Time  Nick       Message
17:36 Krock      sfan5: what's the difference between connect_timeout and timeout in cURL?
17:37 sfan5      --connect-timeout <seconds> Maximum time in seconds that you allow curl's connection to take.
17:37 sfan5      -m, --max-time <seconds> Maximum time in seconds that you allow the whole operation to take.
17:37 sfan5      (quoted from manpage but also applies to libcurl, you can probably guess which is which)
17:38 Krock      alright I see
17:39 Krock      does the client media downloader no longer need a timeout in case connection is lost?
17:40 Krock      nvm
17:40 Krock      it's now hardcoded in HTTPFetchRequest
17:43 Krock      thanks for the explanation
17:44 Krock      now the code only needs to compile properly
22:02 sfan5      merging #10803, #11135, #11136 in 15m
22:02 ShadowBot  https://github.com/minetest/minetest/issues/10803 -- Add `math.round`; fix `vector.round` by v-rob
22:02 ShadowBot  https://github.com/minetest/minetest/issues/11135 -- Sort out cURL timeouts and increase default by sfan5
22:02 ShadowBot  https://github.com/minetest/minetest/issues/11136 -- CGUITTFont optimizations by sfan5
22:25 pgimeno    10803 is yucky, it doesn't round 0.49999999999999994 and -0.49999999999999994 properly, for a good 'round to nearest or even' see https://stackoverflow.com/a/58411671/2428487
22:28 sfan5      this would've been better to know before merge
22:28 pgimeno    if I knew that the PR existed I would have tweeted :)
22:29 sfan5      sure I'm not blaming you
22:30 pgimeno    15m is not enough for me to check the chan and see what's being merged
22:30 sfan5      you don't have any obligation to do that, was just a general comment
22:33 MTDiscord  <appguru> pgimeno: would math.floor(num +  0.49999999999999994) not work?
22:33 pgimeno    it returns 1, which is incorrect
22:33 pgimeno    er, sorry
22:35 pgimeno    I need to give that some thinking; anyway, round to nearest or even is the best rounding method
22:38 Andrey01   Core devs: #10924 ! Please consider it again, all requested changes have already been made by me long ago
22:38 ShadowBot  https://github.com/minetest/minetest/issues/10924 -- Add modslist formspec for /mods command. by Andrey2470T
22:39 pgimeno    Andrey01: there was a recent discussion that I think you have missed
22:40 Andrey01   Hmm
22:40 Andrey01   When was it?
22:40 pgimeno    https://irc.minetest.net/minetest-dev/2021-03-24#i_5802196
22:41 pgimeno    it starts there, but it extends for a while
22:46 pgimeno    @appguru: it seems it would work (it's delicate because the corner cases need to be carefully examined), as long as the CPU uses round to nearest/even, which is the same case where 0.49999999999999994 fails when using round(num+0.5).
22:47 pgimeno    s/CPU/FPU/
22:49 pgimeno    and if you're depending on the FPU's rounding mode for correct results, the algorithm I posted rounds to nearest or even, which is better than rounding to nearest or away from zero.
22:51 pgimeno    "I suggest not following Python's example" <- Python actually does it right
22:54 pgimeno    hm, but #6774 has a very specific requirement: a round function "that matches engine rounding"
22:54 ShadowBot  https://github.com/minetest/minetest/issues/6774 -- Add a new vector.round type function that matches engine rounding
22:55 sfan5      in that case it might be just right
22:56 pgimeno    not sure if that was actually tested or it worked by chance
22:58 Andrey01   What about my PR, if there are objections, then I suggest to add maybe special options for that command which would output it in the plaintext-form and formspec without producing new cmds
22:59 Andrey01   e.g. option '-f' denotes to output formspec, '-p' - chat text correspondingly
22:59 pgimeno    I'd say keep `/mods` without arguments as is, and add e.g. `/mods table` for the format you suggest
23:00 pgimeno    note I'm not a core dev though, so that's just an idea, not a request
23:00 Andrey01   I'd prefer with arguments both
23:01 Andrey01   Because it is unclear in which format it will output
23:13 pgimeno    as long as it outputs text by default when typing `/mods` I'm happy
23:13 pgimeno    and that one always has the option of outputting text
23:14 MTDiscord1 <Jonathon> it would make stuff inconsistent
23:15 sfan5      stuff probably already is inconsistent
23:15 MTDiscord  <Jonathon> true
23:16 MTDiscord  <Jonathon> id still say it should show the formspec, since its arguably better, as long as it sends a text version if the player is not in game
23:16 MTDiscord  <Jonathon> if a game or mod wants they can override that