Time  Nick      Message
15:02 wrrrzr    Why we migrate to catch?
15:26 MTDiscord <josiah_wi> I think I outlined my reasoning in the associated issue. But in summary, Catch is well-known, and provides much more more readable information on test failures. I was running into the problem that I could not even figure out which tests were failing, because the failure messages are buried in the test output.
15:27 MTDiscord <josiah_wi> #13610
15:27 ShadowBot https://github.com/luanti-org/luanti/issues/13610 -- Migrate unit tests to Catch2 framework.
15:38 MTDiscord <luatic> see also e.g. https://github.com/luanti-org/luanti/pull/16040#issuecomment-2816704153
15:59 MTDiscord <josiah_wi> That's an intriguing idea to make test.h use Catch2 under the hood, Lars. It almost seems trivially easy except that I don't think TEST_CASE will work inside a function scope (haven't tried it though).
16:00 MTDiscord <josiah_wi> If I'm right, you can't simply change REGISTER_TEST_CASE to be a TEST_CASE that invokes the described method.
16:06 MTDiscord <luatic> yes, that's basically the problem. and REGISTER_TEST_CASE expects a void function pointer (so a lambda won't work). but i think using macros something should be possible.
16:15 MTDiscord <josiah_wi> Yes, it should... you can end the function block and then do the TEST_CASE. Just like SQL injection.
16:16 MTDiscord <josiah_wi> Oh, but then you also need a mechanism to open another function definition, and make sure that gets called.
16:16 MTDiscord <josiah_wi> So you end up with something bordering on Catch2 macro sophistication.
16:16 MTDiscord <josiah_wi> Sounds like it's complicated enough that we absolutely have to do it to show off.
16:16 MTDiscord <luatic> hehe
16:52 pgimeno   would this patch be considered for minetest_game? http://www.formauri.es/personal/pgimeno/pastes/fix_dry_grass_on_dirt.patch
16:56 pgimeno   it is formatted so it can be imported with `git am`
16:58 MTDiscord <luatic> seems fine to me. though there should maybe be a comment left in the code to prevent someone making this mistake again?
17:12 pgimeno   yeah I thought the same after sending it
17:17 sfan5     if you add a comment we can merge it
17:18 pgimeno   done, refresh
17:20 sfan5     done
17:20 pgimeno   thanks
21:42 MTDiscord <herowl> Pretty sure function pointers can bind to a corresponding lambda.
21:43 MTDiscord <luatic> if there are no captures, yes.
21:43 MTDiscord <herowl> Does that context require captures?
21:43 MTDiscord <luatic> yes
22:11 [MatrxMT] <grorp> sfan5: would you like to have another look at #13125 (after the hack I introduced to fix the "stuck notification" issue), or shall we just merge it now?
22:11 ShadowBot https://github.com/luanti-org/luanti/issues/13125 -- Android: Persistent notification while ingame by srifqi
22:57 sfan5     I'll check
22:58 sfan5     the hack is just to re-show it every 10 seconds? sounds ok
22:58 sfan5     maybe consider making the duration longer
23:30 [MatrxMT] <grorp> yeah, and the user shouldn't notice anything of that since it just replaces the existing notification. it's still observable though, e.g. if you try to dismiss the notification
23:33 [MatrxMT] <grorp> what duration would you suggest? I suppose it's a trade-off between "notification stuck shorter after closing" and "less work", but I'm not sure if the work is significant
23:54 [MatrxMT] <Zughy> any core dev's take on #15916? Before that "possible close" goes stale
23:54 ShadowBot https://github.com/luanti-org/luanti/issues/15916 -- Engine crafting system v2 (in lua)