The Cr-48 and Chrome OS: Google's vision of the net
Posted Jan 23, 2011 2:45 UTC (Sun) by
jhs (guest, #12429)
In reply to:
The Cr-48 and Chrome OS: Google's vision of the net by njs
Parent article:
The Cr-48 and Chrome OS: Google's vision of the net
I think your assessment is close enough so that further nitpicking of the minutia wouldn't be productive. Two final thoughts:
- If an improvement to the architecture is possible, the community would likely be very open to that. That would be a big change, requiring good justification, but the community is very open and flexible. I think the main problem right now is, tooling ("merge" libraries, development/debugging tools) is so much more primitive than the core database, that it's somewhat moot. The wiki's description of conflict resolution is a piece of pseudocode. In the future, it will say "If you use C, use this library; if you use Ruby, use this other." When that happens, the pain point of the history graph may become dominant.
- Tracking true "merges" is possible in "user space" if you will. Like a shadow government, the client can simply track its own history graph using its own mechanism. (In this case, it's just like Couch except merges are recorded as such.) The data is simply a normal key/val part of the record. If the algorithm proves to be superior, it could be baked into couch. (The advantage of Couch's revision tree is, like Unix dotfiles, it is not transmitted to the client unless explicitly requested. Otherwise it's a normal key/val datum called IIRC `revs_info`.)
(
Log in to post comments)
