On 04/13/2012 11:00 AM, Amar Tumballi wrote:
On 04/05/2012 03:56 PM, Florian Haas wrote:
On Thu, Apr 5, 2012 at 11:13 AM, Vijay Bellur<vijay@xxxxxxxxxxx> wrote:
Thanks for the submission. The Gerrit submission would suffice and
the patch
will go through the review process from there.
Thanks. One question about the Gluster Build System which I guess
applies to general Gerrit policy: the commit as submitted contains a
new autoconf --enable-ocf flag, which is off by default. Hence, the
build system builds with no change, really. And that default is sane,
I think -- people who have no use for OCF RAs can safely build without
it.
However, that of course means that the changes the patch introduces
aren't really tested in the build system. Does the project have a
general policy of what to do in this instance?
That is true as of now. The idea is to make sure the default setting
is not broken by any of the new patches. In future we may enhance the
build system to do more of automated tests, but for now, we are doing
it after the patches go in.
One of the possibilities that we have been contemplating is to up the
minimum number of VERIFIED votes in Gerrit to 2. With this scheme,
Gluster Build System would still provide a +1 after build and basic
sanity checks on the default configuration. Apart from that, it would
also be necessary for one more vote (from a human mostly :-)), which
confirms that the patch has been tested and is good to go. Please stay
tuned as we evolve this or an equivalent policy which ensures that the
situation you mention is covered.
Thanks,
Vijay
