Galaxy Updater is here!

With skepticism, reason, and testing?

Why yes, yes it is.

…testing and reason you say? :wink:

Sorry to be slow to respond, but I am still digging out from all the chaos of an international move. My point is that I think by the time you add all of the infrastructure to make nuget work as something it wasn’t designed for, then what’s left in terms of what nuget wins you is pretty trivial. The design choice I made is to use git (via libgit2sharp) as the versioning engine with an Azure storage back-end, supporting all the nuget concepts, plus branching. Branching becomes important when you want to bifurcate a population of devices and create independent histories for them. Again, you can simulate this with nuget, but again, all that bending of nuget ends up eclipsing the cost of just doing a fit-to-task store.

And that’s all aside from the fact that this is the equivalent of ‘off-label’ use of a prescription drug. Policy changes at nuget could break your model and you wouldn’t get so must as a disinterested shrug from them since this is an off-label use. There’s just no SLA for this type of service bending.

I do understand that different requirements might end up driving different design decisions - this is just what I ended up with for the requirements I have.

1 Like

@ mcalsyn - Sounds good to me. Seems like you’ve reasoned out your solution.