Time Nick Message 08:55 jas_ i'm rebooting my computer 08:55 jas_ bye 14:30 ANAND I'm unable to find MTG in CDB via the in-game content browser 14:31 ANAND I'm certain I've previously installed and updated MTG in this manner. 14:31 ANAND Was it removed? 15:10 sfan5 maybe it's just not visible? 15:39 ANAND I selected the "Games" category. MTG was within the first couple of pages, the last time I checked. 15:41 rubenwardy it's been removed as it was causing issues 15:43 ANAND Ahh, the file deletion issue? 15:54 rubenwardy no, the games being installed to /usr/share issue 15:57 sfan5 surely wouldn't be hard to have the client test writability first 16:00 rubenwardy the other issue is that it falsly says there's an update available, because MTG was installed separately 16:01 rubenwardy I shouldn't have added the meta to MTG to make it appear as installed in CDB 16:03 sfan5 that wouldn't be bad if it could track the installed version 16:03 sfan5 but the bigger problem is that people on 5.2.0 engine would get updates for 5.3.0 MTG 16:57 rubenwardy CDB supports minetest versions 17:02 sfan5 you mean protocol versions? 17:03 rubenwardy yeah 17:06 sfan5 that would be a problem for minetest_game 17:09 rubenwardy it's only a problem if the protocol version isn't bumped between minor releases 17:09 rubenwardy https://content.minetest.net/packages/Minetest/minetest_game/ 17:11 sfan5 it's not bumped if there is no need so uh... 17:11 rubenwardy what would the better approach be? 17:11 rubenwardy API versions? 17:12 rubenwardy complex dependency checking :O 17:12 rubenwardy raw minetest version, which can be different from the advertised version? 17:13 sfan5 my suggestion would be to use the engine version and if it is somehow(?) detected that you're running a fork, do approximate version checking based on the protocol version and warn the user 17:14 sfan5 this approach is somewhat fork-friendly and also doesn't mean any extra work regarding API versioning or similar 17:19 rubenwardy ah good point 17:19 rubenwardy the simple way is to try to find the version in the DB 17:19 rubenwardy if it doesn't exist, find it by protocol version only 17:20 rubenwardy if it does exist, check the protocol version 17:21 rubenwardy the checking done by SQL, obviously 17:23 sfan5 I feel like I'm missing context of CDB's inner workings to understand what you said 17:23 sfan5 ¯\_(ツ)_/¯ 17:28 rubenwardy ¯\_(ツ)_/¯ 17:28 rubenwardy basically the same as you did