I created a package which allows one to disable a mergebox. B disabling mergebox one chooses to send possibly unnecessary data to clients (because they already have it) and not maintain on the server side images of clients’ data, thus reducing CPU and memory load on servers.
This is really untested and I do not really know if it even works. Feedback welcome.
Hmm how is this (or is it not) different from fetching data through a method or “non-reactive” find?
That it is reactive. 
It behaves exactly the same as normal publish, only without caching on the server side.
I just read the extended readme. It would be great if you could provide some real life use cases as examples. Although the benefits seem very clear, I could not bring myself to try it out due to the big list of “be careful with” situations.
If you do know why would you use it, you do not have to use it.
Have you read the merge box: https://github.com/meteor/meteor/issues/5645
See also: https://youtu.be/783BL__zIjY?t=1056
Yes I have but honestly I have not come across a usecase other than
chat/logging/geotracking where there’s high velocity only/mostly write data
and a few pointers might help break mental blocks like mine
Yes. I also do not need it in any of my projects. But people needed it, I knew how to do it, so I did it.
Fair enough. And double-high-five for taking the time to contribute something you don’t even need yourself!