Обсуждение: psql's \d versus included-index-column feature

I noticed that psql's \d command doesn't do very well with included
index columns.  Given the regression db's test case,

CREATE INDEX tbl_include_reg_idx ON tbl_include_reg (c1, c2) INCLUDE (c3, c4);

we get

regression=# \d tbl_include_reg_idx
Index "public.tbl_include_reg_idx"
 Column |  Type   | Definition 
--------+---------+------------
 c1     | integer | c1
 c2     | integer | c2
 c3     | integer | 
 c4     | box     | 
btree, for table "public.tbl_include_reg"

So there are two problems with that: first, where's the definition
info for the included columns, and second, how's the user supposed
to tell which columns are key columns vs which are included?

(No, I don't accept the answer that you're supposed to know that
an omitted definition means an included column.  For one thing,
that approach cannot scale to handle included expression columns.
For another, the documentation doesn't say any such thing.)

I looked into the reason why the definition is empty, and the answer
seems to be somebody's poorly-thought-through decision that
pg_get_indexdef_worker's attrsOnly flag could be made to serve two
purposes, with the result that pg_get_indexdef_ext with a nonzero column
argument will print nothing for an included column.  That doesn't seem
like a sane (or documented) behavior either.  So the attached patch
splits it into two flags, attrsOnly and keysOnly, and with that we get

regression=# \d tbl_include_reg_idx
Index "public.tbl_include_reg_idx"
 Column |  Type   | Definition 
--------+---------+------------
 c1     | integer | c1
 c2     | integer | c2
 c3     | integer | c3
 c4     | box     | c4
btree, for table "public.tbl_include_reg"

which is better, but it's still not very clear what's what.
Do we want to consider that good enough, or do we want to
mark included columns in this display, and if so how?

One idea is to do it like this:

regression=# \d tbl_include_reg_idx
Index "public.tbl_include_reg_idx"
 Column |  Type   | Definition 
--------+---------+------------
 c1     | integer | c1
 c2     | integer | c2
 c3     | integer | INCLUDE c3
 c4     | box     | INCLUDE c4
btree, for table "public.tbl_include_reg"

If we wanted to have pg_get_indexdef_ext() include "INCLUDE" in its output
for an included column, this could be achieved with just a couple more
lines of code.  But that behavior seems rather ugly to me, and likely to
break other use-cases for pg_get_indexdef_ext().  So I think we likely
want to do this on the psql side instead, which will take more code but
also offers more flexibility for the display layout.  For instance, maybe
we want something like

regression=# \d tbl_include_reg_idx
Index "public.tbl_include_reg_idx"
 Column |  Type   | Key | Definition 
--------+---------+------------------
 c1     | integer | t   | c1
 c2     | integer | t   | c2
 c3     | integer | f   | c3
 c4     | box     | f   | c4
btree, for table "public.tbl_include_reg"

Thoughts?

            regards, tom lane

diff --git a/src/backend/utils/adt/ruleutils.c b/src/backend/utils/adt/ruleutils.c
index 065238b..a38aed2 100644
--- a/src/backend/utils/adt/ruleutils.c
+++ b/src/backend/utils/adt/ruleutils.c
@@ -320,7 +320,8 @@ static void decompile_column_index_array(Datum column_index_array, Oid relId,
 static char *pg_get_ruledef_worker(Oid ruleoid, int prettyFlags);
 static char *pg_get_indexdef_worker(Oid indexrelid, int colno,
                        const Oid *excludeOps,
-                       bool attrsOnly, bool showTblSpc, bool inherits,
+                       bool attrsOnly, bool keysOnly,
+                       bool showTblSpc, bool inherits,
                        int prettyFlags, bool missing_ok);
 static char *pg_get_statisticsobj_worker(Oid statextid, bool missing_ok);
 static char *pg_get_partkeydef_worker(Oid relid, int prettyFlags,
@@ -1097,7 +1098,9 @@ pg_get_indexdef(PG_FUNCTION_ARGS)

     prettyFlags = PRETTYFLAG_INDENT;

-    res = pg_get_indexdef_worker(indexrelid, 0, NULL, false, false, false,
+    res = pg_get_indexdef_worker(indexrelid, 0, NULL,
+                                 false, false,
+                                 false, false,
                                  prettyFlags, true);

     if (res == NULL)
@@ -1117,8 +1120,10 @@ pg_get_indexdef_ext(PG_FUNCTION_ARGS)

     prettyFlags = pretty ? (PRETTYFLAG_PAREN | PRETTYFLAG_INDENT | PRETTYFLAG_SCHEMA) : PRETTYFLAG_INDENT;

-    res = pg_get_indexdef_worker(indexrelid, colno, NULL, colno != 0, false,
-                                 false, prettyFlags, true);
+    res = pg_get_indexdef_worker(indexrelid, colno, NULL,
+                                 colno != 0, false,
+                                 false, false,
+                                 prettyFlags, true);

     if (res == NULL)
         PG_RETURN_NULL();
@@ -1134,10 +1139,13 @@ pg_get_indexdef_ext(PG_FUNCTION_ARGS)
 char *
 pg_get_indexdef_string(Oid indexrelid)
 {
-    return pg_get_indexdef_worker(indexrelid, 0, NULL, false, true, true, 0, false);
+    return pg_get_indexdef_worker(indexrelid, 0, NULL,
+                                  false, false,
+                                  true, true,
+                                  0, false);
 }

-/* Internal version that just reports the column definitions */
+/* Internal version that just reports the key-column definitions */
 char *
 pg_get_indexdef_columns(Oid indexrelid, bool pretty)
 {
@@ -1145,7 +1153,9 @@ pg_get_indexdef_columns(Oid indexrelid, bool pretty)

     prettyFlags = pretty ? (PRETTYFLAG_PAREN | PRETTYFLAG_INDENT | PRETTYFLAG_SCHEMA) : PRETTYFLAG_INDENT;

-    return pg_get_indexdef_worker(indexrelid, 0, NULL, true, false, false,
+    return pg_get_indexdef_worker(indexrelid, 0, NULL,
+                                  true, true,
+                                  false, false,
                                   prettyFlags, false);
 }

@@ -1158,7 +1168,8 @@ pg_get_indexdef_columns(Oid indexrelid, bool pretty)
 static char *
 pg_get_indexdef_worker(Oid indexrelid, int colno,
                        const Oid *excludeOps,
-                       bool attrsOnly, bool showTblSpc, bool inherits,
+                       bool attrsOnly, bool keysOnly,
+                       bool showTblSpc, bool inherits,
                        int prettyFlags, bool missing_ok)
 {
     /* might want a separate isConstraint parameter later */
@@ -1297,15 +1308,13 @@ pg_get_indexdef_worker(Oid indexrelid, int colno,
         Oid            keycolcollation;

         /*
-         * attrsOnly flag is used for building unique-constraint and
-         * exclusion-constraint error messages. Included attrs are meaningless
-         * there, so do not include them in the message.
+         * Ignore non-key attributes if told to.
          */
-        if (attrsOnly && keyno >= idxrec->indnkeyatts)
+        if (keysOnly && keyno >= idxrec->indnkeyatts)
             break;

-        /* Report the INCLUDED attributes, if any. */
-        if ((!attrsOnly) && keyno == idxrec->indnkeyatts)
+        /* Otherwise, print INCLUDE to divide key and non-key attrs. */
+        if (!colno && keyno == idxrec->indnkeyatts)
         {
             appendStringInfoString(&buf, ") INCLUDE (");
             sep = "";
@@ -1352,13 +1361,12 @@ pg_get_indexdef_worker(Oid indexrelid, int colno,
             keycolcollation = exprCollation(indexkey);
         }

-        if (!attrsOnly && (!colno || colno == keyno + 1))
+        /* Print additional decoration for (selected) key columns */
+        if (!attrsOnly && keyno < idxrec->indnkeyatts &&
+            (!colno || colno == keyno + 1))
         {
             Oid            indcoll;

-            if (keyno >= idxrec->indnkeyatts)
-                continue;
-
             /* Add collation, if not default for column */
             indcoll = indcollation->values[keyno];
             if (OidIsValid(indcoll) && indcoll != keycolcollation)
@@ -2197,6 +2205,7 @@ pg_get_constraintdef_worker(Oid constraintId, bool fullCommand,
                                                               false,
                                                               false,
                                                               false,
+                                                              false,
                                                               prettyFlags,
                                                               false));
                 break;

On 2018-Jul-18, David G. Johnston wrote:

> ​-1 for printing a boolean t/f; would rather spell it out:
> 
> CASE WHEN "Key" THEN 'Key' ELSE 'Included' END AS "Data"

+1

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> On 2018-Jul-18, David G. Johnston wrote:
>> -1 for printing a boolean t/f; would rather spell it out:
>> 
>> CASE WHEN "Key" THEN 'Key' ELSE 'Included' END AS "Data"

> +1

I can sympathize with the eyestrain argument against t/f, but the
above doesn't seem like an improvement --- in particular, "Data"
as the column header seems quite content-free.  My counterproposal
is to keep "Key" as the header and use "Yes"/"No" as the values.

I'd be OK with "Key"/"Included" as the values if someone can
propose an on-point column header to go with those.

            regards, tom lane

On 2018-Jul-18, Tom Lane wrote:

> I can sympathize with the eyestrain argument against t/f, but the
> above doesn't seem like an improvement --- in particular, "Data"
> as the column header seems quite content-free.  My counterproposal
> is to keep "Key" as the header and use "Yes"/"No" as the values.

I think "Key: no" is a bit obscure -- using "included" is a bit more
self-documenting and lends better to documentation searches.

> I'd be OK with "Key"/"Included" as the values if someone can
> propose an on-point column header to go with those.

"Role"?

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Oleksandr Shulgin <oleksandr.shulgin@zalando.de> writes:
> I don't think there is an established practice on how to display this sort
> of info, but I see that both styles already exist, namely:

> =# \dL
>                        List of languages
>     Name    |  Owner   | Trusted |         Description
> ------------+----------+---------+------------------------------
>  plpgsql    | postgres | t       | PL/pgSQL procedural language
>  plproxy    | postgres | f       |
> ...

> and

> =# \dC
>                                          List of casts
>          Source type         |         Target type         |      Function
>     |   Implicit?
> -----------------------------+-----------------------------+--------------------+---------------
>  abstime                     | date                        | date
>      | in assignment
>  abstime                     | integer                     | (binary
> coercible) | no
>  abstime                     | timestamp without time zone | timestamp
>     | yes
> ...

> I like the second option more, for readability reasons and because it is
> easier to extend if ever needed.

> Given that the documentation refers to included columns as "non-key
> columns", it seems natural to me to name the \d output column "Key?" and
> use "yes/no" as the values.

WFM, anyone want to argue against?

            regards, tom lane

> Given that the documentation refers to included columns as "non-key
> columns", it seems natural to me to name the \d output column "Key?" and
> use "yes/no" as the values.

WFM, anyone want to argue against?