cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From sebgoa <run...@gmail.com>
Subject Re: External fork of Cloudstack (was Re: [GitHub] cloudstack pull request: Is the project attempting a fork on Githu...)
Date Sat, 19 Mar 2016 18:12:27 GMT

> On Mar 19, 2016, at 3:59 PM, Sam Ruby <rubys@intertwingly.net> wrote:
> 
> The initial action (creating the fork) was not clearly done by the PMC
> as a whole, that is now being rectified by a retroactive vote being
> taken.  Even that I wouldn't suggest be done any differently: many
> parts of the ASF prefer a Commit then Review (CTR).  Under normal
> circumstances, a vote would not be necessary.
> 

To be transparent and avoid any hint at back door conversations (which are sadly too often
alluded too). Let me highlight a few historical steps:

- We switched to GitHub PR almost a year ago and stopped using RB.
- Through fall Remi and co worked on a new commit workflow to speed releases, their productivity
was “damaged” by lack of github access, namely: -no labels, no ability to close PR, no
access to issues, no access to triggers/hooks.
- Remi reached out to VP infra (David) couple times I believe and a discussion in Dublin (~Oct).
-Through our PMC project report (written by me as VP and LGTM’d by the PMC), I made several
mention of our intent to use Github (and docker Hub) more efficiently. I even said that if
we where getting early access to the “Whimsy experiment” we would quickly VOTE in our
community to decide on such a move.
- I never got any replies or comments (even though I mentioned that lack of github access
was putting our community at risk of fragmentation).
- Wido reached out to Github back in december when he saw there was already a cloudstack org
( not sure who created it and controlled it actually.)
- Schuberg forked and created cosmic in feb sometime.
- Finally last week, I took unilateral decision to click ‘fork’ on apache/cloudstack and
put a copy in cloudstack/cloudstack , thinking this would kick the tires and would give us
a playground to start experimenting with CI taking advantage of full control of github settings.

As a member of ASF I had witnessed several threads around the use of GitHub and there are
clearly strong feelings on the issue. I tried to participate in those threads and even demoed
with another member that we could clearly guarantee provenance by signing all commits with
ASF issued GPG keys, but that was not even acknowledged. 

Moving to Github (in some fashion) is indeed problematic to the ASF and a fully independent
open source project. We end up depending on a vendor for hosting our repo and this puts us
at risk if Github were to go bonkers. We can argue about such a risk, and we can also raise
examples of other vendor/proprietary software being used/depended on by ASF projects (JIRA,
Slack…). It also “moves” the community. While we are on the dev@ and other ML, the issue
that we may move the community to Github and give the impression that project is not at ASF
project is real. But to be honest there is already such confusion in OSS with projects on
ASF v2 license being perceived by most as ASF projects.

Personally I have always thought that this is a very serious issue and trend in open source
projects and that ASF (and the board in particular) should try to proactively address. What
is the future of ASF in a GitHub world ? Can an ASF project live outside of ASF infra, especially
in a Cloud world ? Sadly I never saw any clear proactivity from the board.

Hence yes, as my last action as VP I clicked on “fork”, and now I hope we will move forward,
and get the board to proactively do something.

-Sebastien







Mime
View raw message