fix[dace][next]: Fixes expanding of Not Supported Memlets

src/gt4py/next/program_processors/runners/dace/transformations/gpu_utils.py

@@ -176,7 +176,8 @@ def restrict_fusion_to_newly_created_maps(
     ) -> bool:
         return any(new_entry in new_maps for new_entry in [map_entry_1, map_entry_2])
 
-    # Using the callback to restrict the fusing
+    # Now try to fuse the maps together, but restrict them that at least one map
+    #  needs to be new.
     sdfg.apply_transformations_repeated(
         [
             gtx_transformations.MapFusionSerial(
@@ -192,8 +193,20 @@ def restrict_fusion_to_newly_created_maps(
         validate_all=validate_all,
     )
 
-    # Now we have to find the maps that were not fused. We rely here on the fact
-    #  that at least one of the map that is involved in fusing still exists.
+    # This is a gross hack, but it is needed, for the following reasons:
+    #  - The transients have C order while the non-transients have (most
+    #       likely) FORTRAN order. So there is not an unique stride dimension.
+    #  - The newly created maps have names that does not reflect GT4Py dimensions,
+    #       thus we can not use `gt_set_iteration_order()`.
+    #  For these reasons we do the simplest thing, which is assuming that the maps
+    #  are created in C order and we must make them in FORTRAN order, which means
+    #  just swapping the order of the map parameters.
+    #  We further assume here, that we only have to process the maps that we have
+    #  newly created.
+    # NOTE: We can stop relying on this once [PR#1913](https://github.com/GridTools/gt4py/pull/1913)
+    #   Has been merged, which is currently blocked by a DaCe PR that has not been
+    #   merged.
+
     maps_to_modify: set[dace_nodes.MapEntry] = set()
     for nsdfg in sdfg.all_sdfgs_recursive():
         for state in nsdfg.states():
@@ -202,17 +215,13 @@ def restrict_fusion_to_newly_created_maps(
                     continue
                 if map_entry in new_maps:
                     maps_to_modify.add(map_entry)
-    assert 0 < len(maps_to_modify) <= len(new_maps)
 
-    # This is a gross hack, but it is needed, for the following reasons:
-    #  - The transients have C order while the non-transients have (most
-    #       likely) FORTRAN order. So there is not an unique stride dimension.
-    #  - The newly created maps have names that does not reflect GT4Py dimensions,
-    #       thus we can not use `gt_set_iteration_order()`.
-    #  For these reasons we do the simplest thing, which is assuming that the maps
-    #  are created in C order and we must make them in FORTRAN order, which means
-    #  just swapping the order of the map parameters.
-    # TODO(phimuell): Do it properly.
+    # We did not found any of the newly created Map. Thus we **hope** that all new
+    #  Maps have been integrated into other Maps, that have the correct names.
+    #  But as written above, this is a gross hack!
+    if len(maps_to_modify) == 0:
+        return sdfg
+
     for me_to_modify in maps_to_modify:
         map_to_modify: dace_nodes.Map = me_to_modify.map
         map_to_modify.params = list(reversed(map_to_modify.params))