Brian Aker wrote:
Hi!
On Sep 11, 2008, at 12:24 PM, Jim Starkey wrote:
select cast (string_field as date) from someplace
is useful for debugging. I imagine there's a legitimate use for it
as well...
That part is staying, it is just the conversion for char types that
will go away.
My thinking about blobs has changed. Blobs originally were an escape
from fixed length SQL types. Now that I've abandoned fixed length
SQL types, the utility of blob as a declared type diminishes to about
nothing. Nimbus will retain clob/blob types to humor the
traditional, but they will (probably) be synonyms for string and
bytes, respectively. There will still be a storage type for blobs,
but it will be dynamic, based on a length threshold. (Why, I can
hear you asking? Simple: There is no point to slop around a
high-res jpeg to update a last_reference column in the same row.)
In Drizzle we have two type types of scalars. We have what is
essentially is bounded string of UTF8 characters that we treat with a
collation for sorting. We also have a straight byte array which will
sort based on binary position (its in there... not found a good reason
to remove it). The binary type is unbounded for size (well... it can
be larger then 4gigs).
The only reason for bounding on the UTF8 type is because of the layout
of row[]. Once I can refactor that out to be being, hold your breath,
a value object, I will remove that limitation.
A value object? A Value Object? A VALUE OBJECT!!!!
Cool.
--
Jim Starkey
President, NimbusDB, Inc.
978 526-1376
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
