edit: is about 1:30 a pretty common length you shoot for or did it just turn out that way? Any comments on lengths of loops?
In my experience two things most often impact song length: tech limitations and budget constraints. A third factor can be context. In the case of the Mechlab Theme, I think all three played a part. Most of the time clients want to save the big, more complex cues for more meaningful parts of the game. The menu screens are often viewed as lower priority or filler. Many clients want something to fill that space, fit that mood/visual but don't want to give heavy budget and tech resources to that part of the game. They'll save that for pivotal points in the game or climatic endings/etc.
Of course it's a case by case scenario, but for mobile games often 60-90 second loops work best. Some games may have a longer cue if that was the only music for the entire game. One of the first things I discuss with clients is the overall arch of the game and how they want to spread out the music cues.
So, how do you go about composing and getting the parts into the editor? I've realized I don't have the keyboard skills to use a MIDI controller on some of the parts I write, so I score out the MIDI in a separate program (Guitar Pro) and then export it into Reason. Do you play your parts, or is there a way to edit them in the programs that you use?
I play them all in because that method feels the most organic and effective to me. Being a pianist, it's easier for me to simply play the ideas directly into the session instead of using the ol' mouse point-n-click method. If needed, I'll open up the notation or piano roll view to see what I've played while creating new layers but most of the time I just rely on my ears. Then if any editing is needed, mainly related to velocity and modulation settings, I'll do that in the piano roll view.