On 11/5/07, Tom Hart <tomhart@coopfed.org> wrote:
> Hey everybody. I'm running postgresql 8.2 on a windows 2k3 server
> machine. I have a table (two tables actually, in table and table_import
> format). The _import table has all text type fields, while the main
> table has datatypes defined.
>
> I wrote a PHP script that checks the various type fields (e.g. integer
> is a number, date is a date, etc.) and sets a bool flag is_ok to true
> for the row if the data all checks out. The script gets a dump of the
> data from a SELECT * statement, then takes each row, verifies the
> various data fields (successfully, I've tested) and sets a variable
> baddata. At the end of the checking, if baddata still equals 0, then it
> crafts an UPDATE statement to change the value of is_ok. There are a
> relatively small amount of rows (~1500, small time to you guys I'm
> sure), and the script runs fairly fast.
>
> OK, enough of the background, here's my issue. For some lovely reason,
> even though my script reports running an UPDATE query 1563 times (out of
> 1566 rows), only 316 rows have is_ok set to TRUE. I've tried a few
> times, changing this and that, and it always updates those 316 rows (no
> reason for this, the data is actually really good when it comes in, I'm
> just trying to build an extra layer of assuredness). Of particular note,
> I'm trying to keep the script fairly uniform and work for all our
> tables, so the UPDATE statement looks something like
>
> UPDATE table SET is_ok = 'TRUE' WHERE var1 = value1 AND var2 = value2
> AND var3 = value3.....
Can you create a repeatable test version of this? I.e. have one row
that does and one row that doesn't update? And give it to the list as
a .sql file to be loaded along with the queries you're using?
I understand the general idea of what you're saying, but I'm afraid
there's not enough detail in your post to really help.
