Recently physicists have managed to build "attosecond lasers", lasers which emit pulses 10^-18 seconds long, interrupted by much longer periods of darkness (at least 10^-14 seconds). Before them, lasers emitting femtosecond (10^-15 seconds) pulses have been around. Assuming they produce visible light, what colour is it?

I took a look at the posted answer and I see how it is valid if the blanking time between pulses is infinity (i.e. a single laser pulse).

However, if the blanking time is only 10^-14 s, I don't see why the spectrum cannot be discrete.  Please see my solution.  If the signal repeats then the frequency spectrum doesn't need to be continuous.  In fact, for any finite repeating blanking time, the spectrum appears to be discrete.

If the blanking time is arbitrarily long then I agree that the spectrum is that of the sinc function cited in the website.  In fact, in a lab, I would think you would capture a single pulse only.

However, if you capture a train of repeating pulses (at 10^-14 s for example) then I don't see how the spectrum is not discrete over the visible as in my posted answer.  I am just trying to clarify and do not intend to question the validity of the solution.