[Gzz-commits] (no subject)

[Benja Fallenstein <address@hidden>]

Update of /cvsroot/gzz/gzz/test/gzz/impl
In directory subversions:/tmp/cvs-serv2158/test/gzz/impl

Modified Files:
        TestSimpleDim.java 
Added Files:
        singleslicer.test 
Removed Files:
        sliceversion.test 
Log Message:

Ok, here's the big commit:

1. New Slicer interface. A Slicer is a component of Space on the same
   level as CellTexter. It's responsibility is to divide a space into
   slices, allowing slices to be swapped in and out. This means that
   turning slices into SliceVersions is part of its responsibilities
   (as attaching and disattaching slices (resolving and recreating
   preflets) is encapsulated in it).

   So far, this is only able to get the SliceVersion for s.0 (it is
   unfinished).

2. Update Space, Gzz.py etc. to use Slicer and the first implementation,
   gzz.impl.SingleSlicer (manages a single slice).

3. In CellTexter, add a Set getCells() method to get the cells that have
   a content mapping. (The problem is that we need some cooperation from
   CellTexter for saving, because there might be an infinite number of
   cells... maybe we should be able to get and set a Map from cells to
   implementation-defined serializable objects, but that would be less
   general. OTOH, SingleSlicer currently depends on CellTexter being
   a VStreamCellTexter: not nice...)

4. In Dim, remove addRealNegSides(Set). Instead, put in Set getNegsides().
   (I discussed this with Tuomas on IRC at some point.) The set contains
   cells, while addRealNegSides() put in cell ids-- now, we can simply
   return the keySet() of the Map in the positive direction.

5. The javadoc of Space.getDims() reads, "Get the set of all dimensions
   in this space." I understand this as meaning the set of Dim objects.
   AbstractDimManager, though, returned Cells. I've changed
   AbstractDimManager for now and put a comment in Space's javadoc.
   Tuomas, please decide this and clarify the javadoc in Space.

6. Remove parts of javadoc in VStreamCellTexter that tell about how to
   handle the Lists handling by methods that use Enfilade1Ds now.

7. Remove AbstractDim.getHeadcells(). This returned null and had a comment,
   "XXX REMOVE THIS ONCE THE NEW API HAS SETTLED." Now that we're rid of
   the many basalt/ dims I didn't want to change when I put that comment
   in, we should also be rid of the bad default "implementation."

8. Added a second constructor to ModularSpace, taking a SliceVersion from
   which to initialize the space. I'm not sure where this should be.
   I do feel that once constructed, a space should have one "starter"
   slice (s.0)...

9. Removed field obsTrigger and method getObsTrigger() from ModularSpace.
   AbstractSpace handles them already.

10. Add test/tools.py with tools needed by multiple tests.
    Fairly unorganized so far.

11. Move spaceArgList() and deltaspaceArgList() from test/test.py
    to test/tools.py.

12. Move SliceVersion from gzz.impl to gzz, as it is referenced by a
    core interface (Slicer) now. I think this helps interoperability.

gzz/test/gzz/impl singleslicer.test,NONE,1.1 TestSimpleDim.java,1.8,1.9 
sliceversion.test,1.2,NONE
Mon Jul 29 06:16:55 EDT 2002