Guild icon
DDraceNetwork
Development / developer
Development discussion. Logged to https://ddnet.tw/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet
Between 2021-05-02 00:00:00Z and 2021-05-03 00:00:00Z
Avatar
GitHub BOT 2021-05-02 04:30:23Z
Author icon
ddnet-maps
[ddnet-maps:master] 1 new commit
d910e6a M Wonderfully2, A mood - ddnet-maps
Avatar
GitHub BOT 2021-05-02 08:43:00Z
Author icon
Jupeyy
[ddnet/ddnet] Pull request opened: #3802 Make pixel alignment numerical more robust
Notice the game mode "DDRace" jumping/jittering while scrolling, probably bcs it was near the next floating point (smth like 0.9999f) but rounded down: https://user-images.githubusercontent.com/6654924/116807307-c06cc380-ab32-11eb-8076-fd8d8aeb5a09.mp4

Checklist

  • [ ] Tested the change ingame
  • [ ] Provided screenshots if it is a visual change
  • [ ] Tested in combination with possibly related configuration options
  • [ ] Written a unit test if it works standalone, system....
Avatar
GitHub BOT 2021-05-02 09:01:02Z
Author icon
bors[bot]
[ddnet:master] 2 new commits
fc1b877 make pixel alignment numerical more robust - Jupeyy ddf3212 Merge #3802 - bors[bot]
09:01
Author icon
bors[bot]
[ddnet/ddnet] Pull request closed: #3802 Make pixel alignment numerical more robust
Avatar
Ryozuki 2021-05-02 09:26:56Z
https://lore.kernel.org/lkml/CAHk-=whs8QZf3YnifdLv57+FhBi5_WeNTG1B-suOES=RcUSmQg@mail.gmail.com/
09:27
monkalaugh
Avatar
Learath2 2021-05-02 09:35:20Z
Maybe one day we'll come up with a better way to do shared libs
Avatar
Deleted User 2021-05-02 09:45:47Z
static linking + lto = best xd
Avatar
GitHub BOT 2021-05-02 10:06:33Z
Author icon
Zwelf
[ddnet/ddnet-web] Pull request opened: #126 Make csv more standard-conform
This makes it opening in Github and libreoffice easier. And fixed the attribution line in the explain/ pages. If I got some more time, I'll change how the groups are stored. I rerun the script to update the svgs and they stayed the same as expected.
10:12
Author icon
def-
[ddnet/ddnet-web] Pull request closed: #126 Make csv more standard-conform
10:12
Author icon
def-
[ddnet-web:master] 3 new commits
d08fd59 Make csv more standard conform - Zwelf 038285f Fix wrong comma in attribution - Zwelf fcb3cc7 Merge pull request #126 from Zwelf/pr-fix-csv - def-
Avatar
Deleted User 2021-05-02 10:58:01Z
the performance of native wayland is really good, outperforms x11 a bit with ddnet alteast
Avatar
deen 2021-05-02 14:50:51Z
Nice, didn't expect that
Avatar
GitHub BOT 2021-05-02 15:34:44Z
Author icon
Patiga
[ddnet/ddnet] Pull request opened: #3803 [map_optimize] Fix duplication of ITEMTYPE_EX (uuid index) items
This resulted in duplicate items, with the exact same id and type_id (and content, but id + type_id are supposed to be unique) Credits to heinrich5991 for his feeling that the UUID index duplication could originate in this file.

Checklist

  • [x] Tested the change ingame
  • [ ] Provided screenshots if it is a visual change
  • [ ] Tested in combination with possibly related configuration options
  • [ ] Written a unit test if it works standalone, system.c especially
  • [x] Con...
Avatar
Scrumplex 2021-05-02 15:53:07Z
[2021-05-02 17:52:38][client]: starting... [2021-05-02 17:52:38][sdl]: SDL version 2.0.14 (compiled = 2.0.14) [2021-05-02 17:52:38][gfx]: Created OpenGL 3.3 context. [2021-05-02 17:52:38][gfx]: setting resolution to 640x480 and trying again [2021-05-02 17:52:38][sdl]: SDL version 2.0.14 (compiled = 2.0.14) [2021-05-02 17:52:38][gfx]: Created OpenGL 3.3 context. mesa: for the --simplifycfg-sink-common option: may only occur zero or one times! mesa: for the --global-isel-abort option: may only occur zero or one times! mesa: for the --amdgpu-atomic-optimizations option: may only occur zero or one times! mesa: for the --structurizecfg-skip-uniform-regions option: may only occur zero or one times! [2021-05-02 17:52:38][gfx]: out of ideas. failed to init graphics [2021-05-02 17:52:38][client]: couldn't init graphics
15:53
head scratch
15:53
Jumpeyy's pr_gl_cleanup1 branch running on wayland (edited)
15:58
nvm
15:58
env DDNET_DRIVER=GLES
Avatar
Tsin 2021-05-02 16:28:51Z
Dear DDNet team u rly need to re-think the point system on many maps. Good example is the newest map by @Ravie It have 5 stars and 5 points and lets face it. Its not a hard map. it's a lot easier from older map "Planet Mars" by Aoe (4 stars and 4 points). If you won't clean it up, you will end like I did with ddmax.easy, ddmax.next. Where level of difficulty is just messy. Not to mention ddmax.pro where some maps should never be there in the first place. I've never had time to clean it up and ehhhh. Don't do my mistake =]
f3 1
Avatar
Im 'corneum 2021-05-02 18:42:10Z
i suggest to add half stars. it doesnt mess it up too bad but you get double the rating range
Avatar
Avatar
Scrumplex
env DDNET_DRIVER=GLES
Deleted User 2021-05-02 19:10:26Z
And, how well does it work for you? In the PR we already discussed some bugs, what GPU are you on?
Avatar
Avatar
Deleted User
And, how well does it work for you? In the PR we already discussed some bugs, what GPU are you on?
Scrumplex 2021-05-02 19:13:32Z
it works pretty well. I didn't notice any flickering (edited)
19:13
Just the color-banding thingy on the title screen
Avatar
Avatar
Deleted User
And, how well does it work for you? In the PR we already discussed some bugs, what GPU are you on?
Scrumplex 2021-05-02 19:14:49Z
Radeon Vega 56 btw, using AMDGPU (edited)
Avatar
Deleted User 2021-05-02 19:15:28Z
Ah too bad. So we all on amd Vega or newer. Thanks for testing tho
Avatar
nori 2021-05-02 19:46:54Z
@Jupstar ✪#5615 https://github.com/libsdl-org/SDL/commit/354cabd4a7ae73121a527980d7cef9dae41e9d3a is this the issue we had on wayland? it's fixed on master if it is
egl: favor truecolor configurations. · libsdl-org/SDL@354cabd
If app requested <= 16 color depth and there is a 24-bit config available, favor that. This fixes things that quietly expect to get truecolor output but don't request it (...like SDL...
Thumbnail
Avatar
Deleted User 2021-05-02 19:50:21Z
Ah nice
Avatar
lynn 2021-05-02 21:52:05Z
I don't think adding half stars fixes the difficulty rating system, what does fix it is allowing players to vote on difficulty
Avatar
Deleted User 2021-05-02 21:53:11Z
but half stars defs give more space for more precise difficulty
21:54
one question would be, are longer maps always harder, or should there be a category that gives just more points for longer maps
Avatar
lynn 2021-05-02 21:54:58Z
not that they'd use it. this is a conversation about 5* maps being voted 3* and vice versa, not tiny 1 star discrepancies
Avatar
Deleted User 2021-05-02 21:55:10Z
would be nice to see the estimate map length when voting too
Avatar
lynn 2021-05-02 21:55:40Z
I don't like the length argument too much, length does not affect difficulty unless it's faily, something that should not apply to novice
Avatar
Deleted User 2021-05-02 21:55:57Z
ah i dont meant novice only
Avatar
lynn 2021-05-02 21:56:02Z
then sure
Avatar
Deleted User 2021-05-02 21:56:32Z
kog (in the past atleast) did this i think, more points for longer maps, but difficulty only set the map difficulty (edited)
Avatar
lynn 2021-05-02 21:56:46Z
I think length can really only change a map by about a star unless it's a brutal, then it probably shifts around by 2-3
21:57
insane difficulty changes based on length by, well, pretty much the entire range
Avatar
Deleted User 2021-05-02 21:57:14Z
well that would be the argument for the category btw
21:57
then u could split difficulty from length
21:57
and just give more points for longer maps
Avatar
lynn 2021-05-02 21:57:56Z
I don't think it matters to make a distinction. there is already a ton of criteria you need to consider when rating a map. length is just one tiny part of those considerations, it is not notable above the others
21:58
we need guidelines for these criteria and a voting system, period
Avatar
Deleted User 2021-05-02 21:58:49Z
for shorter maps u generally need less tries to finish if faily
Avatar
lynn 2021-05-02 21:58:52Z
so many decisions are made without the popular opinion of the community already, ddnet is a modern game now and needs to respect its community a bit more imo
Avatar
Deleted User 2021-05-02 21:58:55Z
even if only 10% of the map is faily
Avatar
lynn 2021-05-02 21:59:56Z
and we have room to. the game is not proprietary, the game's contributions are already made by the community. this can extend to our processes and future decisions related to the game
Avatar
Deleted User 2021-05-02 22:00:25Z
yeah but community votes will be abused
22:00
i dunno if i'd like that
Avatar
lynn 2021-05-02 22:00:49Z
I have a distaste for things like a "testing team" :P
22:01
you still have people who oversee things, it wouldn't be total control
Avatar
Deleted User 2021-05-02 22:01:27Z
the biggest problem is, that the testing team would require novice, moderate, hard and insane players all in their "league/category" to be more precise
Avatar
lynn 2021-05-02 22:02:04Z
I think you only need one person around to veto "meme votes" but otherwise they do nothing
22:02
normally things just get voted on difficulty
Avatar
Deleted User 2021-05-02 22:02:22Z
you could also just analyze the gameplay
22:02
how long do teams require and how much finishes are made etc.
Avatar
lynn 2021-05-02 22:02:41Z
yes that's true
22:03
another thing we should do is make testing maps more mainstream, esp if we want average players to have some stake in how maps are rated / approved
22:03
this is ofc something I have always talked about but no one agrees with, so whatever lol
22:04
if you're worried about noobs having misinformed opinions about things we can add a points requirement, but then the accounts issue comes up again
22:06
or something like "x number of y maps completed in order to rate a map in y difficulty"
22:07
though I'm not really a fan of this because it would mean difficulty-limited players will cause maps to be under-rated
22:10
mm I just thought of a better system but it's hard to explain
Avatar
heinrich5991 2021-05-02 22:12:10Z
dynamic rating? 😉
Avatar
lynn 2021-05-02 22:12:24Z
you would check clears around a voted difficulty and weight them based on closeness (to vote a map as brutal 1*, you would either need a few mod 5* clears or 1 brutal clear (some sort of quadratic score weighted towards the target difficulty) (edited)
22:12
yes exactly
Avatar
heinrich5991 2021-05-02 22:13:10Z
ah, I meant a different kind of dynamic rating
Avatar
lynn 2021-05-02 22:13:15Z
assign a score like 50 for exact rank and then decrease that score per map cleared away from the target
22:13
and make the req 50
22:13
huh, what's your idea
Avatar
heinrich5991 2021-05-02 22:16:29Z
something similar to 1/number of finishes
Avatar
lynn 2021-05-02 22:17:09Z
oh weird
22:17
I think that would inflate unpopular maps and hurt the rank of popular ones
22:17
but maybe there's some compensation
22:18
my points system idea is the best anyway ;D
22:18
weight points based on difficulty + race time
22:18
total points would be weighed by your best ranks and farming would be much less effective
Avatar
louis 2021-05-02 22:19:26Z
bad
22:19
separate points and rank points is better
Avatar
lynn 2021-05-02 22:19:39Z
farming is boring
22:19
I would rather spend my time improving
Avatar
louis 2021-05-02 22:19:56Z
playing the same map 500x can be boring for some
Avatar
lynn 2021-05-02 22:20:04Z
doesn't necessarily mean that
22:20
remember that beating a map still awards base points
22:20
just clear more difficult maps
22:20
if your number one score is a moderate, beat a brutal
Avatar
louis 2021-05-02 22:20:52Z
thats like osu pp system
Avatar
lynn 2021-05-02 22:20:54Z
yes
Avatar
louis 2021-05-02 22:20:55Z
which isnt good
Avatar
lynn 2021-05-02 22:20:58Z
wrong
Avatar
louis 2021-05-02 22:21:00Z
for ddrace
Avatar
lynn 2021-05-02 22:21:00Z
xd
22:21
I think it's better
Avatar
louis 2021-05-02 22:21:13Z
top players are already really close together
Avatar
lynn 2021-05-02 22:21:19Z
points = played longer
Avatar
louis 2021-05-02 22:22:02Z
having ur points get less and less large over time isnt too good i think
Avatar
lynn 2021-05-02 22:22:22Z
btw it wouldn't be exactly like osi
22:22
osu
Avatar
louis 2021-05-02 22:22:23Z
just make some overall ranking system that combines points and times but leave it separate from points
Avatar
lynn 2021-05-02 22:22:27Z
all maps have value
22:22
just less value if it's novice
22:22
yeah I mean there can be the existing system that's fine
22:22
just another system for new rank
22:23
but I don't really think it's important to distinguish them. just means your novice scores are worth less than brutals, which is already how it is, just more accurate in terms of your skill
Avatar
louis 2021-05-02 22:24:17Z
but points doesnt rly measure skill it measures progress
22:24
idk
Avatar
lynn 2021-05-02 22:24:32Z
yes but we should have something that actually tracks skill, no?
22:24
I have very little interest in tracking my "progress"
Avatar
louis 2021-05-02 22:25:14Z
just track skill based on what maps u can beat
22:25
plus for that to work u need huge map rerates anyways since a lot of things are too highly or lowly rated
Avatar
lynn 2021-05-02 22:25:47Z
think that's independent of that
22:25
plus I don't think it will happen anyway
22:26
same with accounts, been talked about too long and no progress
Avatar
louis 2021-05-02 22:26:11Z
maybe someones "skill" rating would be like an avg of the hardest maps theyve done
22:26
plus their best times idk
Avatar
lynn 2021-05-02 22:26:21Z
I mean accounts is maybe going to change, idk
22:26
still talked about on GitHub
22:27
that's just the osu system u realize lol
22:27
but it doesn't have to be weighted as heavy as osu
22:27
osu is sooo heavy. you do noob maps and it doesn't do anything to your pp. I'm suggesting that all maps always continue to raise your score, just less the more difficult maps you have beaten.
22:28
since it should be as you say, an average of your best times/clears
22:30
I mean, rank is already like this (with times) but it doesn't do anything to incentivize clearing harder maps, just easy to optimize maps
Avatar
louis 2021-05-02 22:30:18Z
i guess that would be a good way but its basically impossible since theres no standardized rating for ddnet
Avatar
lynn 2021-05-02 22:30:27Z
that's also fixable
Avatar
louis 2021-05-02 22:30:51Z
and i wouldnt say that since harder maps have less ranks so its easy to get rpts
22:31
and its alr broken in osu where maps only have 3 elements, gl fixing it in ddrace
Avatar
lynn 2021-05-02 22:31:16Z
that's true but you can still do more moderate maps faster and see the same results
22:31
it doesn't really matter where you get rank 1, just depends on how easy it was to take, until someone beats it
22:32
it's also bad because it's based on absolute positioning, 1-10
22:32
when it should not be this way, especially for solo where there are many close times
Avatar
louis 2021-05-02 22:34:18Z
yeah the absolute position sucks
22:34
but what im saying is calculating someone's ddnet skill without seeing them play is very difficult
Avatar
lynn 2021-05-02 22:35:08Z
osu has the same issues
22:35
not possible to do it perfectly
22:35
not suggesting thag
22:35
that
22:35
just suggesting am improvement
22:35
sry I'm on mobile and ordering food so probably typos
Avatar
Deleted User 2021-05-02 22:35:49Z
someone can help me, i am working on a mod. i edited 2 files who give lovely effect (heart that fly above tee) then im trying to transform health to armor, the command work but he give health instead armor. the image is a part of code you can see "m_Type = POWERUP_ARMOR;" but that give heatlh
Image attachment
22:40
have u checked client side, if the client also gets it as armor
22:42
i think
Avatar
Avatar
lynn
same with accounts, been talked about too long and no progress
Learath2 2021-05-02 22:42:51Z
There is atleast a plan that more than one person agreed to in place now, so it's just finding someone to implement it
Avatar
lynn 2021-05-02 22:43:02Z
ok cool
Avatar
Learath2 2021-05-02 22:43:31Z
Though I guess the safety of it kinda depends on @heinrich5991 getting QUIC in, so maybe it will take some time 😛
Avatar
lynn 2021-05-02 22:43:54Z
how would migration work? everyone would have to apply to convert a name to an account or would all accounts created from scratch stay that way?
Avatar
heinrich5991 2021-05-02 22:44:21Z
my view on it: that discussion is so far away that it's not worth spending energy on
22:44
if you do want to spend energy on it, please head to github
22:44
your thoughts here will be forgotten
22:45
https://github.com/ddnet/ddnet/issues/3411
Accounts · Issue #3411 · ddnet/ddnet
If anyone wants to summarize the discussion, feel free to. I just wanted to create the issue now.
Thumbnail
Avatar
Learath2 2021-05-02 22:45:04Z
yeah just getting an implementation made would be nice for now
Avatar
Avatar
Deleted User
i think
Deleted User 2021-05-02 22:45:07Z
i'd say check if that code is really called in the server
22:45
i dont see why the client should bug around for no reason
Avatar
Avatar
Deleted User
have u checked client side, if the client also gets it as armor
Deleted User 2021-05-02 22:45:20Z
if you want see by yourself, i dont see any errors
22:45
https://github.com/Nouaaaa/Block
Avatar
lynn 2021-05-02 22:45:58Z
people who advocate for accounts have nothing to contribute on GitHub. just that they would prefer accounts. asking a bunch of people to just say "I like the idea" in a github comment is imo just noise and pointless
Avatar
Learath2 2021-05-02 22:46:22Z
That's why we don't ask for that, we asked for technical feedback
Avatar
lynn 2021-05-02 22:46:48Z
talking about it on discord is important because if there's no pressure to implement it, who would even take the time?
22:47
the more it's not talked about the more no one will care to code it :P
Avatar
Learath2 2021-05-02 22:47:01Z
And ideas on contentious points like migration
Avatar
Avatar
Deleted User
i'd say check if that code is really called in the server
Deleted User 2021-05-02 22:47:05Z
how can i check that?
22:47
set breakpoint inside that function
22:47
in the screenshot
Avatar
Avatar
lynn
talking about it on discord is important because if there's no pressure to implement it, who would even take the time?
Learath2 2021-05-02 22:47:36Z
we talked about it for months or maybe years before that issue was created, no one did it 😛
22:48
It's just that none of us seem to have time to implement it for now
Avatar
heinrich5991 2021-05-02 22:50:19Z
I'm still working on https://github.com/ddnet/ddnet/pull/3772
Add client-side HTTP server info by heinrich5991 · Pull Request #37...
Summary The idea of this is that clients will not have to ping each server for server infos which takes long, leaks the client's IP address even to servers the user does not join and is a D...
Thumbnail
22:50
I still have some problems with the client
Avatar
Learath2 2021-05-02 22:50:33Z
you can just do #3772 btw
Avatar
DDNet BOT 2021-05-02 22:50:34Z
https://github.com/ddnet/ddnet/pull/3772
Avatar
heinrich5991 2021-05-02 22:51:14Z
currently, the client probes all four master server in a random order and takes the one with the smallest latency
Avatar
lynn 2021-05-02 22:51:22Z
maybe I should start a bounty for accounts ;D
Avatar
heinrich5991 2021-05-02 22:51:25Z
the one with the smallest latency that also works
22:51
the problem I have with this is: what happens if one of them doesn't respond?
Avatar
Learath2 2021-05-02 22:52:40Z
@lynn I wouldn't, this is one feature that is extremely contentious, with dozens of opinions around. You might not like the implementation we decide on, then you don't pay out the bounty then the guy gets mad.
Avatar
heinrich5991 2021-05-02 22:52:41Z
I need some kind of timeout. but if I set it too strict, it will time out on legitimate connections, if I set it too coarse, it'll have a bad user experience if the first master that is probed times out
Avatar
Learath2 2021-05-02 22:53:30Z
@heinrich5991 500ms max should be good enough for like everywhere no?
Avatar
lynn 2021-05-02 22:54:06Z
there would be very clear goals on the bounty, and it wouldn't cover all of the expectations after implementing the base of it
22:54
I get what you're saying
Avatar
heinrich5991 2021-05-02 22:54:24Z
I think it can happen that the DNS response takes even longer tbh
Avatar
Learath2 2021-05-02 22:54:54Z
@heinrich5991 Have we considered getting rid of this master1 master2 master3 master4 stuff and moving into the future? 😄
Avatar
heinrich5991 2021-05-02 22:55:15Z
I want some redundnacy tbh
22:55
if you have a better solution for redundancy, go ahead
Avatar
Avatar
Deleted User
set breakpoint inside that function
Deleted User 2021-05-02 22:55:52Z
ok its fixed, i forgot to change "LoveEvent" to "ArmorEvent" lmao
22:56
pogfish
Avatar
Learath2 2021-05-02 22:56:02Z
I was going to suggest one load balancing node where we could do things like geo-ip to give the best mirror
22:56
don't even need to do it with dns, we could do a 301
Avatar
heinrich5991 2021-05-02 22:56:39Z
I'd like to avoid a single point of failure, one thing that can go down if cloudflare goes down, kicks us out, etc.
22:56
it's the thing that will make it impossible for players to play at all
Avatar
Learath2 2021-05-02 22:57:04Z
Yeah, I noticed that too, the master%d scheme is more robust
22:57
Well if the dns response is taking more than 500ms the old system wouldn't have worked either
Avatar
heinrich5991 2021-05-02 22:58:04Z
oh really? interesting
Avatar
Learath2 2021-05-02 22:58:30Z
I mean it'd have to wait for the dns server to respond, if I remember the relevant part of the code
Avatar
heinrich5991 2021-05-02 22:59:51Z
ah, it seems to simply wait DNS resolution out
Avatar
Learath2 2021-05-02 23:00:42Z
anyway, interesting question about the UX though, maybe we can add a prettier loading indicator to fool the user 😛
Avatar
heinrich5991 2021-05-02 23:01:30Z
anyway, that's good news, I don't need to think of something complicated
23:01
it's mostly a first launch issue anyway, the user won't be in the serverbrowser before I select the master I guess
Avatar
Learath2 2021-05-02 23:04:03Z
@heinrich5991 well that's assuming they don't have an awful connection, if you assume that your initial issue isn't an issue at all :/
Avatar
heinrich5991 2021-05-02 23:06:41Z
I was thinking that I could choose a longer timeout on the basis that this is only bad UX on the first launch, and probably not even there, because the user will take a little time to get to the serverbrowser
23:07
e.g. 10s timeout
Exported 199 message(s)