On Tue, Oct 10, 2017 at 4:47 PM, Dan Kenigsberg <danken(a)redhat.com> wrote:
On Mon, Oct 9, 2017 at 11:12 AM, Piotr Kliczewski
<piotr.kliczewski(a)gmail.com> wrote:
> All,
>
> Recently you could notice few patches being merged and later reverted.
> I would like to say what happened. Sometime ago I pushed a tag to
> vdsm-jsonrpc-java and later reused that command from my history to
> push 6 patches [1][2][3][4][5][6] to vdsm for a review. I did not
> notice that I pushed for `heads/master` instead of `for/master` this
> caused to bypass gerrit and the patches were merged. When I noticed
> what happened I pushed [7] to revert it. I hope no-one was affected.
>
> Thanks,
> Piotr
Thanks for your report, Piotr.
I believe that [7] should have been sent properly, via gerrit, and not
by direct git merge. Unless we close git access, errors would happen.
But when you realize you've made a mistake, better fix via gerrit.
You are right. I should have done it via gerrit. My only justification
is that I wanted to
reduce potential issues that people could have.
In particular, I think [7] has mistakes; it is not clear to me how you
have produced it.
I did git revert --no-commit for all the patches. Unfortunately my
source tree was not
clear.
fc1174c91~ is the last commit before the mess occurred, right? da18b865 is [7].
But
$ git diff --name-only fc1174c91~..da18b865|cat
.cache/v/cache/lastfailed
init/systemd/systemd-vdsmd
is not empty. It introduces two files with no good reason.
I noticed this as well and pushed [8] which removes those files.
[8]
https://gerrit.ovirt.org/#/c/82648/