Ah, I see. I think I can understand why that was done initially, but since
using UPDATE_HORIZON_CONFIG is already going beyond the scope of the enabled
file since its directly affecting an exposed setting, I think we should just
use the setting (and document it appropriately in the settings.rst)
Rob
On 2 August 2016 at 23:39, Richard Jones
> wrote:
On 3 August 2016 at 00:32, Rob Cresswell
> wrote:
Hi all,
So we seem to be adopting a pattern of using UPDATE_HORIZON_CONFIG in the
enabled files to add a legacy/angular toggle to the settings. I don't like
this, because in settings.py the enabled files are processed *after*
local_settings.py imports, meaning the angular panel will always be enabled,
and would require a local/enabled file change to disable it.
My suggestion would be:
- Remove current UPDATE_HORIZON_CONFIG change in the swift panel and images
panel patch
- Add equivalents ('angular') to the settings.py HORIZON_CONFIG dict, and then
the 'legacy' version to the test settings.
I think that should run UTs as expected, and allow the legacy/angular panel to
be toggled via local_settings.
Was there a reason we chose to use UPDATE_HORIZON_CONFIG, rather than just
updating the dict in settings.py? I couldn't recall a reason, and the original
patch ( https://review.openstack.org/#/c/293168/ ) doesn't seem to indicate why.
It was an attempt to keep the change more self-contained, and since
UPDATE_HORIZON_CONFIG existed, it seemed reasonable to use it. It meant that
all the configuration regarding the visibility of the panel was in one place,
and since it's expected that deployers edit enabled files, I guess your concern
stated above didn't come into it.
I'm ambivalent about the change you propose, would be OK going either way :-)
Richard
__
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev