Wrap brittle operations in subqueries

distribution/lib/Standard/Database/0.0.0-dev/src/Data/Table.enso

@@ -813,7 +813,7 @@ type Table
             new_columns = partitioned.first
             problems = partitioned.second
             on_problems.attach_problems_before problems <|
-                self.updated_context_and_columns new_ctx new_columns
+                self.updated_context_and_columns new_ctx new_columns subquery=True
 
     ## Returns a new table with a chosen subset of columns left unchanged and
        the other columns pivoted to rows with a single name field and a single
@@ -1020,8 +1020,24 @@ type Table
        Arguments:
        - ctx: The new context for this table.
        - internal_columns: The new columns to include in the table.
+       - subquery: A boolean indicating whether the operation should be wrapped
+         in a subquery. This is a simple workaround for operations which may be
+         affected by further operations if not wrapped. For example, a group-by
+         may need to be wrapped in this way if a filter is to be performed on it
+         later on. Ideally, this should be done only on demand, if the
+         subsequent operation needs it and operations like join should try to
+         avoid nesting subqueries without necessity. However, for now, for
+         simplicity, we are always wrapping brittle operations. This may be
+         revised in the future, to generate better and more concise SQL code.
     updated_context_and_columns : Context -> Vector Internal_Column -> Table
-    updated_context_and_columns self ctx internal_columns = Table.Value self.name self.connection internal_columns ctx
+    updated_context_and_columns self ctx internal_columns subquery=False = case subquery of
+        True ->
+            setup = ctx.as_subquery self.name [internal_columns]
+            new_ctx = Context.for_subquery setup.subquery
+            new_columns = setup.new_columns.first
+            Table.Value self.name self.connection new_columns new_ctx
+        False ->
+            Table.Value self.name self.connection internal_columns ctx
 
     ## PRIVATE
 

distribution/lib/Standard/Database/0.0.0-dev/src/Internal/Postgres/Postgres_Dialect.enso

@@ -90,7 +90,7 @@ type Postgres_Dialect
         distinct_expressions = new_key_columns.map (Database_Distinct_Helper.make_distinct_expression text_case_insensitive problem_builder)
         # TODO check order_by
         new_context = Context.for_subquery setup.subquery . set_distinct_on distinct_expressions
-        table.updated_context_and_columns new_context new_columns
+        table.updated_context_and_columns new_context new_columns subquery=True
 
 ## PRIVATE
 make_internal_generator_dialect =

distribution/lib/Standard/Database/0.0.0-dev/src/Internal/SQLite/SQLite_Dialect.enso

@@ -108,7 +108,7 @@ type SQLite_Dialect
         new_key_columns = key_columns.map c-> column_mapping.at c.name
         distinct_expressions = new_key_columns.map (Database_Distinct_Helper.make_distinct_expression text_case_insensitive problem_builder)
         new_context = Context.for_subquery setup.subquery . set_groups distinct_expressions
-        table.updated_context_and_columns new_context new_columns
+        table.updated_context_and_columns new_context new_columns subquery=True
 
 
 ## PRIVATE