Yeah, I probably shouldn't have bothered if I wasn't going to put any effort into it. But I didn't want to rant. Hey, let's rant!
First, $400/$200 on sale doesn't buy you a click-and-export mobile solution. The initial setup instructions have been improved since I started from "very bad" to "bad," but still require a cross-reference to another page of instructions, including finding and installing an archived version of the Java JDK. As we are all more-or-less bright and clever monkeys, this isn't a huge hurdle, but it's silly.
What's more problematic is that YYG doesn't provide basic functionality in a timely manner. On iOS, GMS2 currently returns the incorrect resolution for the iPad Pro, see bug #0030436, meaning a person can't use the full screen. On Android, as I recently wrote in another thread, GMS2 doesn't support Android's built-in function for drawing to the portion of the screen next to a notch/hole punch/etc. The IAP for Android, Google Licensing, Google Analytics, MoPub, and Facebook haven't been updated since 2016 (or wait, maybe FB was finally updated last week?); AdColony since February 2017; and Google Play Services since May 2018. It took until that February 2017 update to add support for rewarded ads---which is completely nuts. We lagged behind adaptive icons for an entire cycle, and instead had a
hacky workaround. We're currently wondering whether Google's removal of support for Google+ is going to affect the out-of-date Google Services extension. And I'm just an amateur! Goodness knows what people depending on the Android support professionally are dealing with.
(Maybe widespread random crashes that nobody can figure out.)
More generally, I can't understand why anyone would pay money for and rely on GMS2 at all. I dug myself into a healthy sunk-cost fallacy because GML is genuinely easy to learn. But now, two years down the road, I've had my fill of all the shortcomings that overwhelm its ease of learning. Nesting data structures takes backflips--what the heck? Handling data is the most basic, ground-level existential purpose of programming, and GML kind of sucks at it. Every object has a slew of useless built-in variables. Functionality is sprayed across the room editor, code stashed inside the room editor, the object-node editor, and a lengthy hodgepodge of events. It's common advice to avoid much of the built-in "game engine" stuff---built-in collision handling, the "game restart" function, screen-wrapping. A new user doesn't know what to avoid and what to use. Add to that a bizarre two-prong compiling system: VM if you don't care about speed and want to be sloppy, YYC if you want speed but are willing to adhere to an unwritten set of semantic requirements that nobody entirely understands. If you've got the mobile export, you can't test for YYC using the test compiler so you have to wait for a full compile to a device, which, given the sloth-like compile time, gets to be a huge drag after the first forty or fifty times. Now I'm learning C#, and it's just... smarter, more rational, built for people who think firstly in terms of information flow. GML
feels like a hacky language that grew from drag-and-drop visual-programming roots.
So, all that would apply even if it were free. But it's not, it's really expensive. And there's nothing that GMS2 provides over and above [popular open-source 2D engine] or [popular free-with-logo 2D/3D engine] that cost $0.