After receiving the newsletter for about 13 years, I unsubscribed today.

I see no improvement for the readers ... looks like it was done to improve ad placement, i.e. the narrow width and, hence, longer page length to accommodate ads. ...

Excellent! I'm looking forward to the follow-on articles.

... here's a stab at it:

SELECTDISTINCT

COALESCE(b.ID, a.ID), COALESCE(b.RowNo, a.RowNo), COALESCE(b.Value, a.Value)

FROM#Test1 a FULL OUTER JOIN

#Test2 b

ONa.ID = b.ID

ANDb.RowNo IS NOT NULL

ORDER BY

1,2

It may be all wrong ...

David is absolutely 10% correct.

The "hazard" is given here:

"Hastily, we consult MSDN"

and

"A quick scan of the article"

and

"Obviously this must be the ticket, so let’s give that a try"

Phil...

I didn't really ... because I now have to stick my neck out, and put up or shut up on the...

CROSS APPLY (

VALUES ('Revenue', JanRev, FebRev, MarRev, AprRev, MayRev, JunRev

, JulRev, AugRev,...

... May we see a "full year" version of the cross apply? Then we'll start adding years...

Good to know ... I was unawares.

Here's the last example done with pivot/unpivot ... a bit more code, performance not as good ... but much more maintainable/flexible (IMO) given the...

Maybe this to account for signs . . .

EXP(SUM(LOG(ABS([RowValue])))) * SIGN(CHECKSUM_AGG([RowValue]))

Good info . . . thanks!

Though dated a little, this thread is the closet to my project I've been able find. Any more info/progress to report?

Really . . . another ridiculous article! I am responding to this reader's reply to illustrate the problem. Misinformation breeds misinformation . . . STR is deterministic!!!!!!!!!!!!!!!!!! ...

The article's title/subtitles and other wording imply ISNUMERIC does not work properly. Here's a better way to explain the issue:

http://www.simple-talk.com/community/blogs/johnm/archive/2009/06/11/73729.aspx

The issue is ISNUMERIC returns true for values which are valid numeric data types and are not "only numeric digits" (look at the article intro again). I suppose "-1"...