Re: varbit question

On 15 May, Peter Eisentraut wrote:
> Laurent Duperval writes:
>
>> I have an Oracle script that I want to convert to pgsql. I've got a sed
>> script that does most of the mappings but it doesn't look like raw is
>> supported. In the docs, there is mention of varbit(n) but if I use the
>> (n) argument, the script fails. I can only use varbit alone. How do I get
>> around that?
>
> If you're not using version 7.1 or later you might want to upgrade. The
> bit string types do not work correctly in earlier releases. If you are
> using that version, please post more detail.
>

Turns out it's a bug in the docs, I think. The table of all data types
mentions the following:

Type Name Aliases Description
bit fixed-length bit string
bit varying(n) varbit(n) variable-length bit string

but if you read a bit further in the Bit string types section, it
says:

There are two SQL bit types: BIT(x) and BIT VARYING(x); the x specifies the
maximum length. BIT type data is automatically padded with 0's on the right
to the maximum length, BIT VARYING is of variable length. BIT without
length is equivalent to BIT(1), BIT VARYING means unlimited length

No mention of varbit(n) and the table doesn't say that bit(n) is acceptable.
So I fixed my script according to this and it works ok now.

L

--
Laurent Duperval <mailto:lduperval(at)microcelli5(dot)com>

Au lieu de tenir des propos teintés de paresse mélangée de fainéantise de
couleur assortie, introduisez une plaque de métal dans l'appareil.
-Léonard le génie