There is a good reason why no one has unseated excel: it's easy to build the really easy 80%, but the last 20% is incredibly difficult and time consuming.
It's a shame Scheme in a grid appears to be abandoned (if anyone manages to build it under, say, the latest Debian Stable, please let me know):
Most spreadsheets I've dealt with don't hold relational data. Are there people who use them as such? Sure. But I don't think that's the typical use case.
I would say relational data is the dominant use case for spreadsheets. It's a natural fit given the rows and columns. What other kind of data do you see in spreadsheets? Hierarchical data is not a great fit for spreadsheets.
Ironically, Excel fails miserably at implementing the easy parts of a database.
Seriously, how much does doing a simple "Select Distinct" suck in Excel?
So much that I wrote this: http://scpike.com/uniqify/
(I know you can do filter => unique only => copy somewhere) but it's stupidly slow.
It's really easy to implement the easy parts of spreadsheets, getting the hard parts right is why Excel continues to dominate the space.