[Pacemaker] Resource agents: parameter type enforcement and normalization

[Lars Marowsky-Bree]

On 2009-07-14T18:01:15, Florian Haas <florian.haas at linbit.com> wrote:

> That is true, and I am not contesting that they need to validate. I was
> actually asking for, what shall we call it, instance attribute value
> normalization.

Yeah, I agree. Though RAs can simply "enforce" this by only accepting
'true/false' or 'yes/no'. I think we made a mistake in trying to cope
with every possibly meaning, but hey, one lives and learns ;-)

> > "ocf_is_true"
> Funny. If you had used that one in your own drbd OCF RA, I would have
> caught that sooner. :)

I only added ocf_is_true in March when I was refactoring he LVM code. I
hadn't actually touched/normalized the other RAs since, my bad. Sorry
about that.

> > Yes, the meta-data needs more extensive functionality; at the risk of
> > being shot, I've contemplated borrowing the HTML "FORM" syntax.
> Why borrow that? IIUC "enum" is the supported syntax for cluster
> properties already, why introduce something new?

True.

HTML FORM - basically, a configuration page for the RA - though would
allow things like JavaScript validation/helptexts, the whole nine
yards. The downside would of course be validation in the CLI, so enum
likely is the better choice, but I wanted to throw this into the
discussion ;-)


Thanks,
    Lars

-- 
SuSE Labs, OPS Engineering, Novell, Inc.
SUSE LINUX Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)
"Experience is the name everyone gives to their mistakes." -- Oscar Wilde