Re: SQL Humor

From: JT <>
Date: Fri, 19 Aug 2005 14:59:12 -0400
Message-ID: <unM5zBPpFHA.1416_at_TK2MSFTNGP09.phx.gbl>

GB sized XML documents are typically loaded into structures such as datasets which swap the data into and out of memory as needed. I don't see why an application would need to load the entire contents into a flat memory object like a string. If you want to write an application that loads GBs of data into memory, then I'm sure the end users will express their own opinions about the design when their PC or server bogs down due to depleted resources.

"Mikito Harakiri" <> wrote in message
> JT wrote:
>> Once you start dealing with text of length > 4000 (Unicode) 8000 (ANSI),
>> you are practically working with documents not attributes, so this data
>> should be stored in a document management system with numeric or URL
>> pointers stored in the relational database. The TEXT datatype is there
>> for
>> large blobs of data for those so inclined, but it's an entirely different
>> case usage than VarChar.
>> On the application side, if you are storing gigabytes or even 10s
>> megabytes of data in a string or array, then it is time to start
>> re-considering the application design.
> Remarkable. You suggest that arbitrary limitations on datatype length
> provide invaluable insight on application design weakness? So if my app
> processes 1GB strings all over the places (think XML!), I'm fine, but
> if I get 3GB string, then I should redesign my application?
Received on Fri Aug 19 2005 - 20:59:12 CEST

Original text of this message