[haypo]
> No user complained past years.

Raymond said "We've previously had this problem with MT (since resolved, where it is was landed in a very non-random zone)."  Do you believe he was wrong?

> I don't think that we should worry so much, because it looks like
> reading more data from /dev/urandom can be a more serious and
> concrete issue.

Why do you say that?  Most links people have found and posted here clearly say that the Linux warnings about urandom are basically nonsense.  Please supply references to back up "serious and concrete" (or point to earlier references, if you found them convincing).

I'm with Raymond on this.  There is no useful theory that allows us to predict the characteristics of the produced sequences from a set of possible seeds, so limiting the set of possible seeds is potentially dangerous.  The theory about equidistribution (etc) is very useful, but relies on mathematical analysis of the _entire_ period of the generator.  The only way to span the entire period is to allow for all possible seeds.