Re: Re: why data translate error when selecting from view?

Great question, Joe.

The answer is that your CTE rewrite will give a plan that is basically
identical to the one produced by use of the view. There are some
cases--generally for performance or due to things like UDFs with external
actions, etc.--when the optimizer can't/won't merge the CTEs or views into
the outer select, but in general, if you can see a way to do it (and
sometimes even when you can't), the optimizer is probably going to do it,
too.

Tim Clark
DB2 for IBM i

"MIDRANGE-L" <midrange-l-bounces@xxxxxxxxxxxxxxxxxx> wrote on 05/16/2019
05:02:07 PM:

From: Joe Pluta <joepluta@xxxxxxxxxxxxxxxxx>
To: midrange-l@xxxxxxxxxxxxxxxxxx,
Date: 05/16/2019 05:02 PM
Subject: [EXTERNAL] Re: why data translate error when selecting from

view?

Sent by: "MIDRANGE-L" <midrange-l-bounces@xxxxxxxxxxxxxxxxxx>

Thatnks for that explanation, Tim. In a related question: do CTEs allow

me to effectively "hardcode" the order, causing the SQL engine to work
in my predefined sequence?

For example, if in this case I did this:

with t1 as (select * from contr where length(trim(cpcsid)) = 10),
t2 as (select a.cpcono cono, case ... end csno from t1)
select * from t2 where csno = 1346100

Would that force the SQL engine to create my T1 in its entirety (thus
skipping short fields) before then doing the select? Or is the engine
just so darned smart that it ignores my sequence and just optimizes me
into errors? Of course, upon further reflection, if
length(trim(cpcsid)) = 10, then no need to check for blanks, so no CASE
statement. But I digress...


On 5/16/2019 4:36 PM, Timothy P Clark wrote:

You can think of the query engine "inlining" the view into the final
select statement.

<...snip lots of relevant and cool stuff...>

Keeping in mind that SQL has no rules about predicate evaluation

order, it

should be clear that if the optimizer chooses to evaluate the 2nd
predicate before the first predicate, we can run into the error you're
seeing.

--
This is the Midrange Systems Technical Discussion (MIDRANGE-L) mailing

list

To post a message email: MIDRANGE-L@xxxxxxxxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: https://urldefense.proofpoint.com/v2/url?

u=https-3A__lists.midrange.com_mailman_listinfo_midrange-2Dl&d=DwIGaQ&c=jf_iaSHvJObTbx-

siA1ZOg&r=oCmVbiwufH6yM8U29GriOcB5LKDoEG8y1HZD-01RVtc&m=uIFIUpL4kLjo726zd1hGUB8znDqd5lwGjlGB_n14Gds&s=upUivQxU54IEPST7W6WD7ClY4OABHgBLq47FZQ9fPvQ&e=

or email: MIDRANGE-L-request@xxxxxxxxxxxxxxxxxx
Before posting, please take a moment to review the archives
at https://urldefense.proofpoint.com/v2/url?
u=https-3A__archive.midrange.com_midrange-2Dl&d=DwIGaQ&c=jf_iaSHvJObTbx-

siA1ZOg&r=oCmVbiwufH6yM8U29GriOcB5LKDoEG8y1HZD-01RVtc&m=uIFIUpL4kLjo726zd1hGUB8znDqd5lwGjlGB_n14Gds&s=9hUfo5-

Jsc8rw-Nc0PpdYV26783BHKMUqEKwpQr4JME&e=.

Please contact support@xxxxxxxxxxxx for any subscription related

questions.

Help support midrange.com by shopping at amazon.com with our affiliate

link:

https://urldefense.proofpoint.com/v2/url?
u=https-3A__amazon.midrange.com&d=DwIGaQ&c=jf_iaSHvJObTbx-

siA1ZOg&r=oCmVbiwufH6yM8U29GriOcB5LKDoEG8y1HZD-01RVtc&m=uIFIUpL4kLjo726zd1hGUB8znDqd5lwGjlGB_n14Gds&s=ryY7pNASzUnVB0hx1rBpE2qPcN3fHc375U4UJD2dmrk&e=