NP questions take a long time to answer.
P questions take much less time.
The statement "P = NP" means that any NP question can be transformed into a P question. Answering the P question gives you an answer to the NP question.
So, if P = NP, then we can answer questions that we thought were slow, much more quickly than we would have thought possible.
Proving P = NP would (presumably) give you that method for transforming NP problems into P problems. So you would suddenly be able to figure things out easily that currently are hard and take a lot of computing power.
It's often assumed that P = NP would make cryptographic problems much easier to break, but apparently that's not really the case: http://world.std.com/~reinhold/p=np.txt
(Though, as you say, there could still be an O(n) algorithm - this is not known).
What I said is true even for linear programming (a P-complete problem) - it has polynomial and efficient algorithms - though the latter are actually not polynomial in the worst case. :)
Wikipedia should answer most of your questions regarding the consequences:
NP problems can be solved, it's just slow (think brute force method, running through all the possibilities).
http://en.wikipedia.org/wiki/P_versus_NP_problem#Polynomial-...
The first historical example of an important proof which was non-constructive was Hilbert's Basis Theorem in 1888, which solved a famous problem posed by Paul Gordon. It is now viewed as in some sense being the first salvo in the debate over constructivism in the 20th century. See http://en.wikipedia.org/wiki/David_Hilbert#The_finiteness_th... for more.
Could you explain what some of those consequences be? Does P = NP really need to be proved to achieve those practical consequences? Can't it just be assumed to be true and then see what the result is?