I'm wondering what the direct connect priority is comparing to the other normal requesters. It seems the direct connect request priority is fairly low.

I think LordKTor told me it got the highest priority.

Thanks for the reply! I use direct connect method extensively in my Py playthrough. The biggest issue seems to be that TSM doesn't balance multiple direct connect stations. If the production is less than consumption, one of the request station always get all the supply.

I also noticed the normal lamp-requester will show the time waited, and it follows a first in first out order. Such mechanism is missing for direct connect requesters. It would be nice to make the direct connect request work better. Each component in my current playthrough requires 6-7 ingredient, and being able to set it using direct connect is much simpler than setting up 6-7 lamps and decision-combinators.

I'm playing around with the code a bit. It seems to be a non-trivial change.

In the ideal case
* each direct connect station should act as a dummy "publisher" with "tick" and "proc_priority" attributes as normal lamps.
* the update events for direct connect should have a similar cooldown as normal publisher.
* unsatisfied direct connect request should also be registered to newrequest table. Then the find_best_match() during train state change on supply station should be able to find the pending request with the right priority and wait time.

Here's the PR to show the above ideas: https://github.com/billbo99/train-pubsub/pull/2 There are probably more corner cases than I original anticipated.

I play with friends in a dedicated server, do you know if it's possible to run a modified/unpublished mod in multi player?

I play with friends in a dedicated server, do you know if it's possible to run a modified/unpublished mod in multi player?

yup, the server and all the connect players must have the unpublished mod for it to work