Open Stack

On Wed, 26 Aug 2015, Renat Akhmerov wrote:

> What would be your recommendation for now? Just to cap wsme version
> and hold on with changes adjusting to WSME 0.8.0? Or you think most
> likely these changes in new WSME will remain on?

The fixes in WSME are fixes for genuine bugs. If code using WSME was
relying on that behavior it is probably best that it be fixed.

For example if you have code that is silently accepting invalid
fields in a POST, you probably don't want that.

The other changes are primarily related to the error messages that
are generated when incoming data (usually via a POST) is determined
to be invalid.

If you're unable to change the code calling WSME to reflect these
changes then pinning to <0.8.0 is a potential solution.

In ceilometer and aodh it took <30 minutes to find and resolve the
issues related to the changes.

If you need some help with that you can find me (or lucas) in the
#wsme channel and we can help you figure out what changes might be
needed to fix any failing tests.

Dealing with this is a pain, but I think it is worthwhile: Those
services that are using WSME (for now) need to be able to rely on it
doing what it says it is supposed to be doing, not accidental
behaviors that are present as a result of bugs.

-- 
Chris Dent tw:@anticdent freenode:cdent
https://tank.peermore.com/tanks/cdent