>Hi Stephen, et.al.,
>Our user group newletter has yet to come out so I have placed my latest article on my site, Rushmore Optimization for your comments. It took considerable effort, so I am shamelessly checking the forward a copy to these members, hoping you all will give it a read. Included are charts showing the number of bytes used by the deleted tag per number of records, a TestData class generator, and other test info. For those who don't wish to check it out, the summary is, "DelTag, not so good."
>Charlie

That's a robuts article. 6 pages and full of good solid facts!

I agree with most of what you state here and question other statements. Logical fields their index's. I love the use of logicals, less space, less options, fast in queries.

Your findings of set deleted on vs off warrant the need for a tag on deleted(), if you never used it. But what about recycling rows? Now there is a necessary use for it?

Other issues are opening tables on your high speed segment. The deleted() tag aids the loading of a 1.5 M row table, when set deleted is on. I tested this a year ago with William Fields in Phx. The OPENING time droped when the deleted tag was present. About half to a third as I remember. That was Feb or March of this year.

So I'm not saying that you should remove your tag or not. I'm saying that you should consider your environment first and then go from there. You did the same in your article.

We both agreed that we nenver really tested the deleted tag in the beginning. I think that I ran down the path of recycled records quickly though.

Another thing is that SYS(30054) as a final word? Sorry but remember who made it!

I use time as the difference between start and end for my final analysis. SYS(3054) says that it was unoptimized, yet that particular statement returns the best performance. That is thourough testing on multiple machines from a cold boot.