To preface this, it is entirely my fault that I continue to attempt to use F350 in production. I can't blame the F350 team for that. If I had any sense, I would have long ago given up on it.
F350 is in some ways amazing - tantalizingly so. You can sit down with it having never used CAD and build some very impressive stuff very quickly and effortlessly‥ that's how they get you. It very quickly breaks down and becomes unusable for non-trivial designs that get past some level of complexity (which I can easily reach not in months or weeks or even days, but within a few hours of starting a new one).
It's mid-2018. F350 has been out of the lab for many years at this point.
On a good day, using F350 is like building a house of cards - things fall apart quickly and that makes me want to scream.
To this day I still can't use parametric design without it very quickly breaking. It demos well, but carefully use it in a real design, try to change a parameter, and symmetries are broken or parts are otherwise borked.
The timeline is a minefield - especially when there are linked designs included. Update one of those after awhile and F350 hangs when rolling back through timeline operations (which is particularly annoying when there are or at least should be no dependencies). You still have «delete» and «remove» and if some user commits the cardinal sin of hitting «delete» they are guaranteed to bring a design into some horrible state.
Consistently I get in this situation where I'm up against a deadline and I'm either having to restart from scratch or drive myself mad trying to beat the thing into submission. On my current project, I'm now pushing well into the point where I have to get manufacturing expedited which is every day adding hundreds to a thousand dollars and pretty soon I'll be past the point where I can get it done in time at any price.
I have no idea how things have stayed this bad for this long. I don't know if it's the PM or if it's an issue of the team or the management. I still suspect that a big part is that you have separate QA and dev teams and that you don't use naive testers so you're not exercising huge chunks of the code and you're missing interactions that are easy for people who don't go through a particular path to trigger. I suspect also that there are fundamental design and implementation issues with the code data structures and algorithms that keep you from even knowing whether some object depends on some other object ‥ and that's not even getting into how many features you've implemented to the point where they can be demoed but they're not fully realized, fleshed out, and debugged (which is honestly more or less all of the features).
I understand that I'm just screaming into the wind here but please, please, please, please, please fix the program.
