IPv6 support for ENet networking module

[akien-mga]

Now that we have IPv6 support in the engine, we need to extend the ENet module to support it too in the NetworkedMultiplayerEnet API.

AFAIK upstream ENet has no IPv6 support, and there is still no sign upstream of willing to add it. There are two pending PRs adding it (not reviewed):

• IPv6 support lsalzman/enet#21 (needs rebased)
• Yet another Ipv6 patch lsalzman/enet#51 (merges cleanly, recently rebased)

As IPv6 support is a growing concern and ENet is quite popular, I imagine that there might be many forks out there that add the feature. We should try to find one which is well maintained, and use it as a replacement to our upstream ENet - hacking it ourselves would be a waste of time (unless there are really no good forks implementing it).

In this issue, we should thus list all forks/branches we can find and try to assess them. Then we could try to get in touch with those developers and see if they'd be up for making their repo an alternative upstream that can be used by Godot (and maybe eventually packaged in Linux distros as new enet version if the lsalzman/enet repo continues ignoring PRs).

[akien-mga]

Forks / IPv6 branches I found so far:

• https://github.com/freeminer/enet/tree/ipv6 (used in the Freeminer game, recent activity).
• https://github.com/Ericson2314/enet (needs rebasing, inactive)
• https://github.com/cgutman/enet (recent activity)
• https://github.com/thewolfwillcome/enet (uses PR21 mentioned above + some more commits)

That's it for the GH forks AFAICT.

Edit: Also this one which is not registered as "GitHub fork", and quite old:

• https://github.com/csm/enet

[akien-mga]

The Dolphin emulator had plans to use the freeminer fork, but as far as I can tell it hasn't happened yet.

At any rate, from the above list, I think the most relevant to review are @proller's @freeminer fork, and @cgutman's.

[Faless]

I hope I'm not hijacking this issue.
I've successfully merged lsalzman/enet PR 51 into godot enet version and updated the enet module to the new IPv6 API. It seems to work so far. It should support both ipv4 and ipv6 (since the original patch uses dual stack sockets).
@vnen , @akien-mga , @punto- can you have a look? https://github.com/Faless/godot/tree/enet_ipv6

[akien-mga]

@Faless I'd prefer we don't modify thirdparty/enet further than syncing with https://github.com/freeminer/enet/tree/ipv6, so if it needs more work, it would be nice to make a PR to the freeminer repo.

I don't want us (yet) to become maintainers of yet another enet fork, and I would very much prefer that everything we need to modify is also proposed upstream via PR 51, so that we can tell distro maintainers that if they want to build against system enet, they need to cherry-pick this PR.

[Faless]

@akien-mga , I see your point, and I agree. I don't want to patch enet either.
I guess we could just use the freeminer fork. I've synced the source and added a single commit (mostly defines) to get the godot build system to work properly. We could try to get the small UWP/WINRT fix into the freeminer branch (but not having windows I can't really try UWP builds). Also, defines could be added to scons somehow, but I don't know how that works either 🎱 .

[akien-mga]

Also, defines could be added to scons somehow, but I don't know how that works either 🎱 .

env_enet.Append(CFLAGS=["-DYOUR_DEFINE"])

[vnen]

BTW, I got the UWP patch from here: https://github.com/ahrnbom/enet-uwp

[akien-mga]

If @proller is up to it, we could probably ask him to merge his own ipv6 branch into his master branch (or into a new "development" branch or similar) where we could make new PRs without polluting the pending ipv6 PR to the upstream repo.

[akien-mga]

Alternatively, we could have a godotengine/enet repo with all the further developments we need in one public place, if the freeminer guys don't wish to become an upstream for another project.

[punto-]

How hard would it be to patch ENet to be able to provide our own socket
connections? That way we eliminate the duplicate code (and the bsd and
winsock network code from ENet), and it'll work with any OS that we support
with our own sockets (including websockets for example)

On 2 November 2016 at 18:55, Rémi Verschelde notifications@github.com
wrote:

Alternatively, we could have a godotengine/enet repo with all the further
developments we need in one public place, if the freeminer guys don't wish
to become an upstream for another project.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#6992 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AGVmPRSRKzST2qXRUDBCQETTJ_s9pHVXks5q6QbpgaJpZM4Kk3BW
.

[reduz]

There is a file that needs to be reimplemented, did not seem that hard but
its work

On Nov 2, 2016 21:33, "punto-" notifications@github.com wrote:

How hard would it be to patch ENet to be able to provide our own socket
connections? That way we eliminate the duplicate code (and the bsd and
winsock network code from ENet), and it'll work with any OS that we support
with our own sockets (including websockets for example)

On 2 November 2016 at 18:55, Rémi Verschelde notifications@github.com
wrote:

Alternatively, we could have a godotengine/enet repo with all the further
developments we need in one public place, if the freeminer guys don't
wish
to become an upstream for another project.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#6992 (comment)
,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/
AGVmPRSRKzST2qXRUDBCQETTJ_s9pHVXks5q6QbpgaJpZM4Kk3BW>
.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#6992 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AF-Z2yquAfb2wf1V-d1wRcbSHRGJ3f9Pks5q6SvigaJpZM4Kk3BW
.