Subject | Re: UTF-8 vs UTF-16 |
---|---|
Author | Roman Rokytskyy |
Post date | 2003-08-27T05:24:19Z |
David,
data from the database (isc_open_blob, isc_get_segment,
isc_close_blob). If you store binary data that are always smaller than
32k, it makes sense to put them into varchar and not into the blob and
access will be faster.
I will agree with you that conceptually this is wrong, but wire
protocol changes are planned for 2.0, and this unicode stuff might
belong to 1.6 already.
Roman
> Anyone that thinks they need a 16K character column isIf you put data in the row, you need three requests more to fetch that
> smoking something which the State of California says
> is OK for medical purposes, but the federal
> government disagrees.
>
> Such a beast belongs in a blob - and use of blobs
> in expressions (where upper(blob) like '%FOO%')
> should have better support.
>
> Seriously, who needs a row of data that large?
data from the database (isc_open_blob, isc_get_segment,
isc_close_blob). If you store binary data that are always smaller than
32k, it makes sense to put them into varchar and not into the blob and
access will be faster.
I will agree with you that conceptually this is wrong, but wire
protocol changes are planned for 2.0, and this unicode stuff might
belong to 1.6 already.
Roman