Please Add IPv6 support

[Nutz]

My initial testing with udp6 is that it is always slightly faster (lower latency) than udp4, even over tunneled connections. It would be great to see mta support it. All of my servers

are IPv6 connected now.

The best mode of operation would be dual stack, where the server process accepted connections simultaneously on ipv4 and ipv6 for the same game instance.

[orappa]

Sounds good, but doesn't IPv6 have really long unmemorable IP addresses?

[MAD_BOY]

if i remember correctly, yes it does

[Harry]

Just 6 instead of 4 digits, most of the time in another notation. But what's the problem? Just use hostnames and not the adresses... Or a ingame browser with names..

[Nutz]

Sounds good, but doesn't IPv6 have really long unmemorable IP addresses?

It doesn't have to. Notation standard is to collapse a long string of zeroes into just ::

A server would typically be on a subnet like 2001:db8:1234:5678::/64

It is perfectly acceptable to manually assign the host id instead of using

automatically generated EIU-64 numbers (based on mac address).

In fact in the real world server addresses are almost always manually assigned.

So a typical address might be 2001:db8:1234:5678::12

where 2001:db8:1234 is the global portion of the address, 5678 is the

site subnet and 12 is the host id. the :: replaces a long string of

zeroes that fill out the address.

If the site/host uses zero instead of 5678 for the site subnet id it would be even more compact:

2001:db8:1234::12

This is the typical length of address you see for servers in the real world

and is not much longer than a typical ipv4 address. In addition for now most addresses will have 2001 as the first 16 bits which is easy to remember.

Finally, even if the addresses were cumbersome, aids like ASE mean you don't have to work directly with addresses.

For blacklisting, you would only ever worry about the first 64 bits so

wors case would be 2001:0db8:1234:5678::/64

However, since most client pc's will have eiu-64 host id's you could block based on the last 64 bits and even if they get a new isp/subnet you still know who they are.

[eAi]

We might consider it for Blue, but tbh even then it'll be very low priority.

[Blokker_1999]

IPv4 will remain active for about 25 more years if i'm not mistaking. offcourse v6 has some benefits but there is no rush to make things compatible.

[Nutz]

IPv4 will remain active for about 25 more years if i'm not mistaking. offcourse v6 has some benefits but there is no rush to make things compatible.

It's never too early to start coding with IPv6 in mind and it usually

involves minor changes. I realize that right now it's not a priority but

eventually it will be (and maybe sooner than you think) The earlier you

make it compatible the less retrofit you will have to do later on.

[Slothman]

I wouldn't say It'll take 25 years. hell, nobody could possibly conceive what will be going on tech wise. I remember the first time I heard of this new protocol. "the world is running low on ip addresses" was their main pitch (they barely mentioned the other advantages)

[Gtatari]

25 years ago people were still homebrewing WAN's, and dialing up local networks; I have a collection of GEOS Software, still boxed, with instructions, and it talks about "Q-link" the wide-area-network for Commodore 128! Its fun, and dated 1985! thats only 20 years ago,... I love historic software and video games. But the future is definatly uncertain, soon we we'll envy the people who have T3s going to their houses while we are stuck with our measly broadband, god forbid we have dial-up. I think mta should be v1.0 before something like this be implemented.

For what its worth, thats my two cents. (probably worth less than $0.02)

)>-GTAtari-<(

[Slothman]

geos!!

I want it!!!