Yep, Marsu nailed it.
Nope... not even close.
When TTL flash metering was developed, back in the film days (also known as "the good old days"...) the meter read the amount of light being reflected off the film surface during the exposure, to determine when enough light had reached the film to provide the "correct" exposure.
When digital cameras were introduced, it was discovered that the surface of the imaging sensor was too reflective compared to film, rendering TTL flash metering unreliable. E-TTL was the workaround developed by Canon (and similar solutions were developed by other companies) to reliably (or at least, more consistently) determine flash metering from a suitable reference surface (usually the front of the shutter curtain) before exposure.
Well, both of you are right. Marsu's description of ETTL was correct. Your description of TTL is also correct. I wonder if sensor reflectiveness is the reason why Canon went down this route. In the days of film, I remember being frustrated that Olympus had an excellent TTL flash system and Canon never developed one for film. This frustrated me as a macro photographer, having to manually use flash with pre-calculated exposures depending on distance / extension. Were the patents for TTL so tight that Canon could not use it?
I would have thought that the reflectiveness of the digital sensor would have led to a technical workaround. Indeed, would you even need to measure light reflected off the sensor, rather than measure flash directly from the sensor? It seems to me that even if this were possible, that ETTL2, with the balancing of ambient and flash lighting, is a more elegant solution.