Re: [HACKERS] Dbsize backend integration

> -----Original Message-----
> From: Tom Lane [mailto:tgl(at)sss(dot)pgh(dot)pa(dot)us]
> Sent: 04 July 2005 14:54
> To: Dave Page
> Cc: Dawid Kuroczko; Andreas Pflug; Bruce Momjian;
> PostgreSQL-patches; PostgreSQL-development
> Subject: Re: [HACKERS] [PATCHES] Dbsize backend integration
>
> "Dave Page" <dpage(at)vale-housing(dot)co(dot)uk> writes:
> > Aside from the fact that's a change to the API that we had
> settled on,
> > it doesn't solve the actual problem of needing a suitable name for a
> > function that returns the size of a table /or/ index.
> pg_relation_size()
> > or pg_table_size() can't be used for precisely the reason they were
> > rejected for that purpose in the first place.
>
> Rejected by whom? pg_relation_size is an excellent choice for that.

Bruce didn't like it
(http://archives.postgresql.org/pgsql-hackers/2005-06/msg01410.php), and
you seemed to object as well
(http://archives.postgresql.org/pgsql-hackers/2005-06/msg01247.php)

Personally I'm beyond caring much now as the amount of time spent trying
to name these simple functions is wildly disproportionate the the effort
take to actually code them. I think we just need to agree there is no
perfect name and rely on the comments and docs to guide people. I think
the current names work OK, and Bruce and Dawid at least agree!

Regards, Dave.

On Monday 04 July 2005 10:11, Dave Page wrote:
> > -----Original Message-----
> > From: Tom Lane [mailto:tgl(at)sss(dot)pgh(dot)pa(dot)us]
> > Sent: 04 July 2005 14:54
> > To: Dave Page
> > Cc: Dawid Kuroczko; Andreas Pflug; Bruce Momjian;
> > PostgreSQL-patches; PostgreSQL-development
> > Subject: Re: [HACKERS] [PATCHES] Dbsize backend integration
> >
> > "Dave Page" <dpage(at)vale-housing(dot)co(dot)uk> writes:
> > > Aside from the fact that's a change to the API that we had
> >
> > settled on,
> >
> > > it doesn't solve the actual problem of needing a suitable name for a
> > > function that returns the size of a table /or/ index.
> >
> > pg_relation_size()
> >
> > > or pg_table_size() can't be used for precisely the reason they were
> > > rejected for that purpose in the first place.
> >
> > Rejected by whom? pg_relation_size is an excellent choice for that.
>
> Bruce didn't like it
> (http://archives.postgresql.org/pgsql-hackers/2005-06/msg01410.php), and
> you seemed to object as well
> (http://archives.postgresql.org/pgsql-hackers/2005-06/msg01247.php)
>
> Personally I'm beyond caring much now as the amount of time spent trying
> to name these simple functions is wildly disproportionate the the effort
> take to actually code them. I think we just need to agree there is no
> perfect name and rely on the comments and docs to guide people. I think
> the current names work OK, and Bruce and Dawid at least agree!
>

Actually I'd agree with Tom, pg_dbfile_size is ugly, and suggest to me I could
use a filename as an argument. ISTM that if we think that functions like
pg_database_size and pg_tablespace_size all make sense, the natural extension
would be functions called pg_index_size to tell us the size of an index,
pg_table_size to tell us the size of a table (table+toast) without it's
indexes, and some form of pg_table_plus_indexes_size for a table and its
indexes for those that feel we need both. I'm not sold we need a function
that can return either an index or table size, but if so something like
pg_object_size seems ambigious enough to work, and is future proof enough to
handle things like materialized views when and if they arise.

Just my .02 :-)

--
Robert Treat
Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL

Robert Treat wrote:
> Actually I'd agree with Tom, pg_dbfile_size is ugly, and suggest to me I could
> use a filename as an argument. ISTM that if we think that functions like
> pg_database_size and pg_tablespace_size all make sense, the natural extension
> would be functions called pg_index_size to tell us the size of an index,
> pg_table_size to tell us the size of a table (table+toast) without it's
> indexes, and some form of pg_table_plus_indexes_size for a table and its
> indexes for those that feel we need both. I'm not sold we need a function
> that can return either an index or table size, but if so something like
> pg_object_size seems ambigious enough to work, and is future proof enough to
> handle things like materialized views when and if they arise.

You are into the cycle we were in. We discussed pg_object size (too
vague) and pg_index_size (needs pg_toast_size too, and maybe toast
indexes; too many functions).

--
Bruce Momjian | http://candle.pha.pa.us
pgman(at)candle(dot)pha(dot)pa(dot)us | (610) 359-1001
+ If your life is a hard drive, | 13 Roberts Road
+ Christ can be your backup. | Newtown Square, Pennsylvania 19073

On Monday 04 July 2005 13:25, Bruce Momjian wrote:
> Robert Treat wrote:
> > Actually I'd agree with Tom, pg_dbfile_size is ugly, and suggest to me I
> > could use a filename as an argument. ISTM that if we think that
> > functions like pg_database_size and pg_tablespace_size all make sense,
> > the natural extension would be functions called pg_index_size to tell us
> > the size of an index, pg_table_size to tell us the size of a table
> > (table+toast) without it's indexes, and some form of
> > pg_table_plus_indexes_size for a table and its indexes for those that
> > feel we need both. I'm not sold we need a function that can return
> > either an index or table size, but if so something like pg_object_size
> > seems ambigious enough to work, and is future proof enough to handle
> > things like materialized views when and if they arise.
>
> You are into the cycle we were in. We discussed pg_object size (too
> vague) and pg_index_size (needs pg_toast_size too, and maybe toast
> indexes; too many functions).

Yeah, I read those discussions, and think you were better off then than you
are now, which is why I went back to it somewhat.

--
Robert Treat
Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL

>>You are into the cycle we were in. We discussed pg_object size (too
>>vague) and pg_index_size (needs pg_toast_size too, and maybe toast
>>indexes; too many functions).
>
> Yeah, I read those discussions, and think you were better off then than you
> are now, which is why I went back to it somewhat.

To be honest, the amount of effort being expended on this naming
discussion far outweighs the benefits. Maybe it's time for a core
member to step in and just resolve it - one way or the other?

Chris