Re: Why are data types size limited?

"Jan-Erik Rosinowski" <spamfilter_at_rosinowski.de> wrote in message news:38dde75e.46662887_at_News.CIS.DFN.DE...

> >I still maintain it's primarily to facilitate B-tree-style
> >indexing and is a performance issue, like why C/C++'s integers are often
> >only 32 bits.
 

> most rdbms cut indexentries after a "reasonable" number of chars.
> sqlaw for example did this iirc after 16 chars. other dbms won't do it
> much different.
 

> if some stupid consideres indexing a blob to be a viable option it's
> his personal problem. no need to limit the db features because of
> this.
 

> the real problems begin when your "records" span multiple "pages".
> this causes drastic penaties and is not considered to be standard
> operation of your dbms.
 

> if you want it you have to pay: take clobs, memos or whatever it's
> called. but it'll be slow as hell

I can imagine a scheme which allows memos to be indexed. Only a "reasonable" number of bytes are actually indexed, and another "reasonable" number of bytes are actually stored in the record, with the rest stored wherever BLOB data is stored currently. If the text data is kept within "reasonable" bounds, performance is near what is currently acceptable for varchar data. However, if someone actually has an email address longer than 255 or so characters, it won't cause havoc in an app.

--
Joe Foster <mailto:jfoster_at_ricochet.net>  Space Cooties! <http://www.xenu.net/>
WARNING: I cannot be held responsible for the above        They're   coming  to
because  my cats have  apparently  learned to type.        take me away, ha ha!