_ (ndesc...@yahoo.com)

My understanding of "schema" that I discovered
in 7.3 (I don't think they were available before)
is that you can have two tables with the same name
if they are in different schemas.

I have done a google search, as well as archive search
but

pg_dump and pg_dumpall are broken if a database
contains schemas.

First of all if there are two tables with the same
name in different schemas pg_dump only dumps out
one table.  There is no way to dump other tables
and I have checked pg_dump man page

Restoring a pg_dumpall is now a nightmare because

I had as superuser

# create schema test authorization httpd

on a database not owned by database owner.
And it works merrily until the time to
dump and restore.

pg_dumpall answers to above create authorization is

\connect - httpd

create schema test

Hell breaks lose with that!  Because httpd cannot
create schema on a database that it does not own.
Why couldn't pg_dumpall does

create schema test authorization httpd

as superuser when the schema was created in that
fashion?

I really don't think anyone is going to pay attention
to this rant since these list does not like/answer anonymous posts
but I have to post just so some poor soul might find
it in the archive and be warned.

My current versions are 7.3.2 and 7.3.3 and I have been using
posgres since 7.1 and consider myself experienced with postgres

Schemas are the best thing since slice breads but
the baker decided to poison the bread.  Nice!

Thanks

__________________________________
Do you Yahoo!?
SBC Yahoo! DSL - Now only $29.95 per month!
http://sbc.yahoo.com

Let's say you drop a column on a table.  Then you dump
that table and load it into a slave server.  Then you
create dbmirror trigger on master database.  Then
dbmirror goes by original definition of master table
before the column was dropped.  In particular it is
looking for a column "........pg.dropped.2........"
where 2 is the column number that was dropped.   What
is really terrible is that not only does it generate
errors and stop mirroring this table it also stops
mirroring all other tables!!

I tried to "VACUUM FULL" the table hoping that
"........pg.dropped.2........" will be gotten rid of
from a low level definition of this table which
dbmirror is referring but no luck.  "VACUUM FULL" does
not get rid of that dropped column from some kind of
system data which dbmirror is referring to.

Is the only way out of this problem dumping the master
table and reloading?  The table is a live one (being
inserted every second on average) and I prefer to
minimize interruptions

Thanks


__________________________________________________
Do you Yahoo!?
Yahoo! Tax Center - forms, calculators, tips, more
http://taxes.yahoo.com/

Ignore previous.  Error is caused by dbmirror trigger
on another table which is triggered by a delete on the
table in question

__________________________________________________
Do you Yahoo!?
Yahoo! Tax Center - forms, calculators, tips, more
http://taxes.yahoo.com/

Hello AllI am a long time user and with a wishful thinking I attempted create a pl/pgsql function with return type "record". Now I can now longer drop that data type even after I drop the function.Here is what I get if I do \dT.. snipped..polygon | geometric polygon '(pt1,...)' real | single-precision floating point number, 4-byte storage record | refcursor | reference cursor (portal name) regproc | registered procedure..----Now if I drop it as as a database owner I get permission denied---what=> drop type record;ERROR: RemoveType: type 'record': permission deniedwhat=> ---So I change myself to superuser and try to drop---what=# drop type record;ERROR: RemoveType: type '_record' does not existwhat=# ---But I can still see that data type if I do \dT as superuser.The problem I have now is that whenever I do a pg_dump I get "owner of data type record appear to be invalid"Wha
t else can I do?3 features request.1. Is there any support in the future for returning data type "record" from a pl/pgsql function? I would like to call a function and get a nice rows of data. Currently I am joining lots of data with separator, returning varchar and then parsing it at my front end.2. When I do pg_dumpall C functions don't end up in the dump. Is it possible to dump and restore everything including C functions?3. Is it possible at all in the future to do major upgrade without having to do initdb? Is there any serious reasons why initdb is required to do a major upgrade? My pg server has to be up 24x7 and I don't see any way to do initdb without stopping the server.Thanks for such a great product.k.w.


---------------------------------
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now