DDraceNetwork - developer

i can also just patch is, we have a microsecond implementation that doesnt cache

00:16

its safer anyway to not use time_get in a different thread

Jupeyy

[ddnet/ddnet] Pull request opened: #3912 Use time_get_microseconds in CChoseMaster's job

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
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--u...

Jupeyy

[ddnet/ddnet] Pull request opened: #3913 Use C++

Lets finally move to c++, there is no real disadvantage, but there are advantages:

std::chrono
no C89 limitations or whatever MSVC uses
no functions with unspecified parameters

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.c especially
[ ] Considered possible null pointers and out of bounds array...

1

1

Jupeyy

[ddnet/ddnet] Pull request opened: #3914 Update server list silently

So there is always a list rendered(not the "Getting serverlist" string) This isn't really finished. The goal should be to also cache the server list to disk and load it, so there is never an empty list. Also currently I just made the "Refresh" button say "(Fetching) Refresh" to be clear the list might change. Personally i think greying out the list could be a way to show it "more elegant". Anyway, didn't really test it, if somebody wants to test out, give feedback or even build upon ...

deen

Would it help to pay for Cloudflare? I think they have better routing for paid accounts

Nope, CF routing is shit also on Paid accounts

07:32

If they have problems in a location, they just shut down It, Milan location for example is online 1 time over 10 /year

07:32

3

Q: Assuming empty caches, a host pinging another host in their own subnet would just do an ARP request, no?

plus ICMP ping

11:48

what do you mean by "pinging"?

Can you see any reason for this insanity?

11:49

Ah at A...

yes, only the ARP request is broadcast

At the switch we'd see ARP Req, ARP Reply, ICMP Req, ICMP Reply. Right?

yes

The way they drew it made me think it was a shared bus, not a switched network

Is it not possible to set up a limit for the console? For the test server it would be very useful because sp someone always comes when we test on the server and causes with his spam timeouts

12:19

I already asked @murpi but he said it's impossible?

It is possible but it requires some engineering and I'm not completely sure how best to fix it

12:34

@heinrich5991 any idea on how to fix it?

#3691

https://github.com/ddnet/ddnet/issues/3691

flooding the rcon console can timeout authed players · Issue #3691 ...

Title 29 pages of this within the same timeframe

I think we talked about it before for a bit but I don't think we came up with any solution that satisfied both of us

can you change map on the testing servers?

12:35

with the rcon

I think it also annoys murpi that I have to ask him all the time if we can get a server with a new pw

Well the map changes only used to cause it because they spammed fron with vote added

Cøke

I think it also annoys murpi that I have to ask him all the time if we can get a server with a new pw

I don't mind

s/fron/rcon/ wtf

12:36

I think the issue was that we are running out of space in the retransmit queue?

yes

12:36

we ignore the error though (not a good thing)

12:36

and when the client asks for retransmits, it'll eventually get dropped because we don't have the packets anymore

how would we even sanely handle running out of space? fwiw imo this shouldn't even be possible to begin with

do we use the nuclear test servers for anything specific?

Some throttling for rcon log lines or maybe merging duplicated lines?

we either throw lines away or disconnect the player

12:38

I don't see many alternatives

rcon messages are VITAL, maybe they shouldn't be vital?

we could also make the buffer larger, making it harder to reach

12:39

we could introduce a new vital message type that is only actually sent if the queue is less than 50% full

or maybe we introduce something like SOFTVITAL, if we can reliably deliver it

12:39

heh, we came up with the exact same idea

no, you want to send it as non-vital, I want to drop it on the server

12:40

it's similar, I guess

The first one I said was sending it non-vital, but that's not really good. I was thinking the exact same idea for SOFTVITAL

ah

if the queue isn't too busy deliver reliably, if not either send it non-vital and pray for the best or drop it

just drop it, shouldn't happen during normal usage

12:41

maybe also introduce a log message that it happened

12:41

and watch out for it in the logs

12:41

(rate limited!)

I don't know anything about this, but how about doing it like twitch, all non testers are allowed to make an entry every 10sec.

then you can't use stuff like up anymore, I guess?

U can use tele

I would take a look at it but I have a computer networks exam tomorrow and even though I know the material I'll feel bad if I don't atleast pretend I studied for it

there are alternatives, sure

12:42

but up won't be usable anymore?

12:43

or quickly respawning at a tele to retry a part?

I don't think so

sounds like a bad idea

We could also just ratelimit the rcon lines per second as a quick patch and just drop the messages

Maybe 1sec is enough?

well everyone can log in on test servers so that's not particularly useful

But as non tester u got only helper role

@heinrich5991 ah I have one more idea, we can also separate out the consoles finally, so not every logged in person gets the entire server log

This guy gave 11.68ms transmission time for one packet in an exercise. What kind of evil entity gives such a horrible number in an exam without calculators?

bors[bot]

[ddnet:master] 2 new commits

942cbba Init mouse diff floats in editor - Jupeyy a4e2457 Merge #3905 - bors[bot]

13:38

bors[bot]

[ddnet/ddnet] Pull request closed: #3905 Init mouse diff floats in editor

bors[bot]

[ddnet:master] 2 new commits

a43d169 Use time_get_microseconds in ChoseMaster class - Jupeyy 4a13ba2 Merge #3912 - bors[bot]

14:14

bors[bot]

[ddnet/ddnet] Pull request closed: #3912 Use time_get_microseconds in CChoseMaster's job

what are the conditions for a successful vote

NeXus

what are the conditions for a successful vote

Exclude Nexus from the vote troll

Oh cmon its quiet easy. You need to go with a free bottle of beer to everyone who can make a vote and try to reason them why they should F3 =]

Include LCSG47 troll

Tsin

Oh cmon its quiet easy. You need to go with a free bottle of beer to everyone who can make a vote and try to reason them why they should F3 =]

It almost feels like you have to seo ur reasons and do everything so people start to vote

@NeXus ingame: all of the players who have veto must agree, and a majority of the players must agree

Veto for map vote I suppose?

Welcome to elections this is how nowadays voting works =]

afk players don't count, people get veto if they're >20min on the server or >20min in the race

15:28

ah yes

15:28

true, vetos only exist for map votes

We try to kick someone and I've noticed this serveral times now, its almost impossible to kick a blocker by vote

15:29

bcs most people just dont vote and I guess their vote counts f4

how long do you have to be afk to make it count (exclude you from voting) ?

seems like 2min?

15:31

on official servers. default is 5min

https://www.bbc.com/news/technology-57443598

Windows 10 to be retired in 2025, as new OS unveils

From October 2025, there will be no new updates or security fixes for Windows 10.

15:44

15:46

https://news.ycombinator.com/item?id=27514130

mikesabbagh

Windows 10 to be retired in 2025, as new OS unveils

hum

15:46

looks weird, I thought win10 was forever

same

I guess it's just new branding?

While PC sales still dominate - 79.4 million shipped in 2020, according to Gartner - Google's alternative is proving popular, with 11.7 million Chromebooks, which run on Google's Chrome OS, shipping in the same timeframe.

15:47

chromebooks creating competence?

Honestly, HN may want to just consider banning any site that runs this news article from being posted on HN, because this claim is basically proof they did not do even cursory research before posting.

>

Here's the 2015 story covering this date announcement, which bears no relation to recent rumors at all: https://www.ghacks.net/2015/07/20/microsoft-to-support-windo...

ye im reading that

15:48

but i dont think they will ban bbc

I'm more commenting on the story itself

15:49

i.e. that it's not clear that this is some actual news

@heinrich5991 || @Ryozuki do either of you know an expression that will easily get optimized out but would fail if a field of a struct doesn't exist?

16:38

s/fail/fail to compile/

16:38

(in rust)

&struct.field;

struct.0?