Time  Nick       Message
04:38 Pexin      okay so.. it seems that since 5.3, modders have made games which assume bounce is not controllable, so rely on the current behavior (which is still not completely consistent). so my change would make some of their puzzles easier than intended.
04:38 Pexin      the only solution for THAT is a game-specific setting to allow controllable_bounce. is there a way to do that other than adding ANOTHER PERM?  >_<
05:49 Pexin      not a perm, a lua binding to a new serverside variable which must be communicated to the client creating another client/server versioning hiccup whicfasdfalksdjflkasjhfdlka
06:51 MTDiscord  <Benrob0329> Player physics override?
09:42 MTDiscord  <luatic> ^
09:43 MTDiscord  <luatic> make an "old_bounce" boolean flag
10:12 sfan5      do we want to formally enter feature freeze?
10:12 sfan5      this'd mean a 5.5 release just before february
10:13 MTDiscord  <luatic> #11877 should really be fixed IMO
10:13 ShadowBot  https://github.com/minetest/minetest/issues/11877 -- New player causes: duplicate key value violates unique constraint "auth_name_key"
10:15 MTDiscord  <GreenXenith> #11952 has 2 approvals, easy to squeeze it in real quick ;p
10:15 ShadowBot  https://github.com/minetest/minetest/issues/11952 -- Give the ASCII console splash a facelift by GreenXenith
10:16 MTDiscord  <luatic> that's neat
10:35 x2048      sfan5: I'd vote for a feature freeze to get over 5.5.0 as soon as possible
10:38 sfan5      that's my reason too
10:42 sfan5      #11437 will probably end up not being solved
10:42 ShadowBot  https://github.com/minetest/minetest/issues/11437 -- `3d_mode` is broken
10:43 sfan5      and for #10632 I'm waiting for an implementation by appgurueu
10:43 ShadowBot  https://github.com/minetest/minetest/issues/10632 -- Retrieve DPI from get_player_information
10:45 x2048      sfan5: I'll close #11951 then, there is little support among coredevs.
10:45 ShadowBot  https://github.com/minetest/minetest/issues/11951 -- Proposal: Limit availability of dynamic shadows in 5.5.0 by x2048
10:46 sfan5      sure
12:49 erlehmann  celeron55 since it has been decided to scrap the “turn the moon upside down and make a lot of people unhappy” PR, could this be closed? https://github.com/minetest/minetest/issues/11901
12:51 Krock      yes, a 5.5.0 feature freeze seems reasonable
13:06 MTDiscord  <luatic> erlehmann: this should be closed by a doc PR
13:08 erlehmann  luatic then the title of the issue must be changed
13:09 MTDiscord  <luatic> indeed, and it should be labeled "documentation request /missing / error"
13:11 MTDiscord  <luatic> sfan5: don't worry about #11957, it was a mistake on my part, should've commented
13:11 ShadowBot  https://github.com/minetest/minetest/issues/11957 -- Minetest does not detect Postgres version, always considers it zero
13:28 sfan5      obligatory reminder for #11857 and #11866
13:28 ShadowBot  https://github.com/minetest/minetest/issues/11857 -- Raise minimum compiler versions by sfan5
13:28 ShadowBot  https://github.com/minetest/minetest/issues/11866 -- Raise max mapgen limit constant to align with MapBlock by sfan5
13:37 erlehmann  btw, i would like the mapblock thing to be solved because mineclonia (and the other mcl games) contains a structure that can spawn near the map border and contain tnt
13:37 erlehmann  so it is unlikely, but possible, that the map generator will place a crashy schematic there
13:37 erlehmann  (i think?)
13:37 erlehmann  i have not tested that lately tbh
13:39 erlehmann  i will certainly revisit other map border bugs after #11866 is merged and report back about if they are gone
13:39 ShadowBot  https://github.com/minetest/minetest/issues/11866 -- Raise max mapgen limit constant to align with MapBlock by sfan5
13:44 sfan5      you will still find some, I promise
13:49 erlehmann  yeah but i bet this removes a bunch
14:41 sfan5      for some reason an `mod_storage` folder recently appeared in all of my worlds
14:41 sfan5      this is a recent change
14:41 sfan5      an empty*
14:44 Krock      could this be due to the sqlite3 transition?
14:47 erlehmann  Krock wdym which PR are you referring to?
14:47 sfan5      #11763 obv
14:47 ShadowBot  https://github.com/minetest/minetest/issues/11763 -- Use a database for mod storage by TurkeyMcMac
14:49 Pexin      ive seen people complaining about that but they didnt go into detail
14:49 erlehmann  well this should not affect any *existing* worlds unless they are migrated, should it?
14:52 Pexin      i will try to track down who it was and ask them to elaborate (preferably here?)
15:06 sfan5      I guess I should tag a new version of irrlichtmt too so all changes are included in an eventual release candidate
15:09 erlehmann  good idea
16:34 sfan5      merging #11857, #11874, #11952, #11950 in 10m
16:34 ShadowBot  https://github.com/minetest/minetest/issues/11857 -- Raise minimum compiler versions by sfan5
16:34 ShadowBot  https://github.com/minetest/minetest/issues/11874 -- Don't call on_dieplayer callbacks two times by savilli
16:34 ShadowBot  https://github.com/minetest/minetest/issues/11952 -- Give the ASCII console splash a facelift by GreenXenith
16:34 ShadowBot  https://github.com/minetest/minetest/issues/11950 -- [NOSQUASH] Two commits by sfan5
16:48 sfan5      I doubt anyone of the devs who haven't commented yet would disagree so I'll declare the feature freeze in effect
16:49 sfan5      since it's two weeks that'd make the planned release date 2021-01-29
16:51 sfan5      (features already in the milestone are exempted of course)
17:02 erlehmann  i suggest to not have a planned release date, but release after all bugs in RC are solved or deemed unimportant
17:03 erlehmann  it would also allow a faster release or not?
17:03 erlehmann  if no new bugs crop up
17:03 erlehmann  come up
17:04 erlehmann  celeron55 could you maybe look into 11866? i assume you may know a bit about the code ;)
17:06 erlehmann  (i want it to avoid crashy crashy at the map border, not out of purity reasons)
17:09 sfan5      bugs do inevitably crop up
17:10 sfan5      and you want to give people time to actually give 5.5 a test
17:10 erlehmann  well the question is if they are dealbreakers
17:10 erlehmann  ah so you freeze it and then you have 2 weeks for testing?
17:11 erlehmann  i must admit, i do not understand it. but i am in favor of not adding new features shortly before release.
17:11 sfan5      not by us, we give it to people and say "this will be 5.6, make sure your stuff works"
17:12 sfan5      5.5*
17:12 sfan5      devs and some users may be running 5.5-dev but especially server owners do not have time to follow development
17:13 sfan5      so it's their chance to find issues of any kind
17:13 sfan5      before release
17:13 erlehmann  oh yeah
17:14 erlehmann  so 2021-01-29 is the release date for the release candidate?!
17:14 erlehmann  that would make sense
17:15 erlehmann  like, polish it, make a RC, then wait for input
17:15 erlehmann  i think i get it now
17:15 erlehmann  thx
17:52 sparky4    guys guys PLEASE DON"T use sqllite to store cliend side mod data! i loose hundreds and undres of waypoints!
17:53 hlqkj      how does that relates to the use of sqlite?!
17:53 sfan5      erlehmann: no, that's the final release date
17:54 Krock      sparky4: migrate
17:54 sparky4    HOW!?
17:54 sfan5      you should not need to migrate
17:54 Krock      done automatically
17:54 sparky4    there will be a butt load of complaits from people whjo use the mods
17:54 sparky4    it did not for me
17:56 Krock      for server mods, there's --migrate-mod-storage  to migrate them explicitly
17:56 sparky4    there is not client side mod migrator
17:56 sfan5      if the migration does not work you should file that as a bug
17:56 rubenwardy I'd like #11955 in as it improves UX a lot and is relatively simple
17:56 ShadowBot  https://github.com/minetest/minetest/issues/11955 -- Update and improve warnings when no games are found or when only devtest is installed by rollerozxa
17:56 sparky4    it is best not to do this
17:56 rubenwardy Other than that, feature freeze sounds fine
17:56 MTDiscord  <luatic> sparky4: elaborate
17:57 MTDiscord  <luatic> for now, all we've heard are baseless claims
17:57 sparky4    i tested out newer 5.50 minetest and it failed to migrate my lient side mod database and i cannot teleport anywhere.
17:57 Krock      https://github.com/minetest/minetest/pull/11763/files#diff-d0e4d904960723de71f0db13669ff088d764e125bbc8133c3cfed05bf0792512R131-R135
17:57 erlehmann  Krock so if i do not use --migrate-mod-storage what happens then, it continue to use json i hope?
17:57 Krock      client mods assume sqlite3. no migration code,
17:57 rubenwardy #11784 has reached approval but probably too late as that's more risky
17:58 ShadowBot  https://github.com/minetest/minetest/issues/11784 -- Use virtual paths to specify exact mod to enable by rubenwardy
17:58 sparky4    this WILL break many things
17:58 MTDiscord  <luatic> have you migrated, sparky4?
17:58 sparky4    no i dont know how it failed to work
17:58 sfan5      I don't see migration code
17:58 Krock      luatic: there's no CSM migration code
17:58 sfan5      in what Krock linked
17:58 Krock      only for server mods
17:58 sparky4    thje client side mods DO NOT Migrate over
17:58 erlehmann  uh that sounds horrible
17:58 Krock      please open an issue for that
17:59 erlehmann  client side waypoints are like, the most common thing i know from CSMs
17:59 MTDiscord  <luatic> heh, technically we don't support CSM ;)
17:59 erlehmann  luatic do not give them ideas
17:59 rubenwardy Client side API is experimental, breaking compat isn't strictly a problem
17:59 MTDiscord  <luatic> but yeah this should definitely be fixed immediately
17:59 sfan5      throwing away user data != API compat
17:59 erlehmann  rubenwardy declaring it not a problem does not mean it is not a problem …
17:59 rubenwardy I'd at the most migrate but not support files
17:59 rubenwardy API includes file formats
17:59 sparky4    i have a database of hundreds of waypoints to teleport across LF
18:00 MTDiscord  <luatic> sparky4: are you sure that the waypoints are lost?
18:00 Krock      on-the-fly migration should be pretty trivial
18:00 MTDiscord  <luatic> as in, have you checked your FS?
18:00 sparky4    yes but the old database is intact
18:00 MTDiscord  <luatic> yeah, so nothing is really lost?
18:00 erlehmann  just to be clear, server-side nothing will get lost if someone just upgrades from 5.4 to 5.5-dev with the sqlite thingy?
18:00 MTDiscord  <luatic> erlehmann: yes
18:00 sfan5      erlehmann: server-side migrations are not automatic
18:00 erlehmann  sfan5 uh-oh
18:01 sfan5      I know what you're thinking but no
18:01 erlehmann  sfan5 if they are not, this will break A LOT of stuff.
18:01 sfan5      stop talking
18:01 sfan5      the old backend remains supported
18:01 sparky4    this sqllite stuff will break many servers!
18:01 erlehmann  oh good
18:01 erlehmann  sfan5 i shall be silent
18:01 sparky4    and break many client side mod databases
18:01 sfan5      at least this is the behaviour I observed
18:01 sfan5      I can't say for sure because I didn't review the code myself
18:02 sparky4    guys please dont break minetest client side mods!
18:03 Krock      old data is kept intact, and copied to the new backend when specified by the command line argument. there is no data loss
18:03 Krock      sparky4: stop whining and open an issue so that it's tracked
18:03 hlqkj      #11959 and blocking, too
18:03 ShadowBot  https://github.com/minetest/minetest/issues/11959 -- CSM mod storage has no migration path
18:04 Krock      okay
18:04 hlqkj      and my bad for earlier: didn't get he was talking about migration, but sqlite loosing data 0\
18:04 erlehmann  well as long as the old backend is supported, can't that be used for csm mod storage too?
18:05 Krock      yes of course
18:05 erlehmann  good
18:05 sparky4    also nice to meet you all i am mary4 from LF
18:06 sfan5      hmm so do new worlds default to sqlite3?
18:08 sfan5      ah, yes
18:08 Krock      https://github.com/minetest/minetest/pull/11763/files#diff-eed156eda45af76b19e24f3f291ac7cd0d1986c991a4b56323f842544ceab99aR361
18:08 Krock      fallback is "files". https://github.com/minetest/minetest/pull/11763/files#diff-d0ecb51621381529aa2b31bb04fecee4a1b1b7dc0be53978c4878f2c4d4bfe1fR4008-R4009
18:09 erlehmann  sparky4 you can be assured that if the thing would break on servers a lot of ppl would be very unhappy. like, skinsdb would break. mineclone portals could lead to different places, maps would break too i guess.
18:10 rubenwardy Yeah, that's not happening
18:10 erlehmann  i just wanted to reassure sparky4
18:10 sfan5      rubenwardy: [11955] sure if it's fully fixed
18:10 sparky4    thank you and watch the client sid emods too >.> thanks guys!
18:10 sparky4    i gotta get food
18:12 erlehmann  sparky4 in the worst case you can always use waspsaliva client i guess
18:12 erlehmann  cora uses waypoints herself
18:12 erlehmann  (she develops it)
18:12 erlehmann  it is mostly a bunch of CSMs so any CSM breakage is considered important ig
18:12 sfan5      more like just keep using Minetest 5.4 if nothing else helps
18:13 sfan5      no need for any third party software that would probably get you banned off most servers
18:21 erlehmann  sfan5 in my experience people get banned for abusing alternate clients, not for simply joining with them.
18:21 erlehmann  no one cares what client you use if you do not cheat
18:21 erlehmann  but i understand, my advice was wrong
18:40 erlehmann  sfan5, why was this closed? is it fixed? if yes, in which commit? https://github.com/minetest/irrlicht/issues/70
18:40 erlehmann  sfan5, sorry my fault. reading comprehension error.
18:41 erlehmann  it was fixed in https://github.com/minetest/irrlicht/commit/195759100fc723f466b46938128eeaed3edc3a24
18:47 sfan5      ye
18:50 rubenwardy You can be banned for joining CTF with a cheat client, even if you don't use jt
18:50 Krock      make sure you're not caught. that's the general rule
18:53 MTDiscord  <Jonathon> ironically it requires a engine patch to catch people using unpatched df
18:57 erlehmann  rubenwardy wait how would you catch ppl who have a cheat client but set it to “not cheating”?
18:57 rubenwardy That's a secret :)
18:57 erlehmann  like say i override the privilege thing
18:57 erlehmann  i can fly but ofc i do not do it in ctf
18:58 erlehmann  i know a few methods to detect cheats employed, but not one to detect a cheat client
18:58 erlehmann  like, killaura? try to place an invisible entity near the player. if they punch it a lot, cheater.
18:59 erlehmann  same with invisible fake players
19:00 erlehmann  also flying: if someone is accelerating straiht upwards, well … ;)
19:14 v-rob      AAAAAA Why is it that every time I think I have Unicode pretty well figured out and write code for it, I learn yet ANOTHER thing that breaks it all and I have to rewrite the code?! Errg!
19:15 v-rob      And I'm only working with a SUBSET!
19:16 erlehmann  v-rob what is it this time
19:16 v-rob      Grapheme iteration
19:16 erlehmann  v-rob also tell me if you want me to try to break your unicode stuff (but try zalgo text first)
19:17 v-rob      I haven't even gotten it to the rendering part yet, I've always found problems before I get there.
19:17 erlehmann  let me stack hundreds of diacritical marks on a letter in non-canonical form and hope your implementation uses a fast sorting algorithm for figuring it out ^^
19:17 v-rob      Time to go library hunting again...
19:17 erlehmann  tbh i would be happy if utf-8 with unifont worked, how far is that away?
19:18 v-rob      Depends on which languages you want to work.
19:18 Pexin      be vewwy quiet. it's wibrawy hunting season! huhuhuhuhuhu
19:18 MTDiscord  <Benrob0329> inb4 v-rob curses Unicode and swears allegiance to Extended  ASCII
19:18 v-rob      :)
19:18 erlehmann  v-rob all of them in the astral plane
19:18 MTDiscord  <Benrob0329> Welp, time to get the spell book-
19:19 v-rob      Well, some time from now then. Assuming I don't find another problem with my approach, not terribly long before I have a prototype. That's a big if, judging by how many times I've goofed already.
19:20 erlehmann  v-rob i just want to ( ͡° ͜ʖ ͡°) and ? in minetest tbh
19:20 Pexin      ಠ_ಠ
19:22 erlehmann  v-rob vs unicode (2020, colorized) … (ノಠ益ಠ)ノ彡┻━┻
19:23 v-rob      I really don't want to use the ICU library. It's about the largest, cruftiest C++-classes-ported-from-Java with disgusting redundant classes that each have one feature that the others don't, and annoyingly scattered documentation. However, it's the most complete Unicode library there is, and the industry standard at that.
19:25 v-rob      It'd probably double the size of the Minetest executable or something :/
19:58 x2048      sfan5: I think I found a fix for #11437, posted in #11961
19:58 ShadowBot  https://github.com/minetest/minetest/issues/11437 -- `3d_mode = anaglyph` is broken
19:58 ShadowBot  https://github.com/minetest/minetest/issues/11961 -- Reset override material in anaglyph by x2048
20:01 erlehmann  neat!
20:54 v-rob      Results of the library hunt: ICU (yuk), GNU FriBidi + GNU libunistring, or Courier Unicode seem like the most viable options.
20:55 v-rob      Courier Unicode is an interesting case: it's maintained as part of a larger project (Courier Mail Server), but is available as a standalone library.
20:57 v-rob      Both Courier and ICU support bidirectionality as well, so FriBidi wouldn't be necessary with them, whereas FriBidi and libunistring are meant to go hand-in-hand.
20:58 v-rob      So, it's one big ugly library, two smaller libraries, or one obscure but viable library.
20:59 v-rob      Honestly, I see no point in using ICU; it's just too much.
21:01 v-rob      I would like at least one other core dev opinion to know which to work with. The HarfBuzz library would still be included in any case.
21:02 v-rob      I'm fairly certain that I'll never need more libraries than whatever is chosen; those encompass basically all of Unicode.
21:04 erlehmann  -
21:04 erlehmann  sorry, a water bottle fell on the keyboard
21:04 erlehmann  v-rob have you looked at comparable projects already and what they used?
21:06 erlehmann  v-rob like, say, supertuxkart
21:06 v-rob      Hard to say. A lot either don't care about Unicode, or use Pango, which has already been rejected (for instance, it pulls in Cairo, which doesn't play well with hardware acceleration like OpenGL), or embed another GUI system.
21:06 v-rob      I'll look though
21:07 erlehmann  supertuxkart is that other game that forked irrlicht AFAIK
21:07 erlehmann  and i think they have unicode?
21:10 v-rob      Hmm, yes they do actually
21:10 v-rob      Does supertuxkart have text input fields anywhere, or does it only render text?
21:11 v-rob      I can't seem to remember any
21:11 v-rob      Been a long time since I played it
21:11 erlehmann  uh username?
21:11 erlehmann  you could just install it ^^
21:12 v-rob      Too lazy
21:12 v-rob      I'm looking through the code anyway
21:14 MTDiscord  <MisterE> Screenshot (Jan 15, 2022 4:14:13 PM)
21:14 MTDiscord  <MisterE> https://cdn.discordapp.com/attachments/747163566800633906/932020034103021608/Screenshot_20220115-161413.png
21:14 Pexin      when poring through code is considered less effort than installing and trying it out. you know you Might Be too deep into dev.   :]
21:15 v-rob      Apparently, it uses HarfBuzz (like I want to), SheenBidi as opposed to FriBidi, and directly copies grapheme iteration from another library.
21:15 MTDiscord  <MisterE> v-rob look @ screenie
21:15 v-rob      Yes, I'm probably a hopeless programmer :)
21:16 v-rob      Yeah, that makes sense. There's probably a chat somewhere anyway
21:16 v-rob      I don't think assimilating code like they do is a good idea, since you don't get the benefits of bugfixes and updates
21:17 MTDiscord  <MisterE> Bugfixes are important, updates are likely to require bugfixes
21:17 MTDiscord  <MisterE> For libraries
21:18 Pexin      (I take it back, STK is a much bigger download than I expected)
21:19 v-rob      Hmm, the library it copies from seems promising, though...
21:22 v-rob      Well, I guess not. No way of implementing text editing
21:40 v-rob      CEGUI doesn't support graphemes at all, although it has bidi and complex text
21:44 erlehmann  how do you do unicode without graphemes
21:44 erlehmann  like … not render it?
22:57 v-rob      By grapheme I meant language-specific characters (as opposed to codepoints, since characters can be multiple codepoints, and not glyphs either, which are what you render). I prefer "grapheme" to "character" because it's unambiguous what it refers to.
22:59 v-rob      Anyhow, it appears that most projects I've looked at (games and GUIs both) either use some combination of libraries like I'm proposing, or just don't support Unicode.
23:00 v-rob      So, choosing libraries like I've proposed is probably the solution.
23:04 v-rob      If no other core dev weighs in, I'll probably default to GNU libunistring + GNU FriBidi
23:05 v-rob      I figure two more mainstream dependencies is better than one more obscure dependency.
23:05 v-rob      But if anyone would prefer a single dependency over two, please speak up
23:11 erlehmann  well you have evaluated it best
23:11 erlehmann  i guess?
23:14 v-rob      Probably. I just want to make sure the other core devs have a say when adding dependencies.
23:18 celeron55  are those libraries self contained?
23:18 celeron55  i suppose they are
23:18 v-rob      As in, do they depend on anything else?
23:18 v-rob      I'm pretty sure they don't, but I'll check
23:19 erlehmann  it would be a shame if they depend on, say, colors.js
23:19 v-rob      :P
23:19 celeron55  :D
23:19 celeron55  fribidi advertises as not having any dependencies
23:19 erlehmann  but are they telling the truth let me check
23:20 erlehmann  you have been bamboozled, fribidi depends on having a libc!
23:20 erlehmann  (but only that)
23:21 erlehmann  unistring also depends on a libc
23:21 erlehmann  that's basically it
23:22 erlehmann  v-rob i notice that unistring is a c library, does it work with c++ without problems?
23:22 celeron55  and it's LGPL 2.1 for both, right?
23:22 v-rob      extern "C" is nice
23:23 celeron55  looks like no
23:23 v-rob      Hmm, libunistring is a dual license LGPL 3 or GPL 2
23:23 v-rob      I didn't think about that
23:23 celeron55  fribidi is LGPL 2.1, unistring is LGPL 3
23:23 celeron55  but LGPL 3 is fine i guess
23:24 v-rob      Courier Unicode is GPL, so definitely no then
23:24 v-rob      ICU is MIT
23:25 celeron55  hmm
23:25 celeron55  https://www.gnu.org/licenses/gpl-faq.html#AllCompatibility
23:26 celeron55  this makes it quite simple
23:26 MTDiscord  <Benrob0329> Oh no, GPL compatibility
23:26 celeron55  so, it's ok
23:26 celeron55  (using (not copying) LGPL3 code in an >=LGPL2.1 project)
23:26 erlehmann  oh, i have a spiel
23:26 celeron55  if you copied it, then the end result is LGPL3
23:27 v-rob      No, copying promotes it to GPL v3
23:27 erlehmann  https://www.gnu.org/licenses/quick-guide-gplv3-compatibility.png
23:27 celeron55  v-rob: it's not what that page says
23:28 celeron55  how it works is >=LGPL2.1 is promoteable to LGPL3, and if you copy LGPL3 code into an LGPL3 project, well, it's obvious
23:28 erlehmann  https://www.gnu.org/licenses/quick-guide-gplv3.html
23:28 celeron55  but this is about using so it doesn't matter
23:28 erlehmann  > Arrows pointing from one license to another indicate that the first license is compatible with the second. This is true even if you follow multiple arrows to get from one license to the other; so, for example, the ISC license is compatible with GPLv3. GPLv2 is compatible with GPLv3 if the program allows you to choose "any later version" of the GP
23:29 v-rob      Are we 2.1 only or 2.1 and later?
23:29 celeron55  >=LGPL2.1, like i said
23:29 celeron55  that's the shorthand for "or later"
23:29 v-rob      Sorry, I thought we were ==, not >=
23:30 v-rob      That's where I got confused
23:30 celeron55  it of course would be ok to use an LGPL3 library even in a proprietary project, it wouldn't change anything
23:32 celeron55  (what i don't know about is whether some of the statement relating to patents or something differ in 2.1 vs. 3. if so, it might bring troubles to some commercial forks but i doubt it)
23:33 celeron55  so the dependencies are ok and the licenses are ok
23:33 celeron55  i trust you to determine that the design and implementation is up to the task
23:34 celeron55  also i agree that it's best to use mainstream projects
23:35 v-rob      Actually, libunistring does have a dep on iconv on Windows. I can't tell if it's optional or not.
23:39 celeron55  doesn't seem bad in itself, other than it of course makes it 3 libraries
23:39 v-rob      Yeah, it looks required
23:39 celeron55  i would hope people with recent experience in making windows builds would give their opinions
23:40 celeron55  but there's probably not a lot that can be done. being so mainstream it shouldn't be difficult to build them
23:40 v-rob      Actually, it would make 4 -- HarfBuzz is a 100% required library for proper Unicode
23:40 v-rob      Windows uses vcpkg, but I don't see libunistring anywhere on that, which might be somewhat annoying
23:41 v-rob      So, I guess it's 3 deps vs ICU.
23:41 celeron55  oh yes that too... the licene is fine, however what are the dependencies?
23:44 v-rob      As far as I can tell, none.
23:45 erlehmann  celeron55 the GPLv3 patent statements are mostly “you can't say others can use the stuff and then turn around and sue them or you lose the license to this software yourself”
23:46 MTDiscord  <luatic> opinions on making libspatialindex a required dependency?
23:46 erlehmann  celeron55 it's only problematic for big companies that fight patent wars
23:46 erlehmann  or threaten patent litigation
23:47 v-rob      Oh hey, there's a Boost.Locale, which is the Boost wrapper around ICU. Like, replace crufty old Java APIs ported to C++ with highly overblown C++ interfaces XD
23:49 v-rob      No, ICU has no dependencies.
23:50 v-rob      So, it's triple dependencies or one big one
23:51 celeron55  what's this? https://vcpkg.info/port/libunistring
23:52 celeron55  i think all of the libraries mentioned by v-rob are on vcpkg
23:52 v-rob      Odd, when I run `vcpkg search libunistring`, I get no results
23:53 v-rob      That website also doesn't load for me.
23:53 celeron55  i don't know how that can be
23:53 celeron55  wait
23:54 v-rob      Then again, the website might be caused by my internet. It refuses to load other things, like the Web Archive, and just lets it time out.
23:54 v-rob      The vcpkg command line lets other packages show up, though
23:54 celeron55  "supports: !windows"
23:54 celeron55  does that mean it's on vcpkg but can only be installed on... non-windows?
23:55 sfan5      vcpkg is in fact a platform-agnostic package manager
23:56 v-rob      libunistring has build instructions for Windows though. Weird
23:57 celeron55  but, it's a self contained library written in C
23:57 celeron55  what a lazy package publisher
23:57 v-rob      Wait, the main Windows build instructions use Cygwin. That might have something to do with it...
23:59 sfan5      I vaguely remember gettext already including libunistring to it building on mingw is assured
23:59 sfan5      so*