As someone who has been following this tutorial exactly, somewhere around video 21 (possibly due to rerecording?) Wes made a bunch of changes off-camera, failed to revert all of them before resuming, and following along now causes a bug in the SwapCharacter event severe enough to crash the editor when testing, which I assume is due to some variable not actually being set properly in some other function. And… I spent the last 3 days going back over videos with a fine toothed comb looking for what I might have missed, because it’s an official tutorial, and certainly looks like everything was recorded.
The last official tutorial I went through (drag and drop live training) had a similar problem, where he was working from the wrong file and following along requires one to skip ahead to a good chunk of part 3 to catch up.
I also have to second the complaint that once this gets into the real meat of the networking, it stops explaining the logic behind things and just turns into a “do what I do” sort of thing, which doesn’t really teach things well enough to debug this myself.
A few hiccups like this are basically inevitable with this sort of video tutorial, since we’re looking at several hours on a two monitor setup recreating an elaborate sample, explaining on the way, and trying to chop it all into manageable self-contained episodes, and fixing things on after the fact … basically means rerecording the whole thing (or restoring from carefully labeled archives of project states between videos) but… why are these even in a video format at all?
There’s a definite benefit to featuring video of tests between major steps, so people can very clearly see how things should look in action, and tips you’re more likely to remember if you’re keeping up a running narration, but outside of maybe a single introduction to the interface to demonstrate a few features of the GUI, throwing tutorials like these together as just web pages with composited screenshots of complete events and functions with explanatory paragraphs of what they do would be much easier to produce, follow along with, double back and debug, and most importantly, update down the road when errors are caught, or a new engine version changes the most efficient way to do something (or the look of a node).
At this point, I would really love, first off, either a guide to fixing some problems this tutorial left me, and/or a project download/pile of screenshots of how everything should look that I can compare what I have to, and I could also really stand a long-winded explanation of replication, server vs. client events, and… if I’m really honest, while it’s a bit off topic here, the functional nuances between player controllers, game modes, player states, game states, and level blueprints, vs. just dumping not-tied-to-particular-object code into whatever random blueprint. A lot of what-goes-where makes some sense from a basic organizational perspective, but I know I’m bound to break something eventually from how fuzzy I am on this point.