Thanks Karen.
I went with the computed column because it saved the overhead of
crawling through a bunch of code and is therefore likely more foolproof.
But you know what they say - Once you've made something foolproof....
Nope, not what you're thinking - "...the programmer looks like a fool
cuz he missed a clump of code." :)
But you raise an interesting question: Does a computed column evaluate
each time it is referenced? Or does it evaluate only when the value of
the "parent" column changes and then saves that computed value in the table?
Doug
On 6/18/2025 2:35 PM, 'Karen Tellef' via RBASE-L wrote:
Here's my admittedly Low Tech way of doing it. Since you're willing
to add a column to the table anyway, why not have _two_ columns
updated when you print the label? Have a "IsLabelPrinted" column with
a Default of 0. When you print, update the date column with .#NOW and
update IsLabelPrinted = 1. Removes the extra overhead of a computed
column, which I think would have to compute for each row one at a time
as it evaluates your Where.
However, I can't answer how much faster that would be since there's
only 2 possible values in the table
Karen
On Wednesday, June 18, 2025 at 02:31:55 PM CDT, Doug Hamilton
<[email protected]> wrote:
I have an order table with a DateTime column, LabelsPrintedDTS. (DTS =
DateTimeStamp)
The table has a couple hundred thousand rows.
LabelsPrintedDTS gets updated with .#NOW when labels for specified rows
are printed.
Labels to be printed are selected WHERE LabelsPrintedDTS IS NULL.
As the table grows between semi-annual data archiving, the search for
labels to print becomes slower.
I don't think indexing LabelsPrintedDTS column will help because:
1) NULL values aren't indexed (this db is RB 10.5)
2) Indexing this column will bloat RX3, there may be up to 100 duplicate
LabelsPrintedDTS values.
What about:
Creating a calculated Boolean column, PartLabelsPrintedYN =
IFNULL(PartLabelsPrintedDTS,0,1).
i.e. If labels have not been printed, the value of PartLabelsPrintedYN
is 0, if printed, the value is 1.
Then the WHERE clause becomes: WHERE PartLabelsPrintedYN= 0
Would such a Boolean search be faster?
I'd test it but don't think my single-user system would be a fair
evaluation compared to customer's multi-user system.
I saw a Tip-Of-The-Day article about NULLS and indexing in RB 11 but
can't find it.
For whatever reason, I didn't think it would help this situation.
TIA,
Doug
--
This email has been checked for viruses by Avast antivirus software.
www.avast.com
--
For group guidelines, visit
http://www.rbase.com/support/usersgroup_guidelines.php
---
You received this message because you are subscribed to the Google
Groups "RBASE-L" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to [email protected].
To view this discussion visit
https://groups.google.com/d/msgid/rbase-l/5926b3d5-a5a7-4547-ae52-8ed9ab8448f7%40wi.rr.com.
--
For group guidelines, visit
http://www.rbase.com/support/usersgroup_guidelines.php
---
You received this message because you are subscribed to the Google
Groups "RBASE-L" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to [email protected].
To view this discussion visit
https://groups.google.com/d/msgid/rbase-l/220263378.57221.1750275358160%40mail.yahoo.com
<https://groups.google.com/d/msgid/rbase-l/220263378.57221.1750275358160%40mail.yahoo.com?utm_medium=email&utm_source=footer>.
--
For group guidelines, visit
http://www.rbase.com/support/usersgroup_guidelines.php
---
You received this message because you are subscribed to the Google Groups "RBASE-L" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion visit
https://groups.google.com/d/msgid/rbase-l/9f0d20d2-2c44-4a88-830b-6a16843ef180%40wi.rr.com.
