BUG #6017: COPY from CSV file, WITH CSV HEADER fields order not respected

The following bug has been logged online:

Bug reference: 6017
Logged by: pedro mg
Email address: pedro(at)tquadrado(dot)com
PostgreSQL version: 8.4.8
Operating system: Ubuntu Linux 10.10
Description: COPY from CSV file, WITH CSV HEADER fields order not
respected
Details:

Documentation says HEADER is discarded for reads. When uploading millions of
records from other legacy databases, would be an awesome addition to have
Header fields order not discarded for reads.

-Have a table: \d test1

key | integer | not null default
nextval('teste1_key_seq'::regclass)
c1 | character varying(10) |
c2 | double precision |
Indexes:
"teste1_pkey" PRIMARY KEY, btree (key)

- Have a CSV file, /tmp/test1.csv with an HEADER first line with the
following order (note:different from the table):

key,c2,c1
1,1,"Kg"
2,2,"g"
3,3,"mg"
4,4,"µg"
5,11,"l"
6,12,"ml"
7,13,"µl"

- Issue a COPY command:

# COPY test1 FROM '/tmp/test1.csv' WITH CSV HEADER

- Returns:

ERROR: invalid input syntax for type double precision: "Kg"
CONTEXT: COPY teste1, line 2, column c2: "Kg"