You have to explicitly assume that a>0,b>0, c>0 and that they are the sides of a triangle.  That last bit implies a relation that is easy to write down for a right triangle, but involves another quantity (the cosine of an angle) otherwise.  So you are right, but so is Maple!  The issue is that you have more knowledge, from which what you say can be derived.  Without it, there are other cases that need to be accounted for, which is what Maple does [and is the right thing to do].

This is one lesson that most users of computerized mathematics [including myself!] need to learn multiple times before it really sinks in: you need to tell your system all your assumptions, otherwise how else is it going to know what you're talking about?  If you listen to yourself explain your mathematical problem (verbally or in writing like you did above) and compare that to what you actually typed in Maple, you'll notice a non-trivial gap.  That gap results in the answers you get (sometimes!) being other than what you expect.