wrap in an array

bpblanken · bpblanken · commit 416b535f2f73 · 2024-06-13T19:26:03.000-04:00
diff --git a/v03_pipeline/lib/misc/family_entries.py b/v03_pipeline/lib/misc/family_entries.py
@@ -109,7 +109,7 @@ def remove_family_guids(
         # NB: this "should" work without the extra if statement (and does in the tests)
         # however, experiments on dataproc showed this statement hanging with an empty
         # indexes array.
-        family_entries=family_indexes_to_keep.map(lambda i: ht.family_entries[i])
+        family_entries=hl.array(family_indexes_to_keep).map(lambda i: ht.family_entries[i])
         if len(family_indexes_to_keep) > 0
         else hl.empty_array(ht.family_entries.dtype.element_type),
     )
diff --git a/v03_pipeline/lib/misc/lookup.py b/v03_pipeline/lib/misc/lookup.py
@@ -136,7 +136,7 @@ def remove_project(
     ht = ht.annotate(
         project_stats=(
             # See "remove_family_guids" func for why this was necessary
-            project_indexes_to_keep.map(lambda i: ht.project_stats[i])
+            hl.array(project_indexes_to_keep).map(lambda i: ht.project_stats[i])
             if len(project_indexes_to_keep) > 0
             else hl.empty_array(ht.project_stats.dtype.element_type)
         ),

Original file line number	Diff line number	Diff line change
`@@ -109,7 +109,7 @@ def remove_family_guids(`
`109`	`109`	`# NB: this "should" work without the extra if statement (and does in the tests)`
`110`	`110`	`# however, experiments on dataproc showed this statement hanging with an empty`
`111`	`111`	`# indexes array.`
`112`		`- family_entries=family_indexes_to_keep.map(lambda i: ht.family_entries[i])`
	`112`	`+ family_entries=hl.array(family_indexes_to_keep).map(lambda i: ht.family_entries[i])`
`113`	`113`	`if len(family_indexes_to_keep) > 0`
`114`	`114`	`else hl.empty_array(ht.family_entries.dtype.element_type),`
`115`	`115`	`)`