It's interesting, because the issue here is not it being impossible, the issue here is communication.
What the engineer is doing wrong is he's saying things like "it won't work because <reason>", instead of trying to wrap his head around what the customer really wants by saying something like "Splendid idea, I am capable of implementing it. Could you demonstrate to me how you expect the end result to look like?".
The customer obviously doesn't exactly know the implications of her request down to the last detail. The customer and the engineer speak two entirely different languages, and it is the engineer's job to translate the customer's language into their own language without rejecting the customer's ideas.
Actually I think the expert grasps the intention of the client in the end (at least as far as that is possible),
but is faced with a business decision: They can either sell the simple solution (two perpendicular red lines) for a small amount of money or they can sell the same solution but claim it was more difficult (seven perpendicular lines, two of them red, five of them transparent) for presumably more money. So the question is: By how much do you screw your client over?
If the "client" is another department of your company then this is an obvious choice. Sell them the two lines and explain to them, why this gives the same result as 2 lines + 5 transparent ones.
If the "client" is another company, then this can only be decided by somebody who knows the dynamics involved: How high can we put the price before we loose clients to the competition, etc. and this is s.th. that the engineer can't and shouldn't decide, at least not alone.
My "favorite" exchange so far was something like this: "We need you to add this functionality." "That's technically impossible." "Well, you better make it possible, because we already sold it to the customer."
A couple of weeks back at CeBit I was on the other side of that, talking to a guy who tried to sell his product to a friend. It took way more persuation than it should have, before he fetched his engineer who then verified that their product wasn't capable of what the friend had asked for.
I think this (selling s.th. that doesn't exist) is the bigger and more common problem and it can only be prevented by placing the "expert" in those meetings, even if that results in communication problems from time to time. The confusion that arises from it is probably the lesser evil.