Interesting discussion.. Should have jumped in earlier.

1. I'm truly not aware of any place in the standard where the
implementation of any feature, including data types is prescribed.
The standard prescribes externals only and there primarily DML statements

2. Standards prevent lock-in. That in itself is nearly priceless.
Without standards you couldn't use your wireless at the airport,
You'd need an infinite number of power adapters,
you couldn't drive just about any car with your drivers license.
The fact that the wheelbase of a care is based on the width of two
horses sucks if you want to stretch out on the backseat of the car, but
at least your car fits into the parking spot which is way more important.

There are many things I do not like about the standard, but I wouldn't
want to miss it anymore than the knowledge that I can slide in a DVD and
know I can watch a movie (most times).

3. Regarding the original topic:
The right tool for the right job. And just because all you know is a
hammer doesn't mean everything long and pointy is a nail.

Cheers
Serge
--
Serge Rielau
SQL Architect DB2 for LUW, IBM Toronto Lab
Blog: tinyurl.com/SQLTips4DB2
Wiki: tinyurl.com/Oracle2DB2Wiki
Twitter: srielau

Jerry Stuckle <jstucklex (AT) attglobal (DOT) net> wrote:
Again you're so eager to bring an argument in your favour (under-
standable, you so seldom get that chance) that you completely miss
the topic. And thus your argument becomes void.

One could also say you started a completely different discussion,
disproving an argument that was never uttered.

I did not say that the distinction CHAR vs. VARCHAR is useless.
But that the declaration of a column as CHAR is just a hint and
only a hint. And that the DBMS should be free to ignore that hint
(for internal handling, the external view must be kept)

It *might* turn out that ignoring such hints is a bad idea. But
OTOH it might not. In any case such behavior should not count as
a standard violation (and as it seems it doesn't anyway, looks
like I remembered that point wrong)


XL

On 12/30/2010 7:33 PM, Axel Schwenke wrote:
Incorrect. This is the discussion YOU started.

No, the standard does not indicate how the data must be stored
internally; for instance a date field can be stored as y-m-d, it could
be stored as seconds since the universe started, or even the number of
kumquats in a skein.

However, there is a difference between fixed and variable length
columns, and in large databases running on high performance systems that
difference is critical. But never having been introduced to those
systems, you have insufficient background to understand why.

And it's things like this which experienced dba's take into
consideration, and a good part of why they consider MySQL to be a toy
database. It is nowhere near up to the standards of the big boys.

--
==================
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex (AT) attglobal (DOT) net
==================

CHAR always has a fixed length (or NULL).
When you concatenate a '!' to a CHAR(10) of value 'Hello' then you get
'Hello !'
When you do the same with a VARCHAR(10) you get 'Hello!' (no spaces).
Now, that behavior is orthogonal from the way the DBMS chooses to store
the data.
One way to store such data could be with trim any spaces from a CHAR and
store it along with a length (as proposed in the thread).
but the _defined_ type MUST remain CHAR so that the DBMS when it
operates on the stored value can add the blanks back in as needed.

Cheers
Serge


--
Serge Rielau
SQL Architect DB2 for LUW, IBM Toronto Lab
Blog: tinyurl.com/SQLTips4DB2
Wiki: tinyurl.com/Oracle2DB2Wiki
Twitter: srielau