Re: [HACKERS] UTF8 or Unicode

lsunley(at)mb(dot)sympatico(dot)ca wrote:
> In <200502270055(dot)j1R0te122973(at)candle(dot)pha(dot)pa(dot)us>, on 02/26/05
> at 07:55 PM, Bruce Momjian <pgman(at)candle(dot)pha(dot)pa(dot)us> said:
>
> >lsunley(at)mb(dot)sympatico(dot)ca wrote:
> >> Further to my earlier e-mail, there would have to be two lines added to
> >> conversion_create.sql for each alternate function name
> >>
> >> Like:
> >>
> >> CREATE OR REPLACE FUNCTION ascii_to_whatever (INTEGER, INTEGER, CSTRING,
> >> CSTRING, INTEGER) RETURNS VOID AS '$libdir/ascmic', 'ascii_to_mic'
> >> LANGUAGE 'c' STRICT;
> >>
> >> CREATE CONVERSION pg_catalog.ascii_to_whatever FOR 'SQL_ASCII' TO
> >> 'MULE_INTERNAL' FROM ascii_to_whatever;
>
> >I would like to avoid bloating the system with aliases. I think it is OK
> >to have it happen with specific names but doing it for all the
> >combinations is too much, and these duplicates are in the system tables,
> >not in the C code. It just doesn't seem worth it to me.
>
> Good point, perhaps having a script generated that can be executed by the
> DBA to insert any deprecated conversion routine names would be the best
> route to take.

I am unsure why anyone would use the awkward names when you can more
easily supply the actual encoding names, and all the aliases work there.

Also, consider this is going to be a major release change so they are
going to dump/reload anyway. I see I need to update
conversion_create.sql too. I was previously only looking in C files.
Let me do a bigger search.

--
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