Conflict management: About arbitrarily choosing the "winning" revision...

[jblancka]

http://guide.couchdb.org/draft/conflicts.html

"The choice of picking the winning revision is arbitrary."

If I understand the winner picking algorithm well, it picks the revision with the highest revision number when replicating (it is thus not really arbitrary). However, I understand the use of the word "arbitrary" in this sentence, since this is not something you should rely on. Although it can seem confusing to CouchDB beginners. I started with a CouchDB project and kept wondering why the algorithm always picked the revision of the source database as the winner and never the target one. It was because my application only updated the source and never the target. Which on its turn causes the source always to have the highest revision number. The source thus always won.
A note next to the term "arbitrary" might be appropriate (in the sense of "arbitrary, but..."), to prevent unnecessary headache for people who start out with CouchDB.