On Thu, 30 Apr 2015 17:37:51 -0500
Adam Miller <maxamillion(a)fedoraproject.org> wrote:
<snip>
> I started getting playbooks written to deploy phabricator in
infra
> but as with de-bundling and migrating off of
bitbucket.org, got
> sidetracked since it hasn't been a huge priority.
Is there any scope of how much work that would be? As I understand it,
that would be a show stopper for use/inclusion in the Fedora Rel-Eng
toolchain.
I've made notes on the deps which appear problematic and put that in
the git repo which has the spec files [1].
The biggest problem that I've seen thus far is that not all the bundled
deps are vanilla upstream - some are modified locally and those
modifications were not submitted upstream. It may not be a simple issue
of adding deps in the specfile and linking in existing packages for the
deps which already exist in the fedora repos.
[1]
http://ur1.ca/kaka4
<snip>
From the looks of this, phabricator could potentially provide a
solution for the Kanban workflow with it's "Tasks" feature. It would
be a little different than the cards workflow but it might be
workable.
Yeah, the workboards aren't perfect but they can work if projects are
self-contained. I've yet to figure out how to get tasks from multiple
projects on the same workboard, though.
My only main reservation is that I don't know what the team's
general
PHP expertise is and the only reason that is of concern to me is that
in the event we experience a bug, what's the general consensus that we
can we fix and submit upstream? (I have zero PHP background but I'm
not against learning, just thought I'd bring that up because this
concern has been considered a blocker for other things before).
Our experience has been that upstream has been rather responsive to
issues discovered.
Their patch policy requires signing a contributor agreement, so that's
one hurdle but we've yet to have an upstream-acceptable patch and I've
yet to get around to working that particular issue.
Tim