Fix char and varchar typename printing in EXPLAIN and error msgs

Author: ggevay

src/adapter/src/catalog.rs

@@ -1672,6 +1672,14 @@ impl ExprHumanizer for ConnCatalog<'_> {
                     .join(",")
             ),
             PgLegacyChar => "\"char\"".into(),
+            Char { length } => match length {
+                None => "char".into(),
+                Some(length) => format!("char({})", length.into_u32()),
+            },
+            VarChar { max_length } => match max_length {
+                None => "varchar".into(),
+                Some(length) => format!("varchar({})", length.into_u32()),
+            },
             UInt16 => "uint2".into(),
             UInt32 => "uint4".into(),
             UInt64 => "uint8".into(),

src/repr/src/explain.rs

@@ -438,9 +438,11 @@ pub trait ExprHumanizer: fmt::Debug {
     fn humanize_id_parts(&self, id: GlobalId) -> Option<Vec<String>>;
 
     /// Returns a human-readable name for the specified scalar type.
+    /// Used in, e.g., EXPLAIN and error msgs.
     fn humanize_scalar_type(&self, ty: &ScalarType) -> String;
 
     /// Returns a human-readable name for the specified column type.
+    /// Used in, e.g., EXPLAIN and error msgs.
     fn humanize_column_type(&self, typ: &ColumnType) -> String {
         format!(
             "{}{}",

test/sqllogictest/array_fill.slt

@@ -195,10 +195,10 @@ SELECT array_fill(LIST[1], array[3, 2])
 query error array_fill on integer list not yet supported
 SELECT array_fill(LIST[1], array[3, 2], array[2, 3])
 
-query error array_fill on character not yet supported
+query error db error: ERROR: array_fill on char\(1\) not yet supported
 SELECT array_fill('c'::char, array[3, 2])
 
-query error array_fill on character not yet supported
+query error db error: ERROR: array_fill on char\(1\) not yet supported
 SELECT array_fill('c'::char, array[3, 2], array[2, 3])
 
 query error array_fill on map\[text=>integer\] not yet supported

test/sqllogictest/char.slt

@@ -23,7 +23,7 @@ SELECT 'a'::character = 'a'::"char";
 ----
 true
 
-query error coalesce could not convert type character to "char"
+query error db error: ERROR: coalesce could not convert type char\(1\) to "char"
 SELECT pg_typeof(coalesce('1'::"char", '1'::char));
 
 # Fixes database-issues#5191
@@ -90,14 +90,14 @@ SELECT pg_typeof('abc'::text::"char");
 
 # Fixes database-issues#5222
 
-query error db error: ERROR: coalesce could not convert type "char" to character
+query error db error: ERROR: coalesce could not convert type "char" to char
 SELECT COALESCE('a'::char, 'a'::"char");
 
-query error coalesce could not convert type "char" to character varying
+query error db error: ERROR: coalesce could not convert type "char" to varchar
 SELECT COALESCE('a'::varchar, 'a'::"char");
 
-query error coalesce could not convert type character to "char"
+query error db error: ERROR: coalesce could not convert type char\(1\) to "char"
 SELECT COALESCE('a'::"char", 'a'::char);
 
-query error db error: ERROR: coalesce could not convert type character varying to "char"
+query error db error: ERROR: coalesce could not convert type varchar to "char"
 SELECT COALESCE('a'::"char", 'a'::varchar);

test/sqllogictest/cockroach/aggregate.slt

@@ -698,22 +698,22 @@ SELECT avg(b), sum(b) FROM abc
 # ----
 # 3 years 5 mons 7 days 00:00:19
 
-query error db error: ERROR: function sum\(character varying\) does not exist
+query error db error: ERROR: function sum\(varchar\) does not exist
 SELECT avg(a) FROM abc
 
 query error db error: ERROR: function sum\(boolean\) does not exist
 SELECT avg(c) FROM abc
 
-query error db error: ERROR: function sum\(record\(f1: character varying,f2: boolean\?\)\) does not exist
+query error db error: ERROR: function sum\(record\(f1: varchar,f2: boolean\?\)\) does not exist
 SELECT avg((a,c)) FROM abc
 
-query error db error: ERROR: function sum\(character varying\) does not exist
+query error db error: ERROR: function sum\(varchar\) does not exist
 SELECT sum(a) FROM abc
 
 query error db error: ERROR: function sum\(boolean\) does not exist
 SELECT sum(c) FROM abc
 
-query error db error: ERROR: function sum\(record\(f1: character varying,f2: boolean\?\)\) does not exist
+query error db error: ERROR: function sum\(record\(f1: varchar,f2: boolean\?\)\) does not exist
 SELECT sum((a,c)) FROM abc
 
 statement ok

test/sqllogictest/map.slt

@@ -524,7 +524,7 @@ SELECT '{a=>1, b=>2}'::map[text=>int] <@ '{a=>1, b=>2, c=>3}'::map[text=>int]
 ----
 true
 
-query error CAST does not support casting from map\[text=>map\[text=>character\]\] to map\[text=>map\[text=>character\]\]
+query error db error: ERROR: CAST does not support casting from map\[text=>map\[text=>char\]\] to map\[text=>map\[text=>char\(1\)\]\]
 SELECT '{hello=>{world=>a}}'::map[text=>map[text=>char]] <@ '{hello=>c}'::map[text=>char]
 ----
 false

test/sqllogictest/type-promotion.slt

@@ -1662,3 +1662,117 @@ values ('2023-01-01T00:00:00.666'::timestamp(6)) union all values ('2023-01-01T0
 
 query error numeric field overflow
 SELECT '1e-307'::float8::numeric
+
+statement ok
+CREATE TABLE t1(a varchar(3));
+
+statement ok
+CREATE TABLE t2(b varchar(4));
+
+statement error db error: ERROR: value too long for type character varying\(3\)
+INSERT INTO t1 VALUES ('123456');
+
+# From the Postgres docs: https://www.postgresql.org/docs/current/datatype-character.html
+# "An attempt to store a longer string into a column of these types will result in an error, unless the excess
+# characters are all spaces, in which case the string will be truncated to the maximum length. (This somewhat bizarre
+# exception is required by the SQL standard.)"
+statement ok
+INSERT INTO t1 VALUES ('111   ');
+
+statement ok
+INSERT INTO t1 VALUES ('123');
+
+statement ok
+INSERT INTO t2 VALUES ('1234');
+
+query T multiline
+EXPLAIN WITH (TYPES)
+(SELECT * FROM t1) UNION ALL (SELECT * FROM t2);
+----
+Explained Query:
+  Union // { types: "(varchar?)" }
+    ReadStorage materialize.public.t1 // { types: "(varchar(3)?)" }
+    ReadStorage materialize.public.t2 // { types: "(varchar(4)?)" }
+
+Source materialize.public.t1
+Source materialize.public.t2
+
+Target cluster: quickstart
+
+EOF
+
+query T
+(SELECT * FROM t1) UNION ALL (SELECT * FROM t2);
+----
+111
+123
+1234
+
+query T
+(SELECT * FROM t2) UNION ALL (SELECT * FROM t1);
+----
+111
+123
+1234
+
+statement ok
+CREATE TABLE t3(b char(2));
+
+statement ok
+INSERT INTO t3 VALUES ('ab');
+
+query T multiline
+EXPLAIN WITH (TYPES)
+(SELECT * FROM t1) UNION ALL (SELECT * FROM t3);
+----
+Explained Query:
+  Union // { types: "(varchar?)" }
+    ReadStorage materialize.public.t1 // { types: "(varchar(3)?)" }
+    Project (#1) // { types: "(varchar?)" }
+      Map (text_to_varchar[len=None](#0)) // { types: "(char(2)?, varchar?)" }
+        ReadStorage materialize.public.t3 // { types: "(char(2)?)" }
+
+Source materialize.public.t1
+Source materialize.public.t3
+
+Target cluster: quickstart
+
+EOF
+
+query T
+(SELECT * FROM t1) UNION ALL (SELECT * FROM t3);
+----
+ab
+111
+123
+
+statement ok
+CREATE TABLE t4(b char(3));
+
+statement ok
+INSERT INTO t4 VALUES ('ab ');
+
+query T multiline
+EXPLAIN WITH (TYPES)
+(SELECT * FROM t3) UNION (SELECT * FROM t4);
+----
+Explained Query:
+  Distinct project=[#0] // { types: "(char?)" }
+    Union // { types: "(char?)" }
+      ReadStorage materialize.public.t3 // { types: "(char(2)?)" }
+      ReadStorage materialize.public.t4 // { types: "(char(3)?)" }
+
+Source materialize.public.t3
+Source materialize.public.t4
+
+Target cluster: quickstart
+
+EOF
+
+# Trailing spaces are treated as semantically insignificant in both Postgres and Materialize, so the above
+# 'ab' and 'ab ' end up deduplicated.
+# See also in char-varchar-distinct.td
+query T
+(SELECT * FROM t3) UNION (SELECT * FROM t4);
+----
+ab

test/testdrive/char-varchar-distinct.td

@@ -8,10 +8,11 @@
 # by the Apache License, Version 2.0.
 
 #
-# Make sure the distinct operator inside a detaflow operates correctly
+# Make sure the distinct operator inside a dataflow operates correctly
 # with respect to CHAR/VARCHAR, especially in the presence of trailing
 # spaces.
 #
+# The reason for this being a td and not slt is that td has "" wrapping on the output, which is convenient here.
 
 > SELECT 'a '::char(5) UNION DISTINCT SELECT 'a  '::char(5);
 "a    "