linking to benchmarks

When creating a new space in which I want ot run a job over benchmarks, it appears that the only I option I have is to copy the whole benchmark library. What I would really like is to **link** to the benchmark library, so as the library is enlarged I can recreate and rerun jobs that will use all the new benchmarks, without having to copy them again.

StarExec's web interface is misleading about what it is doing on "copying the hierarchy" operations. It already does what you want. It links the benchmarks, and generates new spaces that reflect the previous space being linked.  This is a bit challenging to see without doing an experiment. I'm going to assume you have access to the space /root/SMT/SMT-LIB benchmarks/. Try out the following steps to confirm this:

1. Explore to the space QF_LIA (space 2004)
   
2. Expand subspaces and select RTCL (space 2076). This contains 2 subspaces each of which contain 1 benchmark.  Here is the layout:
   

   Code:

   RTCL (space 2076)
      b04_tf_15 (space 2077)
        ckt_PROP0_tf_15.smt2 (benchmark 406184)
      b04_tf_20 (space 2078)
        ckt_PROP0_tf_20.smt2 (benchmark 406185)


   
   
3. Drag RTCL into your personal account
   
4. You are now prompted with "confirm copy: do you want to copy the RTCL only or the hierarchy to <username>? [space] [hierarchy] [cancel]"
   
5. Click [hierarchy]
   
6. Now open RTCL in your personal folder.  Here is the space layout I see:
   

   Code:

   RTCL (space 17003)
      b04_tf_15 (space 17004)
        ckt_PROP0_tf_15.smt2 (benchmark 406184)
      b04_tf_20 (space 17005)
        ckt_PROP0_tf_20.smt2 (benchmark 406185)


   
   

Notice that all of the space ids have all changed, but the benchmark ids have not.  So the benchmarks were linked, but the spaces were "copied".  Solvers within spaces and jobs also copied in this manner.*


* Not sure it makes sense to copy jobs, but right now it happens.

But this is not quite what I meant to ask about.
I want to link to the root (or subroot) of a hierarchy of benchmarks within a source subspace, so that any changes to that original subspace will also be reflected in my subspace that has the linked root.
What I understand you to say is that when a 'copy' a hierarchy. all the benchmarks are actually links to their originals, so the contents is not replicated. At least then if the benchmark itself is updated, I would see the new benchmark. I'm asking for the same behavior higher up the hierarchy - i.e., the space ids would not change.