Interesting feedback in this thread. Thanks for the posts.
I know Construct isn't perfect, and there are some serious problems that may be causing people problems with large projects. We hope to get the serious things fixed and release 1.0. But we actually wanted to shift focus and concentrate on Construct 2 soon. I can't guarantee that all problems will be fixed in 0.x and you may have to rely on workarounds. The codebase in 0.x is not fantastic and it can be very difficult to work with. These kinds of problems are part of the motivation behind C2 - it's going to be completely redesigned, architected for reliability and flexibility, and as a side effect things like 30 second saves can be avoided as well (which is difficult to fix with the binary .cap file format in 0.x).
I know this may be frustrating to many people who have invested so much in to Construct 0.x, but it does have serious architectural limitations. I doubt it can be taken much further than it is now, bar attempting fixes for the more serious problems. And C2 like the project already will be a spare time project. I hate to use this as a crutch, but we've never been able to guarantee fixes or support because this is essentially a hobby project. Some people watch TV, we write code. So I'm starting to see Construct 1 as a kind of a very big, complicated proof of concept, and C2 as the real thing getting going - we're much better at programming now, and some of us have professional experience to add to that.
So can you make a major game? I think you can, if you try to keep things simple - but I can't guarantee an easy time, unfortunately. That's not saying things will go completely smoothly with other tools either though!