Hello David and thanks for a great article on yet another good use of SQL CLR.

The only issue I have is with your statement that "median" is not a good case for a CLR aggregate as it might "require a substantial portion of the set to be held in memory". I disagree because it is possible to reduce the size of data in memory by compressing it, as I have shown in this article:

http://www.sqlservercentral.com/articles/SS2K5+-+CLR+Integration/3208/

And that article was done before SQL Server 2008 added the ability to go beyond the 8k memory limit (with a MaxSize set to -1) so adding that option to the mix solves all problems (generally) outside of the possible efficiency of the CPU it takes to compress/decompress the dataset. At that point I would argue that if your situation requires a Median calculation (and it appears to for quite a few people) then a little bit of extra processing time is worth the ability to do this in the first place.