today #370045 has been filed against the (in)famous netselect. The postinst was not setting the setuid bit even the user said otherwise, after a bit of investigation it turned out that chown (or better, the kernel) clears the setuid bit as a security measure as explained by Mike Stone.
Nice security measure, oddly enough I've never noticed that effect (that is, netselect's postinst was correctly working for me) so it must be new or I was being hitted by some bug or it's been there for years and I'm the last one to know. Or all the previous options. Anyhow, netselect -8 is on its way with two postinst lines swapped.