DDraceNetwork - developer

What is the maximum packet size for a client to have no problems with a DDNet based server?

https://github.com/ddnet/ddnet/blob/8d7442e328282bce902aa0c9667da5529a7b392e/src/engine/shared/network.h#L58

Yek

What is the maximum packet size for a client to have no problems with a DDNet based server?

what are you doing?

Yek

Will aarch64 support on Linux ever be added?

is aarch64 not supported? do you mean that you want to have binaries on ddnet.org?

heinrich5991

what are you doing?

I want to limit with my own firewall the pps/bps per IP and the maximum packet length If it is higher, the IP will be blocked for a while

Yek

I want to limit with my own firewall the pps/bps per IP and the maximum packet length If it is higher, the IP will be blocked for a while

the link I sent above gave the answer btw. 1400 bytes of UDP payload

Ty thx

If you try to avoid ddos attacks those are usually spoofed ips with proper teeworlds packets so that wouldn’t be filtered :(

06:17

What about dummy’s multiple clients. Or multiple users sharing the same ip? (@Yek)

Yek

Ty thx

1400 is close to the MTU

07:19

https://en.wikipedia.org/wiki/Maximum_transmission_unit

Maximum transmission unit

In computer networking, the maximum transmission unit (MTU) is the size of the largest protocol data unit (PDU) that can be communicated in a single network layer transaction. The MTU relates to, but is not identical to the maximum frame size that can be transported on the data link layer, e.g. Ethernet frame. Larger MTU is associated with reduc...

https://www.nvidia.com/en-us/geforce/news/dota-2-geforce-rtx-nvidia-reflex-support-available-now/

NVIDIA Reflex Available Now In Dota 2 - GeForce RTX Gamers Can Redu...

Get the lowest latency and best responsiveness by enabling NVIDIA Reflex on GeForce RTX and GeForce GTX 900, 10 and 16 Series graphics cards, desktops and laptops.

08:28

Is thisntrue

08:36

maffs nerd

08:37

also congrats on being top1 active githubber i follow ryo! You make up 90% if my feed while being only 1 of my 26 followed dudes

08:37

of

haha

def-

[ddnet/ddnet] Issue closed: #6009 Updater says it failed when it actually succeeded on win7

09:45

def-

[ddnet-scripts:master] 1 new commit

2a7eb13 Add config_directory.bat - def-

if(s_SelectionPopupContext.m_Entries.size() == 1)
{
    s_SelectionPopupContext.m_pSelection = &*s_SelectionPopupContext.m_Entries.begin();
}

Does this result in undefined behaviour? m_Entries is a std::set<std::string>. I'm want to get a pointer to the first element. As you can see I ensure that there is exactly one element in the set and the set will not be modified in the meantime, so taking a pointer should be fine. But if I dereference the iterator with *, doesn't this create a temporary copy of the entry that I would then take the address of with &? In that case the address would point to a temporary object that is no longer valid outside the scope of this if-block. I could also do the following, which should definitely work, as it uses a reference to the string, albeit it looks a bit odd:

if(s_SelectionPopupContext.m_Entries.size() == 1)
{
    for(const auto &Entry : s_SelectionPopupContext.m_Entries)
    {
        s_SelectionPopupContext.m_pSelection = &Entry;
    }
}

dereferncing doesn't automatically create a copy. You just have to be carefuly to only dereference valid pointers, not null pointers for example

13:01

the first version looks fine to me, just copies it into the new target object

I think the *& might even just get removed during the first optimization pass

you mean &*

That too, but even from a standard perspective seems fine

ok, thanks

Oh actually maybe not now that I think about it. The * is probably overloaded here

13:03

Anyway, first one is indeed sane

You can try adding some copy constructors that print something when the object is copied to verify that copy is only done when expected

Or you could delete the copy constructor if you dont use it

The entries are std::string, so can I even do that?

Assigning to a reference is a good example where the object is definitely not copied

Robyt3

The entries are std::string, so can I even do that?

Then nope

if m_pSelection is a std::string&

13:05

ok, then set gdb breakpoints on them

13:05

but anyway, should be fine

It's a std::string *, because I also need it to be a nullptr when it's empty

wouldnt ub sanitizer show its ub or smth?

ub sanitizer doesn't detect all ub

ah too bad

13:06

the standard library is full of undefined behavior btw

rly?

13:07

well and lot of ppl rely on that right

13:07

thats why the std cant be updated

Now that I think about it a bit. Why do you think the * will create a copy anyway?

13:07

It'll just be a normal lvalue

Ryozuki

rly?

Heard it from here a while ago I think: https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1705r1.html#stdLibraryUB

Learath2

Now that I think about it a bit. Why do you think the * will create a copy anyway?

I thought the copy is already created when dereferencing. I'm not really familar with different value types, I'm coming from a Java world

Searching through the standard

Oh, I see. Yeah I guess Java does do that

deen

the standard library is full of undefined behavior btw

I think this is even intended, no? The point is to allow implementations to optimize agressively iirc

Yes, of course

13:13

fewer branches

and more security issues

chillerdragon

What about dummy’s multiple clients. Or multiple users sharing the same ip? (@Yek)

Game mode does not allow dummy

huh? https://github.com/swarfey/tw-chatonly is gone

14:15

kekw

GitHub

Click to see attachment 🖼️

@deen how is it fixed ?

Sad I will still try xdp to see if with specific filters it does something against an attack

14:22

Chairn

@deen how is it fixed ?

By removing the wrongly added integration test files from updater json: https://update.ddnet.org/update.json

ah okay

14:24

log doesn't say which error triggered the "error updating" UI

the 404s triggered it

14:25

old file missing "error" is ignored

gerdoe

huh? https://github.com/swarfey/tw-chatonly is gone

lol even the username

yeah

I guess since no one reported it before you, and this was already in 16.3, only few people are still using the ingame updater

14:26

@Swarfey hi ^

maybe he wanted to disappear

14:27

updating manually, i always end up overwritting some settings file

14:27

like settings_ddnet.cfg and storage.cfg

most people use steam updater

i use portage

14:29

the gentoo guru maintainer is pretty fast btw

14:29

on the same day

14:29

oh

14:29

i say this but its still 16.4 f

1

14:29

https://github.com/gentoo-mirror/guru

GitHub - gentoo-mirror/guru: GURU: Ebuild repository entirely maint...

GURU: Ebuild repository entirely maintained by Gentoo users - GitHub - gentoo-mirror/guru: GURU: Ebuild repository entirely maintained by Gentoo users