Open Stack

Hi Monty,

Thank you for your reply.

2017年3月28日(火) 23:15 Monty Taylor <mordred at inaugust.com>:

On 03/26/2017 02:55 AM, Akira Yoshiyama wrote:
> Hi,
>
> 2017-03-25 4:54 GMT+09:00 Monty Taylor <mordred at inaugust.com>:
>> On 03/24/2017 10:34 AM, Akira Yoshiyama wrote:
>>> Hi Monty,
>>>
>>> Thank you for your reply.
>>>
>>> 2017年3月24日(金) 20:58 Monty Taylor <mordred at inaugust.com
>>> <mailto:mordred at inaugust.com>>:
>>>
>>>     On 03/19/2017 07:18 AM, Akira Yoshiyama wrote:
>>>     > Hi all,
>>>     >
>>>     > I have developed Yakumo, a pythonic unified OpenStack client
library:
>>>     >
>>>     >   PyPI: https://pypi.python.org/pypi/yakumo
>>>     >   Git: https://github.com/yosshy/python-yakumo
>>>
>>>     Nice library!
>>>
>>>
>>> Thank you :)
>>>
>>> We use it for our smoke tests.
>>>
>>>     > and I found that
>>>     > os_client_config.cloud_config.CloudConfig.get_session_endpoint()
>>>     > didn't return Keystone admin endpoint because of below:
>>>     >
>>>     >
>>>
https://github.com/openstack/os-client-config/blob/master/os_client_config/cloud_config.py#L258
>>>     >
>>>     > Why is it so?
>>>
>>>     It's done that way in os-client-config because not doing it that way
>>>     breaks python-keystoneclient. Or at least it used to - looking
through
>>>     the keystoneclient code it seems jamie has fixed this now.
>>>
>>>
>>> Wonderful!
>>>
>>>     I'm going to nudge Morgan or Jamie to respond too - I think we
might be
>>>     able to get rid of this conditional (which would make me super
happy)
>>>
>>>
>>> How about an extra option like below?
>>>
>>> def get_session_endpoint(
>>>     ....., allow_identity_admin=False):
>>
>> I think that's a great idea - you feel like proposing it?
>
> Yes. I'll push a patch for review soon.

Whoops - this is totally my bad. Since we spoke, Jamie Lennox and I
looked at the code and believe that this
https://review.openstack.org/#/c/450259/ is what the code should be
doing. I totally forgot to come back and follow up to this thread. So
your patch is great and exactly what we talked about - but I think we
can just go ahead and fix the behavior and not need a new option.

What do you think? (also, again, sorry for forgetting to follow up)


450259 looks very good; simple and ideal. But I'm afraid of backward
compatibility. We may have to add a version condition to os-client-config
in requirements of old versions of python modules, e.g. python-keystoneauth.

Again, 450259 is ideal. It's ok to abandon 450500.

Regards,
Akira



__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

-- 
吉山あきら <akirayoshiyama at gmail.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20170328/b1c537bd/attachment.html>