Closed mvahowe closed 4 years ago
My point is that we keep saying that "servers retain history" but, as soon as a project moves between servers, it becomes very hard to reason about that history, unless the burritos themselves contain some metadata to facilitate this. So, eg, PT could potentially grab commit 6 from uW, but to do this it needs to be able to walk back through the basedOn links. Without that, for PT, the uW period of the project is like a night out with too much alcohol - something happened but we have no idea what.
To put it another way, how would we produce the history of an ingredient (names and dates) for a burrito that has moved between idServers? So something like this, from github? This seems like quite basic functionality if we want interoperability to be a realistic option.
I think we can do all this from the metadata (assuming the metadata is available). But without some help from the metadata, I don't think it's possible without introspecting two vc systems which may not even be compatible with each other.
With the basedOn field and an API that encapsulates different idServer's history mechanisms, a client could request the history of a project and reconstruct a combined history of the burrito.
I think we had this conversation and there were no action points so I'm going to close this.
What do steps 4 and 5 look like in terms of vc commands?
(To make it artificially easy, let's pretend that PTX and uW use the same vc system.)