API change: unit and machine nodes

Gustavo Niemeyer gustavo.niemeyer at canonical.com
Mon Dec 26 13:17:25 UTC 2011

>> It feels strange to be storing exactly the same type of information in
>> different formats depending on where we're looking at. I suggest
>> storing the unaltered strings in the former format across the board.
> My point is that it's not *exactly* the same type of information; there
> are certainly similarities, but there's a clear distinction between
> "partial input" and "complete output".

Both take their constraint names and values from a common space, are
stored in zookeeper for later retrieval, and are used to restrict the
deployment made. "I need to do more with that information" vs. "I need
to do less with that information" isn't reasoning for persisting
information that is exactly alike differently.

> Partial input, I think we're agreed, works great as a list of strings.
> Complete output, I think we also agree, works great as a dict (for easy
> consumption by the provider). From my perspective, your suggestion is to
> add an intermediate format for the complete output data (ie, two string

That's not what I suggested. I'm simply advocating that the format for
complete output be the same as for partial output, because they are
exactly alike as far as the _data_ goes.

Gustavo Niemeyer

-- I'm not absolutely sure of anything.

More information about the Juju mailing list