[openstack-dev] upgrading WebOb across all projects

Monty Taylor mordred at inaugust.com
Wed Jan 16 21:42:19 UTC 2013


On 01/16/2013 12:52 PM, Doug Hellmann wrote:
> The last project accepted a patch to use WebOb>=1.2 today, so we are all
> now up to date.
> 
> A big thanks to everyone who helped push these changes through,
> including Monty, Dan Prince, Chuck Short, Sasche Peilicke, Eoghan Glynn,
> Brian Waldon,  Zhongyue Luo, Sean Deague, Michael Still, Joe Heck, Dolph
> Matthews, John Griffith, Huang Zhiteng, Dan Wendlandt, and Gary Kotton.
> As you can see, this migration took a lot of collaboration, so if I've
> left your name out please forgive me.

Thanks everybody! Getting changes like that across projects is ... fun
... to say the least.

Also - my narcissism is not allowing me to not point out that I'm the
only person in that list who has become a full-on single-name person. I
see this as another step towards world domination. Maybe next I'll get
to star in Evita...

> On Thu, Jan 3, 2013 at 11:32 AM, Doug Hellmann
> <doug.hellmann at dreamhost.com <mailto:doug..hellmann at dreamhost.com>> wrote:
> 
>     The ceilometer team is using Pecan [1] for our next-generation API
>     framework. Pecan depends on a newer WebOb than is used by the other
>     OpenStack components, so I spent this morning reviewing the
>     differences in WebOb between 1.0.8 and 1.2.3 (the latest stable
>     release) [2] and what impact those changes might have on nova,
>     cinder, keystone, and oslo (the 4 projects where I found WebOb being
>     used -- please let me know if I missed one).
> 
>     Most of the changes appear to be bug fixes and changes that either
>     have a backwards-compatible alias, are deprecations that were
>     reversed later, or should not have an impact on us because we were
>     not using the changed area of the code directly. The only thing I
>     found to be questionable is the change to Request.path_info and
>     Request.script_name. Those properties are now unicode objects
>     instead of str objects, as WebOb is trying to provide better Python
>     3 support. I found a few uses of those variables in comparisons and
>     concatenation with str objects, but the casting should work
>     automatically as long as the path can be encoded in the default
>     encoding, which should be the case as the paths to our API use only
>     ASCII characters. All of the unit tests pass when I run them locally.
> 
>     I have submitted patches against the 4 projects to update the
>     version number in tools/pip-requires, without any other code changes:
> 
>     Oslo: https://review.openstack.org/#/c/18885/
>     Nova: https://review.openstack.org/#/c/18886/
>     Keystone: https://review.openstack.org/#/c/18887/
>     Cinder: https://review.openstack.org/#/c/18888/
> 
>     If anyone else has done similar analysis in the past and come to a
>     different conclusion about whether it is safe to upgrade, please let
>     me know..
> 
>     Doug
> 
>     [1] Pecan: http://pecan.readthedocs.org/en/latest/
>     [2] The full release notes for WebOb:
>     http://docs.webob.org/en/latest/news.html
> 
> 
> 
> 
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
> 



More information about the OpenStack-dev mailing list