> Joe Kislo <postgre@athenium.com> writes:
> > Hmm, 1G is probably fine :) But is there going to be a blob type with
> > toast? If I want to store a large binary object, and have the ability
> > of retrieving it strictly over the postgre database connection, would I
> > be retrieving a blob column, or a really long varchar column?
>
> If you want binary (8-bit-clean) data, you need to use the 'bytea'
> datatype not 'varchar'. Our character datatypes don't cope with
> embedded nulls presently. This is primarily an issue of the
> external representation as a C string.
>
> Alternatively, you can keep using the old-style large-object support
> (lo_read, lo_write, etc). This may be handy if you are dealing with
> blobs large enough that you don't want to read or write the entire
> value on every access. We need to add that capability to bytea too,
> by defining some access functions that allow reading and writing
> portions of a bytea value --- but no one's gotten round to that yet,
> so I don't suppose it'll happen for 7.1.
What I think we _really_ need is a large object interface to TOAST data.
We already have a nice API, and even psql local large object handling.
If I have a file that I want loaded in/out of a TOAST column, we really
should make a set of functions to do it, just like we do with large
objects.
This an obvious way to load files in/out of TOAST columns, and I am not
sure why it has not been done yet. I am afraid we are going to get
critisized if we don't have it soon.
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026