| Time |
Nick |
Message |
| 00:08 |
|
VargaD_ joined #minetest-dev |
| 00:13 |
|
twoelk|2 left #minetest-dev |
| 00:13 |
|
twoelk joined #minetest-dev |
| 00:32 |
|
benrob0329 joined #minetest-dev |
| 00:35 |
|
kaeza joined #minetest-dev |
| 00:52 |
sofar |
how would I get a handle on m_script (cpp script api) from src/log.cpp ? |
| 00:53 |
sofar |
m_server->getScriptIface() doesn't work, neither does env->getScriptIface() |
| 00:53 |
sofar |
I want to call callbacks from log.cpp |
| 00:54 |
sofar |
curse you maze of classes and no way to figure out what is in scope :( |
| 01:00 |
sofar |
https://gist.github.com/sofar/3c1802e9dc1dba7c8fee5285e4608b13 |
| 01:00 |
sofar |
that's ... like 99% working except for the missing ref to m_server or m_env or whatever it needs |
| 01:09 |
|
Ruslan1 joined #minetest-dev |
| 01:19 |
|
YuGiOhJCJ joined #minetest-dev |
| 01:49 |
|
makayabou joined #minetest-dev |
| 01:52 |
p_gimeno |
entuland: would you give this a try? https://notabug.org/pgimeno/Gists/src/matrix-to-facedir |
| 01:53 |
p_gimeno |
note that you will need to replace [1] with [1][1], [2] with [1][2], [3] with [1][3], [5] with [2][1] and so on |
| 01:53 |
p_gimeno |
(for your matrices) |
| 01:54 |
entuland |
oh wow |
| 01:56 |
p_gimeno |
(table is indexed by balanced ternary, https://en.wikipedia.org/wiki/Balanced_ternary ) |
| 01:56 |
p_gimeno |
since there's no flipping, 2 vectors determine the rotation uniquely |
| 01:56 |
entuland |
that's turkish for me, and I'm not even sure I should be looking at that kind of math (your very file) at almost 3AM here :P |
| 01:57 |
p_gimeno |
same time here, and I'm going to bed right now :) |
| 01:57 |
p_gimeno |
please give it a go when you can |
| 01:57 |
entuland |
sure thing, tomorrow for sure |
| 01:57 |
entuland |
as in, in about 10 hours or so :P |
| 01:58 |
entuland |
thanks a bunch, I think I see what you're doing there and it will be a great performance improvement compared to my string lookup |
| 01:59 |
p_gimeno |
and easier on the memory allocator too |
| 01:59 |
p_gimeno |
really need to go now |
| 01:59 |
entuland |
hf, tx again |
| 02:45 |
|
ssieb joined #minetest-dev |
| 03:00 |
Wuzzy |
hmmmm |
| 03:00 |
Wuzzy |
changelog claims " Change mapgen order to ores > dungeons > decorations (paramat) " |
| 03:00 |
Wuzzy |
but not in v6, it seems :( |
| 03:00 |
Wuzzy |
in v6 decorations seem to come first |
| 03:01 |
Wuzzy |
its dungeons > decorations > ores |
| 03:01 |
Wuzzy |
please tell this paramat when possible. thx :) |
| 04:10 |
|
twoelk left #minetest-dev |
| 05:15 |
|
ssieb joined #minetest-dev |
| 07:35 |
|
ANAND joined #minetest-dev |
| 08:04 |
|
proller joined #minetest-dev |
| 08:10 |
|
proller joined #minetest-dev |
| 08:15 |
|
proller joined #minetest-dev |
| 08:18 |
|
troller joined #minetest-dev |
| 08:31 |
|
troller joined #minetest-dev |
| 08:35 |
|
ssieb joined #minetest-dev |
| 08:35 |
|
Beton joined #minetest-dev |
| 10:27 |
|
fwhcat joined #minetest-dev |
| 11:10 |
|
Gael-de-Sailly joined #minetest-dev |
| 12:17 |
|
entuland joined #minetest-dev |
| 12:43 |
|
Fixer joined #minetest-dev |
| 12:50 |
|
kaeptmblaubaer joined #minetest-dev |
| 12:52 |
|
Krock joined #minetest-dev |
| 13:11 |
|
makayabou joined #minetest-dev |
| 13:18 |
|
kaeptmblaubaer joined #minetest-dev |
| 13:24 |
|
ANAND joined #minetest-dev |
| 13:30 |
|
kaeptmblaubaer joined #minetest-dev |
| 13:36 |
|
kaeptmblaubaer joined #minetest-dev |
| 14:03 |
|
kaeza joined #minetest-dev |
| 14:06 |
|
Wuzzy joined #minetest-dev |
| 14:40 |
Krock |
merging #8176 in 5' |
| 14:40 |
ShadowBot |
https://github.com/minetest/minetest/issues/8176 -- Prevent tooltip_append_itemname to append itemstring when it's not needed by Wuzzy2 |
| 14:45 |
Krock |
merging.. |
| 15:17 |
Krock |
Wuzzy: #8200 |
| 15:17 |
ShadowBot |
https://github.com/minetest/minetest/issues/8200 -- Autojump: Correctly disable in fly mode by SmallJoker |
| 15:18 |
Krock |
I think it works. tested in a singleplayer world shortly |
| 15:18 |
Wuzzy |
that was fast |
| 15:27 |
|
kaeptmblaubaer joined #minetest-dev |
| 16:03 |
Krock |
was an easy fix |
| 16:15 |
|
Ruslan1 joined #minetest-dev |
| 16:29 |
|
kaeptmblaubaer joined #minetest-dev |
| 16:34 |
|
jas_ joined #minetest-dev |
| 17:03 |
Krock |
#8200 updated. LINT succeeded |
| 17:03 |
ShadowBot |
https://github.com/minetest/minetest/issues/8200 -- Autojump: Correctly disable in fly mode by SmallJoker |
| 17:04 |
Krock |
clang needs fixing .. |
| 17:04 |
Krock |
numeric.cpp:204:11: error: call to 'sqrt' promotes float to double [performance-type-promotion-in-math-fn,-warnings-as-errors] |
| 17:06 |
p_gimeno |
isn't that a genuine complaint? |
| 17:07 |
Krock |
well, f64 c2 = std::sqrt((f64)M[10]*M[10] + (f64)M[8]*M[8]); might solve that |
| 17:09 |
p_gimeno |
yes |
| 17:10 |
Krock |
compiles just fine with gcc/linux |
| 17:10 |
Krock |
sfan5: May I push a commit to change that line to what I wrote above? |
| 17:11 |
p_gimeno |
on second thought, sqrtf works just fine there |
| 17:11 |
Krock |
then why is it an f64? |
| 17:12 |
p_gimeno |
I'm trying to wrap my head around that, the goal is to minimize precision loss |
| 17:13 |
Krock |
atan2 above returns a f32 as well |
| 17:13 |
Wuzzy |
Might <https://github.com/minetest/minetest_game/issues/2307> be an MT issue, actually? |
| 17:13 |
Krock |
game#2307 |
| 17:13 |
ShadowBot |
https://github.com/minetest/minetest_game/issues/2307 -- Kelp selectionbox is smaller than the kelp |
| 17:13 |
Wuzzy |
cuz the plantlike_rooted is new |
| 17:14 |
Krock |
it would be quite annoying to have the selection box too large |
| 17:14 |
Krock |
especially when diving |
| 17:14 |
Krock |
and you can only remove the kelp at its roots, so I think that makes sense |
| 17:14 |
Wuzzy |
have u ever tested it? |
| 17:15 |
Wuzzy |
seems like kelp selection box is always 3 nodes high, although the real plant might be much higher |
| 17:15 |
Krock |
it might also be way shorter |
| 17:16 |
Krock |
actually there seems to be a restriction from MTG |
| 17:16 |
Wuzzy |
??? |
| 17:16 |
Wuzzy |
never seen a shorter kelp |
| 17:16 |
Krock |
plantlike_rooted could in theory also be just 1 node high |
| 17:17 |
Krock |
but in this case it seems to be limitedf |
| 17:17 |
Wuzzy |
yea |
| 17:17 |
Wuzzy |
so the plantlike_rooted height can be changed with param2, but the selection box height stays constant? |
| 17:18 |
Wuzzy |
seems so :( |
| 17:18 |
Wuzzy |
Krock: kelp is really just one node. there are not a dozen nodes for each height |
| 17:19 |
p_gimeno |
Krock: your fix sounds best. The goal would be to reduce precision loss in M[10]*M[10] + M[8]*M[8], so performing it as double is fine. But c2 doesn't need to be double, since it's only used once and in atan2f which requires a single. That was a thinko on my side. So: f32 c2 = sqrtf((f64)... + (f64)...); |
| 17:22 |
p_gimeno |
not sure if std::sqrt is usable here |
| 17:23 |
p_gimeno |
is there a std::sqrtf in C |
| 17:23 |
p_gimeno |
C++11? |
| 17:25 |
p_gimeno |
or better, use sqrt to perform it in double precision and let it be converted automatically to single: f32 c2 = std::sqrt((f64)...) |
| 17:26 |
sfan5 |
Krock: lgtm |
| 17:32 |
Krock |
thanks. pushing |
| 17:32 |
Wuzzy |
Krock: maybe a compromise would be a flag to turn on/off dynamic selection box height? |
| 17:35 |
Krock |
Wuzzy: the anticheat code would need to be adjusted for that as well. Currently only selection boxes up to 3m are supported there |
| 17:35 |
Krock |
same as collision boxes |
| 17:35 |
Wuzzy |
ouch! |
| 17:35 |
Wuzzy |
is the height limit for boxes documented? :) |
| 17:36 |
Krock |
yes, there's an issue about it somewhere |
| 17:38 |
nerzhul |
hey team, it seems we are closeest than before to release but not yet ready :( |
| 17:41 |
Krock |
nerzhul: If you have some time it would be great it you could push additional commits to #8164 so that it can be finished earlier |
| 17:41 |
ShadowBot |
https://github.com/minetest/minetest/issues/8164 -- Disable confirmation dialog on localhost by rubenwardy |
| 17:45 |
Wuzzy |
weblate still not fixed, eh? ;-) |
| 17:45 |
Wuzzy |
no rush, but please alert me when its fixed so i can work on translations immediately |
| 17:45 |
nerzhul |
Wuzzy oO |
| 17:46 |
nerzhul |
Krock you want me to fix the PR myself ? |
| 17:48 |
Krock |
nerzhul: just a suggestion to speed things up. Devs have access to all PR branches |
| 17:48 |
Wuzzy |
nerzhul: https://github.com/minetest/minetest/issues/8173 |
| 17:48 |
nerzhul |
i should do this manually again, pfff |
| 17:49 |
Wuzzy |
again? it never worked |
| 17:49 |
nerzhul |
i like transifex integration is better |
| 17:50 |
Wuzzy |
:( |
| 17:50 |
Wuzzy |
is proprietary |
| 17:50 |
nerzhul |
yes, but at least the product is finished... |
| 17:50 |
Wuzzy |
do you believe its a bug in weblate? |
| 17:51 |
nerzhul |
i think, reading the logs |
| 17:51 |
* VanessaE |
mumbles something about a certain RTT commit that needs reverted |
| 17:52 |
Wuzzy |
what do the logs say? |
| 17:52 |
Wuzzy |
nerzhul: i think these warnings have been there for weeks now, even after your previous manual updates |
| 17:53 |
Wuzzy |
it seems your manual pushes (or whatever you did) did nothing |
| 18:02 |
nerzhul |
they works properly |
| 18:02 |
Wuzzy |
how do you test this? |
| 18:03 |
p_gimeno |
for the record: "We offer Transifex for free to Open Source projects that have no funding, revenue, and/or commercialization model." https://www.transifex.com/pricing/ |
| 18:03 |
Wuzzy |
... |
| 18:04 |
Wuzzy |
nerzhul: new strings like "X spread" don't show up in Weblate. Doesn't seem to work properly to me. |
| 18:05 |
Wuzzy |
nerzhul: ohh. i just noticed something |
| 18:05 |
Wuzzy |
in minetest, de.po file is not up-to-date. minetest.pot is larger (has more strings) |
| 18:05 |
Wuzzy |
forgot to run updatepo.sh? :D |
| 18:05 |
nerzhul |
i did it everytime |
| 18:06 |
Wuzzy |
o_O |
| 18:06 |
Wuzzy |
i dont know on which repo or branch you are |
| 18:06 |
Wuzzy |
can you give me a commit id? |
| 18:12 |
Wuzzy |
OMFG!!! |
| 18:12 |
nerzhul |
master |
| 18:12 |
Wuzzy |
fgettext("")!!!!!!! |
| 18:12 |
nerzhul |
and currently i don't have the info i'm on another thing |
| 18:12 |
Wuzzy |
builtin/mainmenu/tab_online.lua |
| 18:13 |
Wuzzy |
One does not simply call gettext with empty string |
| 18:15 |
nerzhul |
very strange |
| 18:16 |
Wuzzy |
nerzhul: empty string has a special meaning in Gettext, look it up. it should not be used normally |
| 18:16 |
Wuzzy |
also, updatepo.sh reports a million errors. |
| 18:16 |
Wuzzy |
when i remove fgettext(""), the errors are gone and updatepo.sh purrs like a cat =) |
| 18:17 |
nerzhul |
send patch ? |
| 18:17 |
Wuzzy |
- .. ";btn_mp_refresh;" .. fgettext("") .. "]" .. |
| 18:17 |
Wuzzy |
+ .. ";btn_mp_refresh;]" .. |
| 18:17 |
Wuzzy |
in builtin/mainmenu/tab_online.lua |
| 18:17 |
Wuzzy |
or do you want me to post a full-blown PR? |
| 18:19 |
Wuzzy |
hmmm i wonder how the fgettext("") even got here. it does not make sense |
| 18:19 |
* Wuzzy |
runs blame |
| 18:19 |
Wuzzy |
nerzhul: also, i wonder how you managed to not notice the million errors and warnings returned by updatepo.sh? Automated script? Silenced warnings? |
| 18:21 |
nerzhul |
i think so, i don't have any return back from the script |
| 18:22 |
Wuzzy |
o_O |
| 18:22 |
Wuzzy |
maybe stderr vs stdout (need to check |
| 18:22 |
Wuzzy |
btw this is the offending commit: https://github.com/minetest/minetest/commit/5ef9056e89855660d93c4e8d98a7dbc349c1a7eb |
| 18:24 |
Wuzzy |
LOL |
| 18:24 |
Wuzzy |
apparently someone already complained... |
| 18:26 |
Wuzzy |
basically if you call gettext("") then you get the metadata of the PO file returned. its at the header of each PO file |
| 18:27 |
rubenwardy |
Please note that I don't currently have access to a compiler due to laptop issues |
| 18:30 |
nerzhul |
interesting, build is broken on my machine |
| 18:31 |
nerzhul |
or i just mispaste in a classname :D |
| 18:32 |
Wuzzy |
nerzhul: ah. updatepo.sh redirects output to stderr if there's an error. maybe your console doesn't print stderr? |
| 18:39 |
nerzhul |
lol, hopefully yes |
| 18:42 |
nerzhul |
ok i have PR fixes + unittests for IPv4 |
| 18:46 |
Wuzzy |
holy crap. yep. updatepo.sh indeed had a serious hiccup because of the empty string. basically, updatepo.sh was confused so it just removed the metadata from the PO files! no wonder Weblate bitched around. |
| 18:47 |
Wuzzy |
I wanna bet this is what will Make Weblate Great Again |
| 18:53 |
nerzhul |
pushing 8164 + my fix additions in two commits to master |
| 18:53 |
nerzhul |
(unittests + calculation fixes) |
| 18:54 |
Wuzzy |
*drummroll* |
| 18:54 |
nerzhul |
#8164 |
| 18:54 |
ShadowBot |
https://github.com/minetest/minetest/issues/8164 -- Disable confirmation dialog on localhost by rubenwardy |
| 18:58 |
Wuzzy |
huh. |
| 18:58 |
Wuzzy |
-0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, 0x7f, 0, 0, 1}; |
| 18:59 |
Wuzzy |
+0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, 0x7f, 0, 0, 0}; |
| 18:59 |
Wuzzy |
why? |
| 19:03 |
sfan5 |
the memcmp with mapped_ipv4_localhost should only compare 13 bytes |
| 19:03 |
sfan5 |
just as the m_address.ipv4.sin_addr.s_addr comparison should only compare one |
| 19:03 |
sfan5 |
IPv4 localhost is 127.xx.xx.xx, not just a single address |
| 19:03 |
p_gimeno |
yep, see https://github.com/minetest/minetest/pull/8164#issuecomment-460180365 |
| 19:04 |
sfan5 |
ah, that was fixed but not in the PR |
| 19:04 |
|
paramat joined #minetest-dev |
| 19:05 |
nerzhul |
sfan5 yeah |
| 19:05 |
nerzhul |
at least we both agree :D and you have the fix in master + unittests |
| 19:07 |
Wuzzy |
ah, its a mask. makes sense |
| 19:07 |
sfan5 |
merging #8206 in 5 mins |
| 19:07 |
ShadowBot |
https://github.com/minetest/minetest/issues/8206 -- Remove fgettext("") in builtin by Wuzzy2 |
| 19:09 |
paramat |
please can we release RC2 containing 3 axis rotations? that really needs testing |
| 19:16 |
|
kaeptmblaubaer joined #minetest-dev |
| 19:18 |
Wuzzy |
3 axis rotations? yaaaay |
| 19:18 |
Wuzzy |
it would test that immediately!! |
| 19:18 |
Wuzzy |
i* |
| 19:18 |
VanessaE |
paramat: and #8187 |
| 19:18 |
ShadowBot |
https://github.com/minetest/minetest/issues/8187 -- Revert "RTT fixes (#7428)" by ClobberXD |
| 19:18 |
VanessaE |
(CI failures be damned) |
| 19:20 |
nerzhul |
mac osx travis shit |
| 19:21 |
VanessaE |
nerzhul: this reminds me, users are starting to yell a bit for Android and iOS builds of rc1 (or rc2 then)... |
| 19:21 |
|
Gael-de-Sailly joined #minetest-dev |
| 19:22 |
VanessaE |
maybe you can put an apk somewhere that I can point people at? |
| 19:22 |
VanessaE |
(e.g. via the /topic in my server channel) |
| 19:25 |
paramat |
yes 8187 should be in rc2 also, thanks for testing that pr |
| 19:26 |
VanessaE |
yw. |
| 19:26 |
VanessaE |
it works fine so far, but I think I'd still give it a few more days before declaring it 'good' |
| 19:27 |
VanessaE |
(you know how testy MT networking can be) |
| 19:27 |
VanessaE |
MT doesn't just have corner cases -- it has "turned 180 degrees and wound itself into a spiral" cases ;) |
| 19:29 |
paramat |
the mapgen order change was to stop ores appearing in dungeon walls if the wall is biome stone. however mgv6 doesn't use the biome api and has hardcoded biome stone, only desert dungeons have walls that are biome stone and ores are not in desert stone. also, mgv6 is meant to be preserved as it is and this is not a bugfix |
| 19:30 |
nerzhul |
VanessaE, i thought we already were in RC1 :p |
| 19:31 |
nerzhul |
i never saw a RC release |
| 19:32 |
VanessaE |
nerzhul: indeed so but now the talk is RC2 |
| 19:32 |
VanessaE |
nerzhul: https://forum.minetest.net/viewtopic.php?f=18&t=22078 |
| 19:33 |
VanessaE |
(I guess there just wasn't a release made via github tags/"releases") |
| 19:36 |
paramat |
Krock i think #8200 can be merged |
| 19:36 |
ShadowBot |
https://github.com/minetest/minetest/issues/8200 -- Autojump: Correctly disable in fly mode by SmallJoker |
| 19:38 |
paramat |
can we merge #8167 if rubenwardy is ok with the code? |
| 19:38 |
ShadowBot |
https://github.com/minetest/minetest/issues/8167 -- Consistent HP and damage types by SmallJoker |
| 19:40 |
paramat |
i've tested it in a small way |
| 19:55 |
paramat |
another blocker issue #8203 rubenwardy |
| 19:55 |
ShadowBot |
https://github.com/minetest/minetest/issues/8203 -- Unzip forces all files to lowercase |
| 19:56 |
paramat |
since you have computer issues i can test anything you suggest |
| 19:59 |
VanessaE |
well, ^ that reveals another bug though |
| 19:59 |
VanessaE |
isn't the engine supposed to be case-insensitive when loading media, code, etc? |
| 20:12 |
rubenwardy |
paramat: yeah, I asked them to make that |
| 20:12 |
rubenwardy |
Its not directly cdb related |
| 20:12 |
rubenwardy |
As in, it's in the existing unzip API |
| 20:14 |
Krock |
paramat: the HUD can only show 10 hearts (20 half-hearts) so you might have much more HP than 20 |
| 20:20 |
Krock |
merging #8200 as "Autojump: Disable in fly mode, support continuous forward" in 10' if there are no objections |
| 20:20 |
ShadowBot |
https://github.com/minetest/minetest/issues/8200 -- Autojump: Correctly disable in fly mode by SmallJoker |
| 20:21 |
paramat |
ok, and ok |
| 20:21 |
|
kaeptmblaubaer joined #minetest-dev |
| 20:23 |
paramat |
rubenwardy can we merge #8167 since you're ok with the code and i've tested? |
| 20:23 |
ShadowBot |
https://github.com/minetest/minetest/issues/8167 -- Consistent HP and damage types by SmallJoker |
| 20:23 |
Krock |
frontend issue needs fixing before release: #8190 |
| 20:23 |
ShadowBot |
https://github.com/minetest/minetest/issues/8190 -- Fix grayed-out modpacks in world configuration dialog by random-geek |
| 20:24 |
Krock |
trivial though |
| 20:26 |
paramat |
yeah, blocker |
| 20:26 |
Krock |
testing alternative fix |
| 20:27 |
paramat |
nicely simple |
| 20:29 |
paramat |
rubenwardy can we take your comment https://github.com/minetest/minetest/pull/8167#issuecomment-460069264 as a +1 now i've tested? |
| 20:32 |
|
Sokomine joined #minetest-dev |
| 20:33 |
Krock |
Will push https://krock-works.uk.to/u/patches/0001-Builtin-Fix-grayed-out-but-enabled-modpacks.patch in 15 minutes unless there are objections |
| 20:34 |
Krock |
Tested. Replaces #8190 to fix #8205 |
| 20:34 |
ShadowBot |
https://github.com/minetest/minetest/issues/8190 -- Fix grayed-out modpacks in world configuration dialog by random-geek |
| 20:34 |
ShadowBot |
https://github.com/minetest/minetest/issues/8205 -- Enabled modpacks arent green when entering config |
| 20:43 |
Krock |
merging autojump... |
| 20:54 |
Krock |
pushing modpack fix.. |
| 20:56 |
Krock |
will address the last issue of the network revert tomorrow |
| 20:56 |
Krock |
I'd also like to merge #8167 on Sunday |
| 20:56 |
ShadowBot |
https://github.com/minetest/minetest/issues/8167 -- Consistent HP and damage types by SmallJoker |
| 21:00 |
|
Ruslan1 joined #minetest-dev |
| 21:12 |
|
diemartin joined #minetest-dev |
| 21:54 |
|
hecks joined #minetest-dev |
| 22:25 |
|
kaeptmblaubaer joined #minetest-dev |
| 23:00 |
hecks |
Okay, why on earth is the minimap more expensive to render than the 3D world? |
| 23:01 |
|
fwhcat joined #minetest-dev |
| 23:02 |
hecks |
I'm guessing, it's some brute force software rendering |
| 23:02 |
hecks |
I don't think it's GPU bound at all |
| 23:04 |
paramat |
not offtopic |
| 23:04 |
paramat |
minimap is well known to be intensive and probably not optimum |
| 23:05 |
hecks |
60 frames to 25 difference on a regular PC, it's pretty bad |
| 23:05 |
paramat |
! |
| 23:05 |
entuland |
probably the need to scan verticals to work out what's the top block or worse, to scan all blocks to compute the "air" in the xray mode? |
| 23:05 |
hecks |
honestly not even a bird's eye camera and plain render would be this expensive |
| 23:06 |
paramat |
surprised it's that bad |
| 23:06 |
Fixer |
i was thinking someone will consider me trolling, but 6870 was powerful GPU back in the day, and i had double core cpu yet minimap enabling had huge impact on gameplay, stuttery, 20-30fps slashed |
| 23:07 |
Fixer |
there must be a way to improve its performance, needs special kind of guy, but it is possible |
| 23:07 |
hecks |
the range you're using does matter; also nothing in minetest is GPU bound, any GL2 capable card can pull it |
| 23:07 |
Fixer |
hecks: so you have 60 to 25 fps when minimap enabled? that's crazy, what CPU/GPU/drivers/OS? |
| 23:08 |
paramat |
#2844 |
| 23:08 |
ShadowBot |
https://github.com/minetest/minetest/issues/2844 -- Minimap code and behaviour problems |
| 23:08 |
hecks |
fermi on windows 7 |
| 23:08 |
|
kaeptmblaubaer joined #minetest-dev |
| 23:09 |
Fixer |
interesting, it has special code to measure smth https://github.com/minetest/minetest/issues/2844#issuecomment-116043264 |
| 23:09 |
hecks |
and a phenom x4; not really something that should struggle with rendering some 2D sprites |
| 23:09 |
Fixer |
i wanna test that |
| 23:09 |
entuland |
for me it's not really enabling it, but moving around - 49FPS in both with or without minimap, but each step drops it under 30, to raise upt to 49 again |
| 23:09 |
hecks |
it feels more stuttery than a straight up frame time increase |
| 23:10 |
hecks |
you're right, it doesn't break anything when you don't move |
| 23:10 |
paramat |
#5294 |
| 23:10 |
hecks |
my guess is; a stutter for each new mapblock in range |
| 23:10 |
ShadowBot |
https://github.com/minetest/minetest/issues/5294 -- Optimize minimap by numberZero |
| 23:10 |
Fixer |
i have not used minimap since upgrade to 4-core cpu... |
| 23:10 |
Fixer |
interesting how it works these days |
| 23:11 |
entuland |
rotating the POV doesn't affect it for me (but right now I'm down in a cave surrounded by walls at 2 blocks of distance |
| 23:11 |
Fixer |
hecks: it lags for you on 5.0? |
| 23:11 |
entuland |
2 blocks of movement are enough to trigger the framedrop |
| 23:11 |
Fixer |
yeah, it lags during movement |
| 23:12 |
Fixer |
maybe I can even measure a stutter |
| 23:12 |
Fixer |
will compile newest first |
| 23:12 |
Fixer |
also, IntelHD on my side |
| 23:13 |
p_gimeno |
just tested on a RPi3b, with air all around the minimap did not cause any performance difference |
| 23:13 |
p_gimeno |
I'll test at ground height and see |
| 23:14 |
entuland |
not even wandering around, p_gimeno? |
| 23:14 |
hecks |
what's the name of the new 5.0 gizmo for enabling map |
| 23:15 |
paramat |
map:mapping_kit |
| 23:15 |
p_gimeno |
correction, ~10FPS performance drop |
| 23:16 |
hecks |
singleplayer 5.0 feels better than 0.4 on someone's server |
| 23:16 |
hecks |
but it's still bad |
| 23:16 |
hecks |
I'm testing the worst case; non-rotating map, zoomed out |
| 23:17 |
p_gimeno |
at ground height, 3 FPS drop |
| 23:19 |
hecks |
measure drawtime, not FPS |
| 23:19 |
hecks |
drawtime: 8 without, 9 with, up to 30 moving |
| 23:21 |
p_gimeno |
drawtime 50ms without, 51ms with (at ground height, 18 FPS) |
| 23:23 |
hecks |
I call bull |
| 23:23 |
hecks |
drawtime when moving around with the map? |
| 23:24 |
p_gimeno |
static |
| 23:25 |
Fixer |
holy crap, had large 1-2 sec page scroll lag on vanilla game within 20 sec of new map start |
| 23:26 |
hecks |
page scroll? explain |
| 23:26 |
Fixer |
please remind me what is minimap hotkey? |
| 23:26 |
hecks |
F9 change minimap mode, shift+F9 toggle fixed/rotating |
| 23:27 |
Fixer |
thanks, seing large fps impact, let me start up my FRAPS |
| 23:27 |
Fixer |
and uncap fps |
| 23:27 |
hecks |
no need for fraps; disable vsync in settings, then F5 and look at the drawtime |
| 23:27 |
Fixer |
i want graphs |
| 23:28 |
Fixer |
fraps has drawtimes |
| 23:28 |
hecks |
hit F5 a few more times and you'll have your graphs |
| 23:28 |
Fixer |
can draw in excel |
| 23:28 |
Fixer |
they are too fast |
| 23:29 |
Fixer |
here is quick glimpse https://i.imgur.com/f5v19Rt.png |
| 23:29 |
hecks |
the thin bars, that's minimap |
| 23:29 |
p_gimeno |
ok, while moving things change drastically... I managed to get a fairly stable 55ms while moving in circles, and the minimap raised it to 128 |
| 23:30 |
hecks |
likewise https://a.uguu.se/dd01Sb7onCfT_screenshot_20190210_002820.png |
| 23:31 |
Fixer |
0.005 to 0.023 stutters during walking with minimap |
| 23:31 |
hecks |
yes, and if you just fly+fast around, it's consistently bad |
| 23:31 |
Fixer |
more like 0.007 to 0.023 but you guess the idea |
| 23:32 |
Fixer |
my vrange was 160 |
| 23:32 |
Fixer |
simple leaves |
| 23:32 |
Fixer |
opaque i mean |
| 23:32 |
Fixer |
no shaders |
| 23:32 |
hecks |
100 view range; it doesn't matter |
| 23:32 |
hecks |
everyone agrees that something's rotten in denmark, so I'll look at the code |
| 23:33 |
Fixer |
yeah, I agree |
| 23:33 |
Fixer |
i can make bigger, fancier graph with fraps |
| 23:34 |
hecks |
who cares |
| 23:35 |
hecks |
found the minimap code, of course it's not in /client/ but dumped in the main dir... okay |
| 23:37 |
Fixer |
fraps crash :D |
| 23:38 |
hecks |
I'll be trying to make sense of the map code now, for 700 lines it's woefully overengineered |
| 23:46 |
|
kaeptmblaubaer joined #minetest-dev |
| 23:51 |
|
Sokomine joined #minetest-dev |
| 23:53 |
Fixer |
longer graph https://i.imgur.com/5DtwVnX.png |
| 23:53 |
Fixer |
slow downs are evident |
| 23:53 |
hecks |
I don't think this needed any further investigation |
| 23:53 |
Fixer |
spikes go as high as 30ms |
| 23:53 |
hecks |
I'm straight up rewriting the minimaps |
| 23:53 |
paramat |
woohoo |
| 23:54 |
hecks |
and generalizing them into a "flatmap" system, which can have other uses |
| 23:55 |
paramat |
the potential to generate a map of the world as you explore? |
| 23:55 |
Fixer |
especially this part https://i.imgur.com/6dxOQvM.png |
| 23:59 |
Fixer |
interesting how instant is stuter, last good frame - 7.5 ms, next - 37 ms, third - 22 ms, fourth - 6.5 ms |
| 23:59 |
hecks |
paramat: perhaps |
