reply to all instead of reply to Eric only. On Mon, Dec 29, 2014 at 10:11 PM, Eric Sunshine <sunshine@xxxxxxxxxxxxxx> wrote: > In v6, you considered this a fatal error in the atomic case, which > caused the entire transaction to be rolled back. However, in this > version, this error has no effect whatsoever on the atomic > transaction, which is a rather significant behavioral departure. Which > is correct? (This is a genuine question; not at all rhetorical.) If > failing the entire transaction is the correct thing to do, then this > is going to need some more work. I don't know. in v6 I thought *any* error would stop and abort the atomic commit. An atomic commit is either completely free of failures or it doesn't work out. However that warning doesn't seem to have any effect as of now (apart from warning the user obviously), and is about the (highly unlikely?) case of a bug in git (or a provided helper script/hook?), I thought we can go with it as well. The transaction wen't through so all is fine. The thing the warning is about is about reachability checks for the shallow case, so I don't see why that should fail the transaction. -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html