fuel-dev team mailing list archive

[Nikolay Markov <nmarkov@xxxxxxxxxxxx>]

Team, we were discussing moving to Oslo with some people from Ironic
project, and here is a number of statements we got:

1) Oslo itself is trying to get rid of eventlet-based logic, because this
approach is bad for moving to Python 3 and Oslo itself should not be nailed
to one concurrency model. That's why we can pass our scopped_session as a
session factory, as we do now, but this is kinda unresearched way for Oslo
right now. So hypothetically there may be some troubles. But there are no
serious issues preventing us from switching our code to it.
2) There is no serious need for moving to another web framework right now,
but if we want to catch up with OpenStack - it should be Pecan. But it's
really low-priority task for now.
3) Oslo supporters are really slow in patch acceptance. This means, if we
run into some trouble - we may need to use our own patched version of Oslo,
with blackjack and other stuff, before our patches will be accepted (it may
last for months).
4) Pecan is really fast in accepting patches, but though it's really small
and doesn't provide much functionality we need (like, pagination). However,
there are some existing patches (in Ironic itself, for example) which
provide this lacking functionality.
5) There is no really common way how to write APIs for OpenStack, but we
may follow any of the ways they already use - for example, copy this part
from Nova or Ironic and dance further.

-- 
Best regards,
Nick Markov