwardy  - Bug fixes are exempt from the roadmap, they're always prioritised
13:35 rubenwardy  - Roadmap PRs are automatically accepted
13:35 rubenwardy  - Non-roadmap PRs must be for an accepted issue or be accepted within a week, else they are closed
13:35 rubenwardy  where "accepted" means "allowed to be left open"
13:36 Zughy[m]1   it'd be nice to change the PR form, having check boxes corresponding to the roadmap, like "goal1", "goal2", "bugfix" and "other", with a warning on the last one
13:38 Zughy[m]1   <rubenwardy "where "accepted" means "allowed "> would it be to harsh if "accepted" would mean "I'll take care of it"?
13:38 sfan5       is the assumption here that only one person would be responsible for accepting it?
13:38 rubenwardy  well, if you accept a non-roadmap PR you should probably review it
13:38 Zughy[m]1   it may help the core dev to understand whether they have the time
13:39 sfan5       I fear that "we want this in principle but I certainly do not have time to review it" cases will be a problem if done like that
13:39 rubenwardy  yeah
13:40 Zughy[m]1   exactly, so that if you don't retain that PR THAT important, you'll just ignore it. And the fault won't be yours, because the roadmap is clear
13:40 Zughy[m]1   with ignore I mean that it'll be closed
13:41 Zughy[m]1   this could also push contributors on working on bugfixes instead (about 320 right now)
13:46 * Zughy[m]1 sent a long message:  < https://matrix.org/_matrix/media/r0/download/matrix.org/ZMBicjaCVjfcxpVmHaqxRqru/message.txt >
13:47 rubenwardy  source of that message:
13:47 rubenwardy  Right, thoughts about the last comment of the roadmap PR? I quote:
13:47 rubenwardy  "what is going to happen to old PRs not aligning with the roadmap
13:47 rubenwardy  1. closed and sayounara
13:47 rubenwardy  2. kept and merged on the go
13:47 rubenwardy  3. kept and merged when/if the next roadmap aligns with them (please no)
13:47 rubenwardy  "
13:48 rubenwardy  I think they should be marked as Roadmap where relevant, and non-roadmap ones would not be subject to the 1 week thing - but can be closed if necessary.  Kinda offputting for contributors though
13:48 sfan5       reevaluated by whoever looks at them and then either 1 or 2
13:48 rubenwardy  yeah
13:50 sfan5       >{"address":"127.0.0.1","port":30000.0}
13:50 sfan5       floating point ports
13:51 sfan5       such advanced technology
13:51 celeron55   frankly if there's not at least someone crying about something, the change wasn't enough
13:51 celeron55   it's like if you brush your teeth and the mirror doesn't get messy, you didn't brush them properly
13:53 rubenwardy  Lua numbers are always floats, it sucks a bit
13:53 rubenwardy  in future versions you do have integers
13:54 sfan5       yeah but write_json could make a difference between 30 and 30.1 just like print() does
13:54 celeron55   i think the most important effect of the proposed method is that it forces communication before making a PR, you can't just fire and forget, so hopefully it gives more value to PRs that way for both contributors and core devs
13:59 rubenwardy  yeah
13:59 rubenwardy  Should the port be floored (?) when saving and loading then?
14:00 rubenwardy  or maybe  assert(math.floor(x) == x)
14:01 sfan5       not that important
14:02 celeron55   could make more problems
14:38 MTDiscord   <w​war> > it's like if you brush your teeth and the mirror doesn't get messy, you didn't brush them properly Lol ?
14:39 rubenwardy  merging #10085 and #10845 in 10
14:39 ShadowBot   https://github.com/minetest/minetest/issues/10085 -- Use JSON for favorites, move server list code to Lua by rubenwardy
14:39 ShadowBot   https://github.com/minetest/minetest/issues/10845 -- Remove dead code by rubenwardy
15:10 Krock       would it make sense to arrange a meeting tomorrow?
15:10 rubenwardy  yeah definitely
15:11 Krock       okay, then I'll prepare the wiki and write an /engine/ post
15:18 Krock       https://dev.minetest.net/Meetings#2021-01-23
15:22 pgimeno     <rubenwardy> in future versions you do have integers  <-- in LuaJIT too
15:23 pgimeno     what are the cons of dropping Lua and supporting only LuaJIT?
15:25 pgimeno     I believe it's an option that should be given serious consideration
15:26 sfan5       uh
15:26 sfan5       I'm pretty sure luajit doesn't have an integer type like 5.3
15:27 sfan5       but it can internally store integers for performance reasons
15:27 pgimeno     maybe not equivalent, but in LuaJIT you can write: a = 1LL
15:27 sfan5       > =type(1LL)
15:27 sfan5       cdata
15:27 pgimeno     it's a ctype, i.e. FFI-related
15:27 pgimeno     yes
15:27 pgimeno     cdata*
15:28 pgimeno     and you can add them and convert number to integer and all that
15:29 pgimeno     there are many servers that can't run on plain Lua for performance reasons
15:30 pgimeno     builtin and the API can benefit from depending on LuaJIT
15:30 sfan5       how?
15:32 pgimeno     https://notabug.org/pgimeno/ffi_accel <-- this is a hack to speed up things, but without the hack, it can be much faster
15:33 pgimeno     using FFI functions doesn't need switching to interpreted mode, they can be compiled
15:34 sfan5       since non-ffi needs to be supported anyway (it might not be available) we don't need to drop plain Lua support for that
15:35 pgimeno     FFI not available? why?
15:35 pgimeno     maybe you're confusing JIT with FFI?
15:35 sfan5       does it work all platforms luajit works on?
15:35 pgimeno     yes
15:35 pgimeno     JIT doesn't, but FFI does
15:36 sfan5       does it work on platforms where you need to disable JIT due to codesigning constraints (iOS)?
15:36 pgimeno     yes
15:36 pgimeno     it runs in interpreted mode
15:36 pgimeno     sadly, in interpreted mode, FFI is actually slower
15:38 pgimeno     https://love2d.org/forums/viewtopic.php?f=4&t=83992&p=237538#p237538
15:46 sfan5       hm
16:30 rubenwardy  easy PR to review, fixes pet peeve:   #10838
16:30 ShadowBot   https://github.com/minetest/minetest/issues/10838 -- Do not close game on escape in error windows by yamanq
16:34 Zughy[m]1   yes please, it's obnoxious
19:44 Wuzzy       rubenwardy: can I ignore the FS() in favor of F(S())? It will require everyone to update their string-collecting scripts which we will probably forget anyway... :/
19:44 rubenwardy  surprised that doesn't support it already?
19:44 rubenwardy  if it doesn't work that way, sure
19:45 rubenwardy  I'll also have to update  conquer and my other mods :'(
19:45 rubenwardy  well actually - shouldn't the script be version controlled in Minetest utils?
19:45 Wuzzy       we dont have any official string-collecting script yet
19:46 Wuzzy       rubenwardy: I recommend this script btw: https://github.com/minetest-tools/update_translations
19:46 Wuzzy       I think this script might actually be good enough for inclusion. maybe not 100% perfect but better than nothing
19:46 Wuzzy       I don't know if Python can be accepted into utils, tho
19:55 rubenwardy  well, perhaps it could be another repo
20:04 Wuzzy       ?
20:04 Wuzzy       Why making things so complicated?
20:04 Wuzzy       everyone will ignore this repo then
20:04 Wuzzy       the bash-completion nonsense can be removed anyway
20:05 rubenwardy  there did use to be python scripts in utils
20:12 sfan5       there is a python script in util/ right now, but that's dev only and not needed by users
20:32 rubenwardy  where does word wrap in StaticText come from?
20:32 rubenwardy  It's quite broken
20:32 rubenwardy  for example, it doesn't support Windows new lines and the support for unicode hyphens is wrong
20:34 rubenwardy  OK, from IRrlicht
20:34 rubenwardy  that explains it
20:37 rubenwardy  It also has an if-statement for is   RightToLeft, and duplicates the code on each side
20:45 rubenwardy  oh right, it wraps in a direction direction but the text is stored as is
20:45 rubenwardy  that's weird
20:45 rubenwardy  RTL should still use the same memory layout, just be rendered the opposite way...
20:46 rubenwardy  Minetest doesn't set RightToLeft to true, so I might delete it for now