Re: [Condor-users] inspiration for DAGMan syntax

["R. Kent Wenger" <wenger@xxxxxxxxxxx>]

On Thu, 1 Apr 2010, Ian Stokes-Rees wrote:

> ...
> I could reduce my DAG file by 20k lines (well, 20k - 2) if it were
> possible to do:
>
> RETRY       * 7
> SCRIPT POST * ./process_output $JOB $JOBID
>
> and DAGMan would notice that the node name contained a regular
> expression and then apply the statement to all DAG nodes it matched.
> There would, of course, be some subtle implementation details to work
> out (ordering/precedence of statements, interaction with other RE or
> non-RE statements, etc.), but I could imagine that something basic could
> be achieved without a huge amount of work.

Yes, this sounds like a good idea.  I've created a ticket for it in our
gittrac wiki.  We had thought about doing something similar in relation to 
splices, but there's no reason this shouldn't apply to DAG nodes in 
general.

Kent Wenger
Condor Team