Обсуждение: weird objectaddress.c entry for transforms

Hi Peter, Alvaro, All,

(Peter, this is your change, Alvaro, you're the objectaddress.c master)

While looking through my oids removal patch I noticed that the
objectaddress.c ObjectProperty entry for transforms looks wrong:

typedef struct
{
    Oid            class_oid;        /* oid of catalog */
    Oid            oid_index_oid;    /* oid of index on system oid column */
    int            oid_catcache_id;    /* id of catcache on system oid column    */
    int            name_catcache_id;    /* id of catcache on (name,namespace), or
                                     * (name) if the object does not live in a
                                     * namespace */
    AttrNumber    attnum_name;    /* attnum of name field */
    AttrNumber    attnum_namespace;    /* attnum of namespace field */
    AttrNumber    attnum_owner;    /* attnum of owner field */
    AttrNumber    attnum_acl;        /* attnum of acl field */
    ObjectType    objtype;        /* OBJECT_* of this object type */
    bool        is_nsp_name_unique; /* can the nsp/name combination (or name
                                     * alone, if there's no namespace) be
                                     * considered a unique identifier for an
                                     * object of this class? */
} ObjectPropertyType;

    {
        TransformRelationId,
        TransformOidIndexId,
        TRFOID,
        InvalidAttrNumber,
    },

this was added in https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=cac76582053e

The InvalidAttrNumber entry initializes the member for name_catcache_id,
instead of -1 as the other entries, and in contrast to the rest of the
entries it doesn't initialize the other fields. In particular objtype is
0 instead of -1 as the rest.

Greetings,

Andres Freund

On 2018-11-19 17:53:22 -0800, Andres Freund wrote:
> Hi Peter, Alvaro, All,
> 
> (Peter, this is your change, Alvaro, you're the objectaddress.c master)
> 
> While looking through my oids removal patch I noticed that the
> objectaddress.c ObjectProperty entry for transforms looks wrong:
> 
> typedef struct
> {
>     Oid            class_oid;        /* oid of catalog */
>     Oid            oid_index_oid;    /* oid of index on system oid column */
>     int            oid_catcache_id;    /* id of catcache on system oid column    */
>     int            name_catcache_id;    /* id of catcache on (name,namespace), or
>                                      * (name) if the object does not live in a
>                                      * namespace */
>     AttrNumber    attnum_name;    /* attnum of name field */
>     AttrNumber    attnum_namespace;    /* attnum of namespace field */
>     AttrNumber    attnum_owner;    /* attnum of owner field */
>     AttrNumber    attnum_acl;        /* attnum of acl field */
>     ObjectType    objtype;        /* OBJECT_* of this object type */
>     bool        is_nsp_name_unique; /* can the nsp/name combination (or name
>                                      * alone, if there's no namespace) be
>                                      * considered a unique identifier for an
>                                      * object of this class? */
> } ObjectPropertyType;
> 
>     {
>         TransformRelationId,
>         TransformOidIndexId,
>         TRFOID,
>         InvalidAttrNumber,
>     },
> 
> this was added in https://git.postgresql.org/gitweb/?p=postgresql.git;a=commit;h=cac76582053e
> 
> The InvalidAttrNumber entry initializes the member for name_catcache_id,
> instead of -1 as the other entries, and in contrast to the rest of the
> entries it doesn't initialize the other fields. In particular objtype is
> 0 instead of -1 as the rest.

Also, uh, shouldn't this actually fill out some of those fields? Most of
them actually apply to transforms.

Greetings,

Andres Freund