Open Stack

On Wed, 26 Aug 2015, Dmitry Tantsur wrote:

> Note that this is an API breaking change, which can potentially break random 
> users of all projects using wsme. I think we should communicate this point a 
> bit louder, and I also believe it should have warranted a major version bump.

Yeah, Lucas and I weren't actually aware of the specific impact of that
change until after it was released; part of the danger of being cores-on-
demand rather than cores-by-desire[1].

I'll speak with him and dhellman later this morning and figure out
the best thing to do.

And post the outcome back here.

[1] After having done quite a few reviews and fixing a few bugs in WSME
for a few months my advice to any one using it is to make a plan to
stop. It is a very difficult design that is extremely hard to
maintain. On top of that it was written from the standpoint of
satisfying some use cases (mostly related to typed input and output
handling and validation) by doing the HTTP handling that got the
desired results _not_ correct HTTP. It is riddled with incorrect
handling of headers and response codes that are very hard to fix
without causing unplanned side effects.

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