On 29 July 2014 22:14, Sandy Ryza <sandy.r...@cloudera.com> wrote: > > http://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-common/Compatibility.html
Sandy is correct is a semantic compatibility issue. Back then the notion of "interface" was defined in the early 1970s by D.L Parnas (see: http://www.cs.mun.ca/~harold/Courses/Old/CS3718.F08/Diary/p330-parnas.pdf), he made clear from the outset that an interface was a combination of (binary signature, functional semantics) -it's just that C++ derivative languages have only focused on the signature, which is what can be checked at compile and link times Policy there is "The behavior of API may be changed to fix incorrect behavior" If an AM is now expected to handle new things, then it is a change in the semantics...though if it was due to a bug fix -things didn't work before- then it could maybe get in. With YARN, there's now a (wide) interface between apps and Hadoop, and right now it is fairly intimate -you do tend to end up using things marked as private because its how you get stuff like the web proxy/filter and other things that I can't remember); I think it's going to take a while to stabilise. Ideally for this patch, I'd like an AM (including my own) to at least degrade to its existing behaviour, even if it doesn't pick up any new features. Is this the case? -steve -- CONFIDENTIALITY NOTICE NOTICE: This message is intended for the use of the individual or entity to which it is addressed and may contain information that is confidential, privileged and exempt from disclosure under applicable law. If the reader of this message is not the intended recipient, you are hereby notified that any printing, copying, dissemination, distribution, disclosure or forwarding of this communication is strictly prohibited. If you have received this communication in error, please contact the sender immediately and delete it from your system. Thank You.