DDraceNetwork - bugs

DDraceNetwork

Development / bugs

Report client & server bugs, especially but not limited to release candidates.

Between 2022-08-09 00:00:00Z and 2022-08-10 00:00:00Z

on the map "Iris" the clous teleport back and fourth and get smaller and bigger, can reproduce on local server. the envelopes look smooth and it also doesn't happen on my renderer https://mapview.patiga.eu/?ddnet=Iris is this a known issue?

10:11

the envelope is synchronized

(I'm investigating)

10:37

hm can someone reproduce other than me? it only happens on vulkan

Patiga

hm can someone reproduce other than me? it only happens on vulkan

What version are you using?

stream version and master, am trying to bisect right now

steam*, probably

yup

10:51

okay I'm confused

10:51

trying to bisect: was hard to find a good commit first, now it gave me a commit where the envelopes are just stuck

10:51

what do I do now?

10:52

bad I guess?

you can do skip

10:52

but only if it's like a couple of bad commits

10:53

if you skip, you'll get adjacent commits

10:53

(i.e. it's no longer binary search)

can I go one step back? I already told it bad

10:54

didn't see it in the help

me neither

10:54

it's probably not good anyway

eh I'll continue for now then

if you don't find it by marking these commits bad, try again and mark these commits good

10:55

idk if you understand how git bisect works

first time I use it, I gave it HEAD as bad, and found an old commit where it worked for good

sounds good

10:56

it's basically doing a binary search between these two commits now

10:56

you tell it whether to look for the bug in the earlier or newer commits

thats how I assumed that it works

10:57

thanks for explaining though

10:57

and success btw, the next was bad again

nice

10:57

if you skip, it can't continue with binary search so it looks at the neighbouring commits

10:58

neighbours to the middle of the tested interval

makes sense

And what is it the nano second patch?

Bisecting: 18 revisions left to test after this (roughly 4 steps)

10:59

gimme a sec ^^

10:59

can you reproduce btw?

10:59

map Iris

Yes but i am only on phone rn ^^

its 53cb36bbe711dab1b333da7173deae94e46ba5f4

11:06

its really small, nice

commit 53cb36bbe711dab1b333da7173deae94e46ba5f4 (upstream/pr/5038)
Author: Jupeyy <jupjopjap@gmail.com>
Date:   Fri Apr 29 15:48:01 2022 +0200

    Fix wrong quad offset for (not caused by transparent) flushes

will still open an issue, I wrote it already :d

well, a hash is not very informative

(edited)

11:07

date, author, pr number and commit description is more informative

dunno, hash is enough most of the time. I guess I couldve written more coz we are chatting here

Patiga

can I go one step back? I already told it bad

git bisect replay, edit the output and restart the bisect

thanks :), tried it out, works nicely

git bisect log > bisect_log
--- edit bisect_log ---
git bisect replay bisect_log

bugged hook line

Image attachment

Image attachment

its either bugged for hookables or for tees

for tee it's bugged vice versa

Image attachment

its either bugged for tees or for fat tees

Exported 53 message(s)