It's also the best option to fix create-release. If we cannot get help to fix the script, it seems the only choice is to maintain CHANGES.txt again in branch-2.
On Tue, Jun 7, 2016 at 2:56 PM, Andrew Wang <andrew.w...@cloudera.com> wrote: > I'm not super happy about the alternative of reverting the auto-generated > CHANGES/RL JIRAs, since it means we have to do CHANGES.txt maintenance > again in branch-2. That's an overhead that we pay for every commit, and > it's particularly bad when doing backports. > > As I commented on HADOOP-12892, I think the backport should be possible, > and hopefully without backporting all the conflicting JIRAs. Maybe another > bash-savvy contributor could attempt it? > > > > On Tue, Jun 7, 2016 at 2:45 PM, Wangda Tan <wheele...@gmail.com> wrote: > >> > >> > I'm not the one who cherry-picked them into branch-2, so I don't >> > care either way. I've been extremely open about targeting my code for >> the >> > past three years for trunk. With few exceptions, I generally only commit >> > code I write or even review to branch-2 if someone practically begs me. >> > That branch needs to die. Given my current situation, I'm certainly not >> > going to spend any unpaid time on it. >> >> >> Yeah, we all knew your preferences about this, you have the choice to work >> on whatever branch you like. :) >> >> Actually this question is to Andrew and Akira, I want to get consensus on >> this. >> >> Thanks, >> Wangda >> > >