Time  Nick       Message
16:05 rubenwardy trivial #8922
16:05 ShadowBot  https://github.com/minetest/minetest/issues/8922 -- Fix error message caused by adding new parameter to background by rubenwardy
16:05 rubenwardy should I remove support for the new parameter in the standard background element, to avoid people using it?
16:05 rubenwardy probably
16:09 sfan5      mixing mtg 5.1 and engine 5.0 is not something we should support
16:09 sfan5      even if it's just removing an error message that causes no harm otherwise
16:11 rubenwardy what about servers running and using 5.1.0 and clients on 5.0.1?
16:11 rubenwardy this is a client-side error
16:20 rubenwardy question: should #8740 be considered a breaking change considering the order was previously undefined?
16:20 ShadowBot  https://github.com/minetest/minetest/issues/8740 -- Make formspec elements real elements for draw order and clipping by DS-Minetest
16:22 rubenwardy I wonder if a way to fix the backwards compatibility of that is to have a sorter which optionally sorts the elements into the previous order
16:22 rubenwardy perhaps using buckets when generating the formspec
16:23 rubenwardy buckets being    LayerID -> [Element]
16:23 rubenwardy if you factored out the bit where elements are added to the GUI root, then that could be easier
16:23 rubenwardy idk
16:24 rubenwardy oh wait, Irrlicht does attachments in the constructor, not as an add method
16:24 rubenwardy I personally prefer   container->add(element);  to   element = new Element(container);
16:52 sfan5      rubenwardy: do you agree with merging the mtg translation PR first and then fixing the issues in a separate PR?
16:52 rubenwardy I don't mind too much either way
16:52 rubenwardy although, the formspec escape issues are troubling
16:55 sfan5      should definitely be fixes but it's not too critical since the injection potential comes from translation files we'd add (and be able to review)
16:55 sfan5      merging game#2466 in 10 minutes
16:55 ShadowBot  https://github.com/minetest/minetest_game/issues/2466 -- Add support for MT 5 game translation (rebasing #2368) by yquemener
17:04 Krock      please wait a little before merging #8922. I'd like to propose a saner version first
17:04 ShadowBot  https://github.com/minetest/minetest/issues/8922 -- Fix error message caused by adding new parameter to background by rubenwardy
17:04 Krock      keyword "version"
17:05 rubenwardy ok?
17:05 sfan5      merging the game PR..
17:09 sfan5      uh done
17:21 pmp-p      funny print(0xffffffffffffff-0xfffffffffffff0)
17:38 Krock      rubenwardy: why does "formspec prepends" send the formspec version string?
17:39 rubenwardy it doesn't
17:40 rubenwardy Minetest, however, always sends the formspec_version first
17:40 rubenwardy it's defined in a macro and prepended to every message
17:41 Krock      SendPlayerFormspecPrepend also sends the version string, which is not needed at all
17:43 Krock      well, it does not do anything but still
17:46 rubenwardy that's probably a mistake
17:56 p_gimeno   pmp-p: expected, Lua double precision numbers only have 52 bits
17:57 p_gimeno   you have 56 bits there
17:59 pmp-p      sure, but the fact it prints 15 instead of NaN is puzzling
17:59 pmp-p      16*
18:02 p_gimeno   no reason to print NaN, the first number is rounded up to 0x100000000000000
18:02 p_gimeno   that's a little off-topic here though
18:04 p_gimeno   I can explain in PM if you want
18:05 pmp-p      no need to, just good to know about precision
18:08 pmp-p      or the lack of it, depends on interpretation of word "precision"
18:15 p_gimeno   "precision" in the context of floating point is how many digits can be stored after the first nonzero number (starting from the left). In hex, doubles store 13 digits, minus the first bit which is always 1 and isn't stored, but that's an implementation detail. The parser that reads 0xfff... notices that the number doesn't fit and rounds it before storing it, getting 0x1000....
18:19 Krock      ^ pro
18:20 pmp-p      i like my integers well rounded :)
18:22 Krock      rubenwardy: I'm currently working on a better way to use formspec_version[]. would it be OK to force real_coordinates for version 2, together with all upcoming elements in 5.1.0-dev?
18:22 sfan5      rubenwardy, Wuzzy: https://github.com/minetest/minetest_game/pull/2471
18:22 Krock      i.e. replace real_coordinates with v2
18:23 Krock      the idea is to assume backwards compatibility on all changed elements
18:24 Krock      "yes" would mean that formspec_version must be specified for each and every formspec, if features newer than v1 are used
18:25 Krock      "no" would mean that real_coodinates[]" will stay, and formspec_version will be only to force a certain version
18:27 rubenwardy Krock: yes
18:27 rubenwardy I think that would be a good approach
18:27 rubenwardy version 2 = real co-ordinates, no errors on unknown parameters, order determines draw order ( #8740 )
18:27 ShadowBot  https://github.com/minetest/minetest/issues/8740 -- Make formspec elements real elements for draw order and clipping by DS-Minetest
18:44 Krock      soon done
18:54 Krock      (still waiting for build)
19:01 Krock      > when your testing world crashes due to a bug in current master
19:33 Krock      #8923
19:33 ShadowBot  https://github.com/minetest/minetest/issues/8923 -- Formspec version by SmallJoker
19:34 Krock      looks simple but dang did it take long to find the best compatible approach
19:43 Krock      sorry sfan5, I cannot find that gettext call in builtin any more. Maybe it was (re)moved
19:43 sfan5      maybe you were thinking of src/settings_translation_file.cpp?
19:43 Krock      it was a single line:  gettext("foo bar")
19:44 Krock      definitely Lua
20:18 Wuzzy      formspec version? this sounds like a long overdue idea *thumbs up*
20:18 sfan5      unfortunately it doesn't make formspecs any less of a mess
20:18 Wuzzy      but it also sounds like you would have to support billions of legacy versions in future. GLHF :D
20:19 Wuzzy      note that I am personally fine with how the syntax of formspec works. i generally think to "prettify the syntax" are not really important
20:20 Wuzzy      instead, its such things as coordinates, but iirc thats been fixed now, right?
20:21 Wuzzy      Also, Krock: Under what conditions will formspec_version be incremented? is there a defined process?
20:22 Wuzzy      like, which changes to formspec API are "safe" (no new version), and which changes require  a version bump?
20:26 Wuzzy      har har har
20:26 Wuzzy      Changes by FORMSPEC_API_VERSION:
20:26 Wuzzy      FORMSPEC VERSION 1:
20:26 Wuzzy      (too much)