Time  Nick       Message
08:59 ANAND      Trivial typo fix: #9917
08:59 ShadowBot  https://github.com/minetest/minetest/issues/9917 -- lua_api.txt: Fix minor typo by ClobberXD
10:10 appguru    rubenwardy: I have replied to your comment to #9907
10:10 ShadowBot  https://github.com/minetest/minetest/issues/9907 -- Shaders as media files (server-sent shaders) by appgurueu
10:31 ANAND      on_player_inventory_action callback passes an InvRef among other callbacks. This InvRef is documented to point to the inventory the player modified, but in reality, it's the player's inventory.
10:31 ANAND      I checked using inv:get_location()
10:35 ANAND      Not sure if this has been the case all along. Should this be fixed? Are there a substantial number of mods relying on this?
10:38 Krock      all inventory actions would be broken if that were true
10:39 Krock      moving in the same inventory results in inventory == player:get_inventory()
10:39 Krock      "take" results in inventory != player:get_inventory()
10:39 ANAND      I came across this while testing player<-->nodemeta transfers
10:41 Krock      I see, yes. player inventory callbacks only provide information regarding the own changes
10:41 Krock      hence inventory always being the player inventory
10:42 ANAND      It's documented to point to the inventory being modified by the player, and that's what I'd logically expect from a callback like this.
10:42 Krock      the documentation and code is correct. everything's from the perspective of the player's inventory
10:42 ANAND      The player's inventory can be retrieved using player->get_inventory()
10:42 Krock      the player puts an item into their inventory, or takes something from it
10:43 Krock      yes, I see it's kinda superfluous
10:43 ANAND      There's no way to determine what inventory the player was interacting with?
10:44 ANAND      It would be possible to add `allow/on_metadata_inventory_*` for certain nodes, but it certainly doesn't make sense to do this for all nodes with inventories
10:45 ANAND      And neither do we have corresponding global callbacks like register_on_metadata_inventory_take
10:45 Krock      there's no "take" information for "put" actions
10:46 Krock      though, all movement actions would optimally contain a InventoryLocation + listname + index for source and destination
10:46 ANAND      Yeah, exactly
10:49 ANAND      Even if we do have global variants of node inventory callbacks, I'd still want to be able to get both the source and dest of player inventory actions. It's just a much more convenient way of presenting the required information, and saves the mod from having to register two callbacks to get two halves of the same information.
10:51 ANAND      All inventory actions are essentially Move actions (within the same inv, or between two different inventories), and as such, should contain both source and destination info. :)
10:53 nephele    always bugged me that in mt picking up an item visually is not an actuall change and inventories can't react to it .. :)
10:53 nephele    would be nice to have a "hand-buffer" inventory or smth
10:54 ANAND      Player inventories do have a "hand" list by default, nephele
10:55 sfan5      picking up an item is not an inventory change? that wouldn't make sense
10:55 nephele    that's unrelated though
10:55 ANAND      Yeah :)
10:55 nephele    sfan5, sure it is, it works that way in the other block game ;)
10:55 ANAND      sfan5: It technically is
10:55 ANAND      Just that there's only one inventory involved
10:55 nephele    affects like stack-spliting and stuff, i also can't immidiently cancel a take action, i have to wait for the player to put it down, which is a bad worse ui
10:55 sfan5      ANAND: indeed, which is why I don't understand nephele's point
10:56 nephele    s/bad/bit
10:56 sfan5      uh
10:56 sfan5      sounds like you have some feature requests regarding the client implementation, not the internals of the inventory system
10:57 ANAND      sfan5: It'd still be nice to have a callback for when that inventory changes. :)
10:57 nephele    why? a transition state should be something i can react to
10:57 sfan5      what is "that inventory" here?
10:57 ANAND      Player inventory
10:57 sfan5      absolutely
10:57 sfan5      and I believe there's both an issue and PR for this
10:57 ANAND      When someone picks up an item, the player inventory is modified.
10:58 ANAND      Yes, there's an issue by Wuzzy
10:58 ANAND      #8390
10:58 ShadowBot  https://github.com/minetest/minetest/issues/8390 -- minetest.register_on_player_inventory_action ignores many inventory changes
10:59 nephele    i mean a transition state like, player picks up an item and puts it into the same place again
10:59 nephele    if i have an inventory slot that only allows taking out items you can still put it back, even though it definetely looks like you took it out
11:00 sfan5      an inventory action only happens when you move the item somewhere else, yeah
11:01 Krock      ANAND: "hand" isn't created by default. mods must create the list
11:01 ANAND      Oh yeah, I remember now :)
11:01 ANAND      Since 5.0, right?
11:01 nephele    sfan5, yeah, that is what i complained about
11:02 nephele    but if there is nothing i can put it "back" to refuse the move back into the slot i can't really do smth :P
11:02 nephele    that back would be a transitional state i think, which would be the item at the cursor
11:04 ne_mo      Hello. I cannot generate a flat map with hills and lakes in minetest-5.2.0. Also in this version I cannot turn mapgen v7 mountains off. This is probably a bug..
11:05 ANAND      This is the issue I was trying to confirm, when I noticed the player_inventory_action oddity: #9918
11:05 ShadowBot  https://github.com/minetest/minetest/issues/9918 -- "Swapping" bypasses inventory callbacks
11:06 sfan5      ne_mo: are you editing map_meta.txt directly
11:06 sfan5      ?
11:07 ne_mo      sfan5: no, I'm doing this with UI in the "All Settings" menu
11:07 appguru    a "flat map" with hills? doesn't make much sense to me
11:07 Krock      5.3.0-dev has those entries in the world creation dialogue btw, if that helps
11:07 Krock      ne_mo: those settings will only apply for new worlds
11:07 sfan5      5.2 has a bug related to that, I'm sure editing map_meta.txt directly will work
11:08 ne_mo      sfan5: thanks, will try.
11:09 ne_mo      appguru: this is the only way to obtain some materials like clay in flat world..
11:09 ne_mo      Krock: yes, I'm creating a new world after the change.
11:12 Krock      will merge #9898 and #9574 in 10 minutes
11:12 ShadowBot  https://github.com/minetest/minetest/issues/9898 -- Modernize include guards and add missing ones by Awkor
11:12 ShadowBot  https://github.com/minetest/minetest/issues/9574 -- Add on_authplayer callback and a last_login parameter to on_joinplayer by sorcerykid
11:13 ne_mo      sfan5: it worked!
11:14 Krock      what was again the reason to keep the "minimal" name for #9450?
11:15 ShadowBot  https://github.com/minetest/minetest/issues/9450 -- Replacement for Minimal Development Test (PR) by Wuzzy2
11:16 sfan5      not directly related: did we have consensus yet to stop shipping minimal with release builds?
11:18 Krock      for release builds, yes I think so
11:18 sfan5      hm I think the PR was stalled
11:19 Krock      #8312
11:19 ShadowBot  https://github.com/minetest/minetest/issues/8312 -- Remove minimal from release installations by rubenwardy
11:22 Krock      merging (2)
11:24 ANAND      Krock: I didn't use any spell-checkers for #9917 - I was just sifting through the player inv action callbacks' docs and noticed "Determinates" :)
11:24 ShadowBot  https://github.com/minetest/minetest/issues/9917 -- lua_api.txt: Fix minor typo by ClobberXD
11:24 Krock      ...
11:24 Krock      GitHub, why did you not take my edited squash message?
11:24 Krock      stupid
11:25 ANAND      lol
11:26 Krock      amend & force push
14:43 ANAND      Is anyone currently working on a fix for #6534? If not, I could try.
14:43 ShadowBot  https://github.com/minetest/minetest/issues/6534 -- Full details of inventory swaps not reported to mods
14:56 rubenwardy appguru may be
15:00 Krock      two PRs = one will be more promising
15:03 ANAND      Aight, I'll work on a fix side-by-side. :)
15:03 Krock      PS sfan5 #9914
15:03 ShadowBot  https://github.com/minetest/minetest/issues/9914 -- POC: Introduce SerializeStream helper class by SmallJoker
15:03 Krock      ANAND: I hope the code battle will be legendary
15:04 Krock      feedback and suggestions are welcome to fix this somehow
15:04 Krock      though would be great to have a meeting today to process some PRs and issues
15:07 ANAND      Krock: More like friendly competition
15:12 sfan5      Krock: from a quick look: maybe have a SerializeStream class for either direction? also I assume a SerializeStream(NetworkPacket *is) constructur is planned?
15:13 sfan5      because if you want to replace the networkpacket with an ostringstream that would lose the preallocation and it's also slower overall (maybe?)
15:18 Krock      sfan5:  SerializeStream(NetworkPacket *is)  would be difficult to implement, because it's not a stream
15:18 Krock      the serialization functions readU8/writeU8 expect streams or byte pointers
15:19 sfan5      I'm sure that could be abstracted over
15:19 sfan5      since in the end all you need is some method to call to put bytes in
15:20 Krock      a class that contains read(), write() and length-getter would be enough
15:21 Krock      though currently existing code mostly uses istream/ostream, which all need modification to take the abstract class
15:21 Krock      inheriting from i/ostream sucks, there are so many functions to implement we'll never need
15:30 ANAND      #9921
15:30 ShadowBot  https://github.com/minetest/minetest/issues/9921 -- InvRef passed to player inventory action callbacks point to the wrong inventory
15:31 ANAND      Krock: I've created a bug report to keep track of the player inv actions issue
15:39 rubenwardy I've made ContentDB load roughly 5-10 faster in the client
15:40 ANAND      Clickbait :P
15:40 rubenwardy providing someone else has viewed it in the last 10 minutes
15:40 rubenwardy although, background updating is on so that shouldn't matter
15:41 rubenwardy turns out the nginx caching never worked because Flask was sending Set-Cookie
15:45 Krock      Wall of text. great.
15:46 rubenwardy 3 messages isn't a wall of text :/
15:46 rubenwardy I've done worse
15:46 Krock      rubenwardy: I meant the issue above
15:46 rubenwardy ohh right
16:14 ANAND      Krock :(
16:15 ANAND      Hm, it is a wall of text...
16:15 ANAND      But it's a well-formatted good-looking one at that :)
18:47 appguru    #9923 for the swapping thingy
18:47 ShadowBot  https://github.com/minetest/minetest/issues/9923 -- Invswap fix by appgurueu
20:41 sfan5      merging #9912 in 10 minutes
20:41 ShadowBot  https://github.com/minetest/minetest/issues/9912 -- [NO SQUASH] Node texture appearance for particles+particlespawners by sfan5