Sxderp

I guess I never gave a full rundown on why this is needed. The master creates a [publish server](https://github.com/saltstack/salt/blob/bd89384259c4fd577f60273b4fc6bb465b98e6f9/salt/master.py#L819-L822). The [factory method](https://github.com/saltstack/salt/blob/bd89384259c4fd577f60273b4fc6bb465b98e6f9/salt/channel/server.py#L968-L970) creates a `MasterPubServerChannel` with the chosen transport using...

@dwoz I think this is fully ready. The CI is not running "core tests" so my test isn't actually executed. I'm not sure if there's something I need to do...

I have rebased onto 3007.x. I have also extended the PR to include the other methods of PublishServer. This will partially fix #66577 (follows are required to completely fix that...

I think the write callback in transport.tcp is a holderver from IPCServer. I could not trace any code that actually used it. Local tests seem to agree.

@dmurphy18 requesting Workflows get ran again.

Small fixups to get tests to succeed.

Well, that's embarrassing. I missed a linter item. Quick fix incoming for that. Since GitHub makes it hard to find previous CI runs when a commit is pushed. Here is...

Flaky tests? They work locally and were successful on the previous commit (before my lint fixup).

Rebased onto 3007.x. No conflicts.

As the upstream changes I keep rebasing and pushing to bring the branch up to date. I'm wondering if I should keep doing that or to leave the branch "as-is"?