On Wed, Sep 23, 2015 at 12:55 PM, Julian Foad <julianf...@btopenworld.com> wrote: >> Johan Corveleyn wrote: >>> [...] stefan2 told me in person that that part of the >>> change in r1572363 was unintentional :-). IIUC, he didn't realize that >>> it would have this effect on the output of dump. > [...] >>>>> I think the dump.c part of r1572363 and r1573111 should be reverted / >>>>> fixed so that we get the previous behaviour again, and this should be >>>>> backported to 1.9. At this point, IMO 'svnadmin dump' is broken in >>>>> 1.9. > > 1. I pretty much agree now that we should revert the change. > > We understand now that the change in 1.9.0 was unintentional, and > after we analyse the situation, we are very unlikely to conclude that > that change was a complete bug fix to the whole issue of no-op > changes. It is surprising and is regarded as losing information, and > is not justified (yet) by some higher purpose. > > It seems fairly clear what the change was, and so how to revert it. > > 2. We might also want to make another change to the behaviour of > 'svnadmin load', so that the result of loading a dump file that people > have *already* produced using 1.9.0-1.9.2 will be the same as if they > had dumped and loaded using 1.8.x. I don't yet understand the details > enough to know whether this option is possible. > > 3. I firmly believe that our handling of 'no-op changes' is mistaken > and a bad idea. I'll explain that, but not in this thread -- that's a > follow-on task. > > Brane wrote: >> I also suggest adding a note to >> http://subversion.apache.org/docs/release-notes/1.9.html#issues . > > And we need to file an issue. > > I'll do both of those things (issue and rel-notes) now.
Thanks. I'm a bit swamped by $dayjob now. -- Johan
