Re: [Condor-devel] RFC: an implementation of classads enhanced with macro evaluation

[Erik Paulson <epaulson@xxxxxxxxxxx>]

On Tue, Mar 13, 2012 at 02:40:54PM -0700, Erik Erlandson wrote:
> On Tue, 2012-03-13 at 15:45 -0500, Alan De Smet wrote:
> > 
> > I like the idea of ClassAds as configuration.  However, there has
> > been resistance to major changes to the configuration system,
> > resistance that lead to the most recent overhaul being far scaled
> > down from the original design.  A complete replacement with a
> > ClassAd based system would be an even bigger step.  
> 
<...>

This idea of Classads as the ideal configuration gets asserted from 
time to time, but there really isn't any evidence that it's true. 

There's a frustration with shortcomings in the configuration language, 
and true, much of what's in the configuration language are ultimately
incorporated as expressions in classads. 

At the same time, other parts of the internal state of the Condor daemons
is _not_ managed via any sort of ClassAd data structure. The configuration
process relies on a known-order and precdence in intrepreting the files,
and some thought was given with how it would interact and behave like
other systems that most Condor administrators manage at the same time
as Condor and follow some of the conventions from those communities.

If Condor went to using a serialized ClassAd as its configuration system, 
it would be the only thing in the universe that does so. Is that really a
win for the pool administrators?

It's a fine idea to revisit the configuration file situation but it's not
a good idea to start with a predefined outcome. Perhaps Condor would be
better off by thinking about ways to make the ClassAd expressions easier
to store in a different semi-structured format.

<...>

On Tue, Mar 13, 2012 at 02:40:54PM -0700, Erik Erlandson wrote:
> 
> I admit to having some geeky plans for writing a spec-compliant
> implementation of classads for C++, in my copious spare time.
> 

To each their own, of course, but isn't the current ClassAd implemention
both in C++ and already reasonably close to the spec?

-Erik