The customer was really persistant that we should not use ANY locks when reading from SQL server, not even Sch-S locks, because "noone else is using locks".

After two days of trying to explain to them the concept of "Dirty Reads" and the practical imposibility to avoid Sch-S locks, they finally gave up.

The best part was when they asked in a quite condescending way "this is funny, why do you think that nolocks causes dirty reads?" and I sent them a link to the MSDN page about nolock that cleary states "Specifies that dirty reads are allowed.".