Hacker News Reader

I strongly suggest Shapiro's paper[0] on CRDTs. In a nutshell, the only really problematic data type are sequences such as arrays or strings. There are some specialized approaches specifically for those, and you can always fall back on a LWW conflict resolution.

In general, I like to think of Paxos as an approach that uses LWW for all value types.

[0] - http://pagesperso-systeme.lip6.fr/Marc.Shapiro/papers/RR-695...

View on HN

Shapiro's name is on about seven different CRDT papers :) it makes citations difficult for the Wikipedia article. Personal opinion, this[0] 2011 paper is probably the best one to read, where Shapiro's and Baquero's teams finally joined forces and put out a good comprehensive paper on the subject. The one you linked focuses a bit too heavily on TreeDoc in lieu of good treatment of CRDT theory. Their survey of known CRDTs[1] is also worth reading.

[0] https://hal.inria.fr/file/index/docid/609399/filename/RR-768...

[1] https://hal.inria.fr/inria-00555588/document

Oops. I had meant to link the survey. His lecture[0] on the subject is quite approachable, for those who prefer visuals to papers.

[0] http://youtu.be/ebWVLVhiaiY

Reply on news.ycombinator.com