Time  Nick       Message
04:46 paramat    celeron55 i have a PR that somewhat sorts out CSM for MT 5.0.0 by disabling client-provided CSM until server-sent CSM is completed, but it's likely to be controversial. i'm wondering if you might approve it? https://github.com/minetest/minetest/pull/7949
09:25 ClockGen   nerzhul: I've managed to write a build process for fdroid, minetest compiles with NDKs r16b and r17b, but now I'm facing another problem: it crashes on startup. I checked logcat and it actually segfaults with null pointer dereference, I've checked the code and I thought it was fixed by this commit? https://github.com/minetest/minetest/commit/119aa5c91977ce5381e399e21eabf390e5cb5f5e
09:25 ClockGen   It's one of commits between 0.4.17 and 0.4.17.1
10:37 nerzhul    ClockGen why you don't have it in your tree
10:37 nerzhul    paramat, remember c55 is the original dev but it's a consensus, and i really think a 1 year old message can be invalid :)
10:38 nerzhul    just stop that please, VanessaE which was one of the -1 1 year ago said it's not the best but it's quite good
10:38 nerzhul    Krock looking
10:38 nerzhul    paramat: remember this battle was one of the reason of my afk
10:59 ClockGen   nerzhul: I do have it, I even compared the code manually
11:00 ClockGen   But for some reason it still happens
13:03 Krock      nerzhul: push_back missing near https://github.com/minetest/minetest/pull/7939/files#diff-2e8891f210fadde431f8a10c08a228edR77
13:03 nerzhul    oops i did this only on server side
13:03 nerzhul    ty
13:04 Krock      np. I see the compilers are yet not happy with it
13:05 Krock      will merge #7947 in 10 minutes
13:05 ShadowBot  https://github.com/minetest/minetest/issues/7947 -- Update minetest.conf.example and run updatepo.sh by SmallJoker
13:06 * Krock    quickly changes the author back to get impressive contributor stats
13:06 nerzhul    yes it's memory problems in the unittests because the lifetime is different than in code due to our historical lifetime on these objects
13:14 Wuzzy      Please someone review my eat sound patch: https://github.com/minetest/minetest/issues/7788#issuecomment-429656014
13:15 ANAND      lol Krock :P
13:15 Krock      merging...
13:16 Krock      no what
13:17 nerzhul    ?
13:17 ANAND      :O
13:17 Krock      fixed the commit. for some reason squashing assigned it to my account
13:19 ANAND      You really did change back the author, didn't ya? ;)
13:21 Krock      yes sure. the stats would shoot through the roof otherwise
13:22 nerzhul    Krock i'm pretty sure you want to be the first commitor, behind c55 :p
13:22 Krock      yes sure. maybe I should re-structure the repository now
13:23 ANAND      Would someone be able to take a look at #7891?
13:23 ShadowBot  https://github.com/minetest/minetest/issues/7891 -- Fix ContentDB packages timing out by using download_file instead by rubenwardy
13:24 nerzhul    Krock: are you for a FF this day ?
13:24 nerzhul    it seem your didn't answered on https://github.com/orgs/minetest/teams/engine/discussions/17 :)
13:24 Krock      nerzhul: what does FF mean in this context?
13:24 nerzhul    not firefox :p
13:25 Krock      https://github.com/orgs/minetest/teams/engine/discussions/17/comments/4
13:25 nerzhul    erf; ... i need to clear my glasses i missed the earlier comments :p
13:25 Krock      well, it's nearly no answer but I'd like to see a feature freeze
13:25 nerzhul    yeah, i was afraid about the changelog since august
13:26 nerzhul    it seems you, sfan5, ruben (except a little content db fix which requires re-handling one feature), and me are okay
13:26 nerzhul    paramat seems okay or neutral i don't really understand :p
13:27 nerzhul    nore, sofar lhofhansl are absent
13:27 nerzhul    Shadowninja too
13:27 nerzhul    Ezhh doesn't vote
13:30 nerzhul    then we have +1 (4), unk (1), afk (5), nc (1)
13:43 Wuzzy      Krock: ???
13:43 Wuzzy      this link does not work: https://github.com/orgs/minetest/teams/engine/discussions/17?from_comment=4#discussion-17-comment-4
13:44 Krock      Wuzzy: these page are restricted to core devs
13:44 Wuzzy      wow, secret link
13:44 Krock      or rather those people who are inside the "Engine" group of the Minetest team
13:44 Wuzzy      GitHub says error 404 :O
13:45 Wuzzy      GitHub lied to me!! haha
13:45 Wuzzy      should be error 403
13:45 Wuzzy      the error has an error :D
14:06 Wuzzy      hello. i have recently worked on making builtin translatable
14:06 Wuzzy      https://github.com/minetest/minetest/pull/7955
14:06 Wuzzy      its WIP and a couple things need to resolved
14:06 Wuzzy      opinions are welcome
14:07 Wuzzy      feel free to discuss it right here in IRC
14:15 Krock      good idea
14:20 Wuzzy      "idea"?
14:20 Wuzzy      translating stuff is kind of obvious thing to do :D
14:21 Krock      yes, but yet it wasn't done in builtin
14:22 Krock      having the translation files too would be great...
14:22 Krock      doesn't intllib have a tool to collect the strings?
14:23 Wuzzy      intllib has nothing to do with that
14:23 Wuzzy      and no, we dont have any scripts to collect the strings. sad but true
14:23 Wuzzy      at least not in our repo
14:24 Andrey01   hi devs, i want to ask what decision did you make for #7930 ?
14:24 ShadowBot  https://github.com/minetest/minetest/issues/7930 -- Support for Mesh Animations in Mesh Nodes
14:24 Wuzzy      Krock: well, there is a script *somewhere*, hold on a minute
14:24 Wuzzy      just not merged into master
14:25 Krock      ah, the one from nore/Ekdohibs
14:25 Andrey01   are you going to add this support in 5.0.0?
14:25 nore       ehm yes I recall that one
14:25 nore       it was never merged, was it?
14:25 Wuzzy      no
14:25 Krock      yes, sadly.
14:26 Wuzzy      what i totally cannot understand
14:26 Wuzzy      its critically important for translators
14:26 rubenwardy Andrey01: if there's no open pr, then it won't be added in 5
14:26 nore       still toying with the idea of writing a mo parser and using mo instead of our own format
14:26 Krock      #6325
14:26 Wuzzy      YES YES YES
14:26 ShadowBot  https://github.com/minetest/minetest/issues/6325 -- Add scripts to find and update translations. by Ekdohibs
14:26 nore       not trivial though...
14:26 Wuzzy      i hate the .tr format
14:26 Wuzzy      Gettext solves many problems
14:26 nore       Wuzzy: to be honest, I hate it too
14:26 Andrey01   rubenwardy: this pr is opened now
14:27 Wuzzy      why was it chosen? laziness? too hard? :)
14:27 Krock      gettext also creates many problems which is why we have this workaround
14:27 Wuzzy      wait what??
14:27 nore       gettext has problems
14:27 Wuzzy      thats news to me
14:27 Krock      Wuzzy: did you seriously miss the whole discussion about this?
14:27 nore       like, not having an API for what we want to do
14:27 nore       so we would need to parse the gettext files ourselves...
14:27 Krock      it does not allow dynamically loading/unloading translations and requires an app-hard-reset
14:27 Wuzzy      oh riiight. your use case...
14:28 nore       and tr are easier to parse than mo :/
14:28 Krock      https://github.com/minetest-mods/intllib/blob/master/gettext.lua
14:28 nore       Wuzzy: well, if there just was an API to parse mo files, I wouldn't have hesitated one second about what to do
14:29 Krock      *cough* link
14:29 Wuzzy      :D
14:29 nore       I asked the question long before I implemented it and no-one ever found one...
14:29 Andrey01   rubenwardy: why did you not close it if it won`t be added in 5.0.0?
14:29 Krock      there's no C++ implementation but that one from inttlib in Lua
14:30 nore       yeah
14:31 nore       and that one isn't suitable as-is for client-side translations
14:31 nore       https://github.com/minetest-mods/intllib/blob/master/gettext.lua#L110 this would be client executing server-sent code
14:31 Wuzzy      uh-oh
14:31 Andrey01   so will #7930 be added in 5.0.0? What decision ??
14:31 ShadowBot  https://github.com/minetest/minetest/issues/7930 -- Support for Mesh Animations in Mesh Nodes
14:32 nore       shouldn't matter with a proper C++ implementation: but it means writing a parser for these expressions
14:32 nore       and it doesn't matter for intllib as well: it is all server-side executed code, so no security holes
14:33 Wuzzy      nore: Krock: Aaahh!! I found the translation updating scripts!! https://github.com/minetest/minetest/pull/6325
14:33 Wuzzy      *someone* has killed the PR for unknown reasons :O
14:33 nore       should *probably* be okay even with untrusted translations server-side, I think the worst you can do is an infinite loop
14:33 Krock      Wuzzy: scroll up. It's the line below your "<Wuzzy> YES YES YES"
14:35 nore       anyway, it's complicated in any case...
14:35 Krock      nore: it doesn't look that insecure. better than nothing for the moment
14:35 nore       Krock: yes of course :)
14:35 Wuzzy      Krock: so the PR was killed for incomplete documentation? All that needs to be done is to write the docs, then it's good?
14:36 Krock      Wuzzy: please read the GitHub discussion again. It was closed due to long inactivity by the developer
14:36 nore       oh
14:36 nore       https://github.com/cbeck88/spirit-po
14:36 nore       however: depends on boost...
14:36 Krock      just seen that too.. but that means such parsers exist
14:37 Wuzzy      Krock: so what do you mean with "possibly also a mod where this can be tested?"
14:37 nore       this one seems quite recent but interesting
14:37 Wuzzy      i think you can't just add a mod into the source tree
14:38 Wuzzy      never in the history of minetest has a mod been bundled with Minetest
14:38 Krock      Wuzzy: at this point client translations were very new. Some kind of testing instructions would be very helpful
14:38 Wuzzy      so you mean just some example code in a README?
14:39 Krock      no, sample code in the PR description as it's done with other script PRs
14:39 nore       <Wuzzy> never in the history of minetest has a mod been bundled with Minetest <-- well, there are lots in the minimal/ game
14:39 Krock      ^  that too
14:39 Wuzzy      standalone mod*
14:39 Krock      however, minimal needs some cleanups and the most recent PR is probably dead by now
14:41 nore       oh https://www.gnu.org/software/gettext/manual/html_node/libgettextpo.html
14:42 nore       so maybe we could get away with only gettext...
14:42 Wuzzy      yay
14:42 nore       well, "libgettextpo" isn't exactly gettext but close enough
14:42 nore       and doesn't depend on boost
14:42 Wuzzy      double-yay
14:42 Krock      well yes, it's a C library
14:44 nore       I wonder how much work it would be to use that instead
14:44 nore       probably not that much, given most of the infrastructure for the code is there (assuming we don't deal with pluralforms for now)
14:52 Krock      does that mean you've got some time to work on it? :D
14:57 p_gimeno   Wuzzy: FYI, in Lua, constant strings need no parentheses around the function call if they are the only arg, e.g. S"string" is the same as S("string"). Also, a common choice for aliasing gettext is _ which would make _"string" be the translation of "string".
14:57 nore       ehm, not sure :/
14:58 Wuzzy      but that's not according to coding style guide afaik
14:58 Wuzzy      also S() has been popularzied by intllib, this convention should be kept
14:58 p_gimeno   I think the convenience for translations would merit an exception
14:59 Wuzzy      i will only change the coding style on core dev request. are you are core dev?
14:59 p_gimeno   no, but it's worth discussing, to avoid further clobbering of the code
15:00 Wuzzy      ok fine
15:00 Wuzzy      parenthesis can go away indeed
15:00 Wuzzy      but i am still not happy with S → _ change
15:00 nore       one problem is that this won't work for translations with one argument
15:01 Wuzzy      heh
15:01 nore       (like what minetest.translate currently does)
15:01 Wuzzy      ok then better keep the parenthesis otherwise its inconsistent
15:02 Wuzzy      eeeeh. i just noticed the code quality of the locale update scripts isnt very good
15:02 Wuzzy      i found a bug
15:02 Wuzzy      the script fails when theres an equals sign in the string
15:02 nore       what happens?
15:03 nore       if it replaces the '=' by '@=', that is normal
15:04 Wuzzy      no
15:04 Wuzzy      i mean thats not what it does
15:04 nore       okay
15:04 Wuzzy      the script only collects half of the string, the rest is missing
15:04 nore       so this is a bug indeed :)
15:04 nore       wat
15:04 Wuzzy      everything after the equals sign is gone
15:04 nore       hmmm
15:05 Wuzzy      why do you claim you made the script when it was Ekdohibs?
15:06 nore       because I am Ekdohibs ‽
15:07 Wuzzy      ok
15:07 Wuzzy      nore's secret identity has been finally revealed :D
15:08 nore       lol that was never secret :)
15:08 Wuzzy      thats what you WANT us to believe... *x files music plays*
15:08 nore       I'm even credited as both in https://www.minetest.net/credits/ :p
15:10 rubenwardy _ is used by underscore already
15:10 rubenwardy And stress
15:11 Wuzzy      ?
15:12 Wuzzy      nore: my first attempt of a documentation: https://ptpb.pw/TR8k
15:12 rubenwardy Underscore is a Lua library, stress is a mineteet library
15:13 rubenwardy Thinking about it, _ would be overwritten by Lua values when marked as not important
15:13 nore       Wuzzy: oh, good
15:13 nore       well, if you want to take the PR over and add the documentation, please feel free to do so :)
15:13 Wuzzy      well then just let's keep using S like we have always done
15:13 nore       yeah, S is good
15:14 Wuzzy      nore: i guess this means you are also unwilling to fix the bug
15:14 nore       Wuzzy: I looked at the code and I don't even understand why there is a bug
15:14 Wuzzy      oh
15:14 nore       it should be a simple string.gsub...
15:15 Wuzzy      i am stupid ... hold on a minute
15:15 nore       hm?
15:15 Wuzzy      has nothing to do with equals sign...
15:16 Wuzzy      it fails for something like this (Lua code):
15:16 Wuzzy      S("This is the [minimal] \"Minimal Development Test\" game. Use [minetest_game] for the real thing."     )
15:16 Wuzzy      there's a hiccup at the \"
15:16 Wuzzy      ahhh right. because the script believes its the end of the string, i gues
15:16 Wuzzy      so it will only collect...
15:16 Wuzzy      This is the [minimal] \
15:17 nore       hm
15:17 nore       yeah, lua string parsing is not obvious to do right
15:18 nore       well, that's an additional reason to move to gettext I guess?
15:18 Wuzzy      obviously the " needs to be escaped, thats not unique to lua :D
15:18 Wuzzy      of course! :D
15:18 nore       heh
15:18 Wuzzy      in Hedgewars we also have just a dumbed-down string parser for collecting Lua strings
15:18 p_gimeno   s/\\"/\\x22/g
15:19 Wuzzy      We only recognize strings within "
15:19 Wuzzy      but we support \"
15:19 Wuzzy      so strings in ' or even [[ ... ]] are ignored
15:19 nore       hmmm yeah
15:19 Wuzzy      in docs we wrote 'only use " when you call the locale function'
15:20 Wuzzy      makes things easier for devs :D
15:20 nore       well if you can fix that pattern "([^"]*)", please do :)
15:20 nore       hehe ^^
15:21 Wuzzy      well it gets painful if you want to support [[ .. ]], [=[ .. ]=], [==[ .. ]==] etc
15:21 Wuzzy      lua strings are a bit crazy
15:24 nore       yeah
15:28 p_gimeno   could I take a look? I added support for them (and comments) in a Lua syntax highlighter written in Lua
15:29 Wuzzy      cool
15:29 Wuzzy      here you go: https://github.com/minetest/minetest/pull/6325
15:29 Wuzzy      findtext.lua line 123
15:31 p_gimeno   thanks
15:32 Wuzzy      its good enough for a start if you just fix \"
15:36 p_gimeno   how well does it deal with strings within comments?
15:37 p_gimeno   looks like it doesn't deal with them at all, from a cursory look
15:39 p_gimeno   I'd need a specification. What is it doing about get_translator?
15:40 Wuzzy      specification?
15:40 Wuzzy      ask nore :)
15:41 nore       it looks for calls of the form local S = minetest.get_translator("context")
15:41 nore       then tries to find calls like S("string to translate")
15:41 p_gimeno   ahh, ok
15:55 p_gimeno   well, parsing comments and strings is not too difficult, but parsing locals and function calls is much trickier. It would be best to standardize to one function call without parentheses, with always the same name (single character), and perform argument replacement in a separate function. Example: F(S"string @1 @2", arg1, arg2)
15:56 p_gimeno   F() doesn't need to be standardized, but S() would, according to this proposal.
15:57 p_gimeno   Otherwise, Lua scope rules come into play, e.g. local S = minetest.get_translator("context") do local S = function() end; S("blah") end. That's a royal pain to track.
15:58 Wuzzy      i am all for standardization
15:58 Wuzzy      S() is luckily pretty standard anyway
15:59 p_gimeno   If the tool only has to search for S<lua_string>, that's much simpler.
16:00 Wuzzy      i like to keep parenthesis tho
16:02 Krock      doesn't S() also accept multiple parameters?
16:05 p_gimeno   I'm suggesting separation into two functions, one to handle translation and the other to handle parameters
16:06 Wuzzy      :(
16:06 Wuzzy      bad idea
16:09 nerzhul    p_gimero bad
16:11 p_gimeno   sigh... ok
16:13 Wuzzy      you only need to care about the first parameter, however :)
16:14 p_gimeno   with this structure and these limitations, I think it's best to replace \" as I suggested earlier
16:23 Wuzzy      yeah
16:24 Wuzzy      p_gimeno: wait, you mean s/\\"/\\x22/g ?
16:24 Wuzzy      nooooo
16:24 Wuzzy      that would look horrible
16:24 Wuzzy      its reasonable to expect that \" works
16:24 p_gimeno   definitely yes, but that would be done internally within findtext.lua
16:25 p_gimeno   I'm working on that already
16:25 Wuzzy      great
16:25 Wuzzy      sorry i misunderstood. phew
16:28 p_gimeno   Lua escapes need to be handled too, once the strings are parsed
16:55 p_gimeno   my challenge is that this is handled correctly: S('bar baz "this" \'that\' foobar')  print("this is in a string S(") x=0 print(") still text")
17:05 Wuzzy      try to care about "" first and ignore '' strings
17:07 p_gimeno   I'm almost there anyway
17:32 p_gimeno   phase 1 complete: https://notabug.org/pgimeno/minetest/commit/d69937544317bf7eef4faae1f1022558441048e0
17:35 Wuzzy      nice
17:40 p_gimeno   I think the function I'm writing should also add @ quotes. How are @ themselves quoted? By doubling them?
17:40 Wuzzy      yes
17:41 Wuzzy      p_gimeno: i think this is already handled in nore's code
17:42 p_gimeno   where?
17:42 p_gimeno   anyway, I'm going through the string to remove \ escapes, so I could as well handle it there as I go, and remove another function
17:47 p_gimeno   oh crap, PUC Lua 5.1 doesn't accept \x
17:47 Wuzzy      haha. oh. nore actually didnt handle @@ :D
17:47 nore       no, it shouldn't be handled
17:47 Wuzzy      oops
17:48 nore       it needs to stay in the translation file
17:48 Wuzzy      ah, right
17:48 nore       well and in the string to translate
17:48 nore       (so that it isn't mistaken for things like "@1")
17:48 p_gimeno   nore: um what? so if I write S("This is blah@=3"), that goes as is?
17:49 Wuzzy      yes
17:49 nore       p_gimeno: yes
17:49 Wuzzy      i now understood the logic
17:49 nore       but you are the one to blame here
17:49 nore       because @ should be escaped in your string
17:49 Wuzzy      i had a brainfart
17:49 nore       :)
17:50 nore       if you want "This is blah=3" you should write it as such (the tool will add the @), if you want "This is blah@=3" you should write "This is blah@@=3"
17:50 p_gimeno   ahh
17:51 nore       And the tool will write "This is blah@@@=3 = ...."
17:51 p_gimeno   understood, thanks
17:51 nore       you're welcome :)
18:53 p_gimeno   finished: https://notabug.org/pgimeno/minetest/src/translation-toolchain/util/findtext.lua
18:54 p_gimeno   eep, bug found
19:00 p_gimeno   and fixed
19:19 p_gimeno   another bug found
19:41 p_gimeno   done