On Wed, May 26, 2021 at 7:55 PM Bharath Rupireddy
<bharath.rupireddyforpostg...@gmail.com> wrote:
>
> On Wed, May 26, 2021 at 7:38 PM vignesh C <vignes...@gmail.com> wrote:
> > > Attaching v5 patch, please have a look.
> >
> > We get the following error while adding an index:
> > create publication mypub for table idx_t1;
> > ERROR: "idx_t1" is an index
> >
> > This error occurs internally from table_openrv function call, if we
> > could replace this with relation_openrv and then check the table kind,
> > we could throw a similar error message here too like the other changes
> > in the patch.
>
> Do you say that we replace table_open in publication_add_relation with
> relation_open and have the "\"%s\" is an index" or "\"%s\" is a
> composite type" checks in check_publication_add_relation? If that is
> so, I don't think it's a good idea to have the extra code in
> check_publication_add_relation and I would like it to be the way it is
> currently.
Before calling check_publication_add_relation, we will call
OpenTableList to get the list of relations. In openTableList we don't
include the errordetail for the failure like you have fixed it in
check_publication_add_relation. When a user tries to add index objects
or composite types, the error will be thrown earlier itself. I didn't
mean to change check_publication_add_relation, I meant to change
table_openrv to relation_openrv in OpenTableList and include error
details in case of failure like the change attached. If you are ok,
please include the change in your patch.
Regards,
Vignesh
diff --git a/src/backend/commands/publicationcmds.c b/src/backend/commands/publicationcmds.c
index 95c253c8e0..b6380d7491 100644
--- a/src/backend/commands/publicationcmds.c
+++ b/src/backend/commands/publicationcmds.c
@@ -16,6 +16,7 @@
#include "access/genam.h"
#include "access/htup_details.h"
+#include "access/relation.h"
#include "access/table.h"
#include "access/xact.h"
#include "catalog/catalog.h"
@@ -523,7 +524,21 @@ OpenTableList(List *tables)
/* Allow query cancel in case this takes a long time */
CHECK_FOR_INTERRUPTS();
- rel = table_openrv(rv, ShareUpdateExclusiveLock);
+ rel = relation_openrv(rv, ShareUpdateExclusiveLock);
+ if (rel->rd_rel->relkind == RELKIND_INDEX ||
+ rel->rd_rel->relkind == RELKIND_PARTITIONED_INDEX)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+ errmsg("\"%s\" is an index",
+ RelationGetRelationName(rel)),
+ errdetail("Index objects not supported by publications.")));
+ else if (rel->rd_rel->relkind == RELKIND_COMPOSITE_TYPE)
+ ereport(ERROR,
+ (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
+ errmsg("\"%s\" is a composite type",
+ RelationGetRelationName(rel)),
+ errdetail("Composite type not supported by publications.")));
+
myrelid = RelationGetRelid(rel);
/*
