marshall.jones (5/23/2009)

---

That would be the case if I was actually doing what's happening in that example.

In practice other constraints on the datasets ensure that the number of reads is in the hundreds if anything.

I would much prefer using a single column to store information in my cases than many separate columns. From an academic perspective a column per value is better but in real life I don't want to have to add a new column to a table, add a new index, return a new column in my result sets, add a new parameter into my stored procedures and anything else I've missed.

The alternative of adding a new bit value starts to look pretty good in that situation. Horses for courses of course.

mike brockington (5/22/2009)

---

I would love to see what this is being compared against, in order to describe this as 'elegant'.

In my dictionary, elegance requires something to be as simple as practicable, but
Bitwise comparisons are really inefficient, and extremely hard to debug by eye.

Andrew Peterson (5/22/2009)

---

Great article. And nice example use of bitmask for searching.

If you approach the bitmask column for the purpose of high frequency searching, really for "reporting" then it does not violate any normalization rules. Report tables are almost always denormalized. I've seen some real world examples where this approach could really add value for end user searches.

And a great twist on tier to tier data transmittal. Consolidating data at the front end tier into a bitmask for rapid transmittal, and then decomposition at the middle or data tier for storage. This has some potential in selected situations.
The twist is that the time to convert to bitmask (both directions) is faster than transmitting a large volume of data - which in today's environment would be an XML file for best efficency. I cannot immediately think of a use, but it is a new tool, that may come in handy when I least expect it.

As with everything in system design and the New Yankee Workshop, more tools are better. Just use the right tool for the right job!

Thanks again!

GregoryAJackson (5/22/2009)

---

I agree.

It's "cool", etc. but it doesnt seem practical.

GAJ

GregoryAJackson (5/22/2009)

---

I agree.
I'm not a fan of this in the world of RDMS's

It's "cool", etc. but it doesnt seem practical.

imagine writing queries and reports against the data, etc ?

GAJ