constraints call notes/proposals/sync

[William Reade]

On Thu, 2013-02-07 at 02:56 +0100, William Reade wrote:
> Consider the following tweaks to the python system:
> 
>   * instance-type values are not checked at parse time.
>   * instance-type values no longer conflict with cores/mem, and are
>     set and inherited independently.

I should note that an additional drawback came to light in a discussion
earlier today: that the rules as stated lead to crackful behaviour in
the following situation:

    $ juju set-constraints instance-type=m1.small
    $ juju set-constraints -s S mem=8G

...in which the more general instance-type would override the more
specific mem constraint on S and fail to accurately reflect user intent.
If the second point were to read:

  * instance-type values can be specified alongside cpu/cores/mem
    without complaint, but the two groups continue to interact as
    before when service constraints overlap with instance constraints.

...then that problem would no longer apply, but I could no longer claim
to have eliminated the concept of constraints. It seems to me that
according to the use cases I imagined, the impact of this drawback is
not enough to materially affect my conclusions; do I appear to have
missed anything here?

Cheers
William