I have found a transaction problem in 6.5.3. This is an example:
create table a (a1 int4,a2 int4);
create table b (b1 int4);
alter table b add column b2 int4;
insert into b select * from a ;
ERROR: INSERT has more expressions than target columns
The db sees new table b, but thinks it has only one column: if I try
"select * from b;" after the alter table command, I receive:
Moreover, I found another bug, but it seems it's the same as reported
Radhesh Mohandas at Feb, 4:
Drop table a;
Drop table b;
-- now the second drop failed, e.g. table b doesn't exist - and I
-- want to roll back the whole transaction
Table a "does" and "doesn't" exist: it cannot be created but every
select from a failes.
I found the only way how to fix the problem: delete these tables from
system tables, this means "delete from pg_class where relname='a'"
and then "delete from pg_type where typname='a'". Then execute vacuum,
which should, I think, synchronize indices. I tried it and it worked.
Don't you have anybody better solution?
pgsql-bugs by date
|Next:||From: Tom Lane||Date: 2000-04-08 23:25:55|
|Subject: Re: insert after table modify bug |
|Previous:||From: Tom Lane||Date: 2000-04-07 22:19:07|
|Subject: Re: 7beta3 has naughty parser bug with ALTER USER .. PASSWORD |