Commercial Apps For Consumer Linux: D.O.A.? (InformationWeek)

Speaking as a developer employed to write a commercial application, I'd say that even if they
don't sell very many copies on Linux, porting the application will propably still be
worthwhile, since the porting effort itself is likely to expose bugs they didn't know about -
and getting those fixed makes the Linux port worth it all alone.  
We recently ported our app to Linux as an in-house project, and just compiling it with G++ (we
use the Intel C++ compiler on Win32) revealed a lot of bugs. Also, getting access to tools
like Valgrind on Linux revealed a few more bugs. Just running it on Linux exposed some areas
of the code that didn't perform well generally (they just happened to run acceptably on
Win32).
The point is, just by changing platform and compiler we found lots of bugs - once we fixed
those bugs we could provide a much better product to our Win32 customers. So even if we never
sell a single Linux license, the Linux port has still paid off massively (and since we try to
write cross-platform code from the get-go, it didn't take more than a few days ;).