Re: estimating RGZPFM runtime

[Sam_L <lennon_s_j@xxxxxxxxxxx>]

Didn't intended to debate your use of RRN...

In what I've done I've always *assumed* the UP with journal seq # 10002 
is always preceded by UB  # 10001 and they relate to the same object and 
record.  In practice, on our machine that has always worked out.  (You 
can do a fairly simple join with SQL to put the two records into one row.)

Now I can't say that I've ever read anywhere this is quaranteed, but 
neither have I read that it isn't so.

Sam

On 11/19/2013 5:49 PM, Stone, Joel wrote:
> Fair enough - but how do you know that 10002 and the prior image 10001 are the same [for example] product row??
>
> You could have 10002 be the AFTER image for an add of "Cheerios grape" and 10001 be the AFTER image for an add of "Cheerios cherry".