On Mon, Dec 8, 2014 at 5:07 PM, Julian Foad <julianf...@btopenworld.com> wrote:
> Stefan Fuhrmann wrote: > > Looking through ^/subversion/branches, I found that there are many > > backport branches that are not mentioned in any STATUS file. > [...] > > > 1.8.x > > Julian: 1.8.x-r1619380 (not modified) > > Ugh. Thanks for reporting this discrepancy. There is a mess here. > A typical case of BCD ;) - Bad Commit Day or Bad Code Day. > First, it IS modified; I committed the branch creation and modification > together ;-) > May bad, I didn't log verbosely. -v revealed that Bert: 1.7.x-issue-4295 is actually modified as well. > It was proposed for backport, but the process went wrong. > > "svn log --stop-on-copy --search 1619380 --diff" shows: > > In r1619394 I "Nominate r1619380 group (diffing a copied dir with props)". > > In r1619401 I "Mention the backport branch for the r1619380 group" ... by > adding this to the nomination: > > + Branch: > + ^/subversion/branches/1.8.x-r1611379 > > Oops, I wrote the wrong branch name (copy-and-paste-o). My apologies. > > Then Bert and Stefan approve this nomination, and in r1640665 the > 'svn-role' robot merges the mentioned (wrong) branch. > I obviously didn't check the branch contents vs. the claimed revisions :( Luckily, the named_atomics code modified in r1611379 should be ineffective in 1.8.x by now. > The r1611379 branch was for a separate nomination, "Fix revprop caching > initialization delays for ra_serf on Windows". It was first mentioned in > STATUS in r1612544 and was WITHDRAWN in r1620644, replaced by r1619774 > "Disable the revprop caching feature". > > But it has been merged anyway. So now we need to: > > * undo the merge of the 1.8.x-r1611379 branch > +1. * re-nominate my original nomination. > +1. -- Stefan^2.
