[openstack-dev] [PKG-Openstack-devel][horizon][xstatic] XStatic-Angular-Bootstrap in violation of the MIT/Expat license (forwarded from: python-xstatic-angular-bootstrap_0.11.0.2-1_amd64.changes REJECTED)

Michael Krotscheck krotscheck at gmail.com
Tue May 5 15:05:20 UTC 2015


On Tue, May 5, 2015 at 1:32 AM Matthias Runge <mrunge at redhat.com> wrote:

> On 05/05/15 04:31, Ian Cordasco wrote:
>
> > Even so, Horizon is deployed in many places, and given the reliability of
> > system packages, it’s increasingly deployed from source.
>
> Ok, I'll bite.
>
> You surely have a source for your statement, or even better, a proof?
>

In the interest of open data, I'll answer this, using basic data extracted
from the Paris User Survey. For the sake of simplicity, I'm only going to
focus on production deployments of openstack, and I'm going to make the
assumption that if a tool is referenced, the official openstack version of
it was used. Here's the link, if you'd like to follow along:

http://superuser.openstack.org/articles/openstack-user-survey-insights-november-2014

As of Paris, the following tools were used to deploy openstack-dashboard in
production.  I've called out which of these are actually a source install,
and which of them are not. I've also skipped the two 1%'ers, because I
don't know those tools well enough to figure out if they're source.

ToolPercentageIs Source?Is Package?Puppet45%NoYesChef20%NoYesAnsible21%NoYes
DevStack7%Yes NoPackStack8%NoYesSalt8%NoYesJuju7%NoYes

As you can see, the majority of the tools that we publish install via
packages. Note that this data _cannot_ be used to infer an argument as to
whether source or packages are used more often, for the following reasons:

   1. The actual percentages from the survey add up to 118%.
   2. "Install from Source" did not appear to be an option.
   3. We are unable to determine the size of the cloud, thus providing a
   'weight' to each install method.
   4. We do not know whether every one of these respondents actually
   install horizon.

In short: We don't have data to support either side of this argument,
though there is a strong case that packages are the de-facto install method.

If I can editorialize for a second and read subtext into what Ian's saying:
The real question seems to be whether packagers have a disproportionate
amount of power to set development goals, tools, and policy. This is a
common theme that I've encountered frequently, and it leads to no small
amount of tension.

This tension serves no-one, and really just causes all of us stress. How
about we start a separate thread to discuss the roles of package
maintainers in OpenStack?

Michael
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150505/ad74f2be/attachment.html>


More information about the OpenStack-dev mailing list