Обсуждение: Array detection in pg_dump

Is there a reason we don't use pg_type.typcategory to detect arrays in
Postgres 8.4?  Right now I see this in pg_dump.c:
   if (g_fout->remoteVersion >= 80300)   {appendPQExpBuffer(query, "SELECT tableoid, oid, typname, "
"typnamespace,"          "(%s typowner) AS rolname, "          "typinput::oid AS typinput, "          "typoutput::oid
AStypoutput, typelem, typrelid, "          "CASE WHEN typrelid = 0 THEN ' '::\"char\" "          "ELSE (SELECT relkind
FROMpg_class WHERE oid = typrelid) END AS typrelkind, "          "typtype, typisdefined, "          "typname[0] = '_'
ANDtypelem != 0 AND "
 
-->  "(SELECT typarray FROM pg_type te WHERE oid = pg_type.typelem) = oid AS isarray "          "FROM pg_type",
username_subquery);   }
 

It seems the appropriate 8.4+ test would be:
t.typtype = 'b' AND t.typcategory = 'A'

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +

Bruce Momjian <bruce@momjian.us> writes:
> Is there a reason we don't use pg_type.typcategory to detect arrays in
> Postgres 8.4?  Right now I see this in pg_dump.c:

typcategory is user-assignable and thus not too reliable; furthermore
it wouldn't prove that the type is the array type for its typelem.
(Consider things like point and name --- there can be multiple types
with the same typelem, but only one is the real array type for that
typelem.)  The typelem back-link check is much safer.
        regards, tom lane

Bruce Momjian wrote:
> Is there a reason we don't use pg_type.typcategory to detect arrays in
> Postgres 8.4?  Right now I see this in pg_dump.c:
> 
>     if (g_fout->remoteVersion >= 80300)
>     {
>     appendPQExpBuffer(query, "SELECT tableoid, oid, typname, "
>               "typnamespace, "
>               "(%s typowner) AS rolname, "
>               "typinput::oid AS typinput, "
>               "typoutput::oid AS typoutput, typelem, typrelid, "
>               "CASE WHEN typrelid = 0 THEN ' '::\"char\" "
>               "ELSE (SELECT relkind FROM pg_class WHERE oid = typrelid) END AS typrelkind, "
>               "typtype, typisdefined, "
>               "typname[0] = '_' AND typelem != 0 AND "
> -->  "(SELECT typarray FROM pg_type te WHERE oid = pg_type.typelem) = oid AS isarray "
                                     ^^

Oh, and what does that 'te' do?  Seems useless.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +

Tom Lane wrote:
> Bruce Momjian <bruce@momjian.us> writes:
> > Is there a reason we don't use pg_type.typcategory to detect arrays in
> > Postgres 8.4?  Right now I see this in pg_dump.c:
> 
> typcategory is user-assignable and thus not too reliable; furthermore
> it wouldn't prove that the type is the array type for its typelem.
> (Consider things like point and name --- there can be multiple types
> with the same typelem, but only one is the real array type for that
> typelem.)  The typelem back-link check is much safer.

Thanks;  I will use the pg_dump code in pg_migrator then.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +

Bruce Momjian wrote:
> Bruce Momjian wrote:
> > Is there a reason we don't use pg_type.typcategory to detect arrays in
> > Postgres 8.4?  Right now I see this in pg_dump.c:
> > 
> >     if (g_fout->remoteVersion >= 80300)
> >     {
> >     appendPQExpBuffer(query, "SELECT tableoid, oid, typname, "
> >               "typnamespace, "
> >               "(%s typowner) AS rolname, "
> >               "typinput::oid AS typinput, "
> >               "typoutput::oid AS typoutput, typelem, typrelid, "
> >               "CASE WHEN typrelid = 0 THEN ' '::\"char\" "
> >               "ELSE (SELECT relkind FROM pg_class WHERE oid = typrelid) END AS typrelkind, "
> >               "typtype, typisdefined, "
> >               "typname[0] = '_' AND typelem != 0 AND "
> > -->  "(SELECT typarray FROM pg_type te WHERE oid = pg_type.typelem) = oid AS isarray "
> 
>                                       ^^
> 
> Oh, and what does that 'te' do?  Seems useless.

Oh, I see it now;  'te' makes the pg_type reference in the subquery
reference the outer pg_type;  Wow that is confusing.  I will clear that
up in my pg_migrator version.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +