Re: [Condor-devel] Gridmanager without Globus

[Erik Paulson <epaulson@xxxxxxxxxxx>]

On Mon, Feb 25, 2008 at 05:05:10PM -0600, Jaime Frey wrote:
> On Feb 24, 2008, at 9:53 AM, Matthew Farrellee wrote:
> 
> >Currently, the condor_gridmanager is only compiled if  
> >HAVE_EXT_GLOBUS is defined, which only happens when Globus is  
> >available.
> >
> >I think this is no longer correct. The gridmanager is used for  
> >Condor-C and ongoing EC2 work, which have nothing to do with Globus.
> >
> >Attached is a patch to V7.0, but probably applies to V7.1, that  
> >always compiles the condor_gridmanager and selectively leaves out  
> >functionality based on available externals. For instance, GT4 code  
> >is left out if HAVE_EXT_GT4GAHP is not defined.
> >
> >The patch compiled on all platforms, and appears to test fine as  
> >well. On my system, x86 F8, the job_condorc_ab_van and  
> >job_condor_abc_van tests pass when Globus is not available.
> 
> 
> The patch looks good to me. I don't see a reason not to commit it.
> 

The gridmanager continues to build with the GT4 code enabled, even if
the GT4 code is not available, right? ie all the gridmanager does is
respond to ASCII responses from the gahp. If anything, it should be
like the startd/schedd, which look to see which starter/shadows are
available at runtime and then decide what they'll support.

Why not just leave it on all the time? Of all the code bloat to worry
about, this seems like the least of our worries.

-Erik