kinda depressing

[Raoul Duke]

done some UE5 messing around. a poor video of some of it. the first ~8
seconds are blank, sorry, and there's no much actual game there ...
https://1drv.ms/v/s!AodgS59KO3aCqF1x-N1pVIxl2g4D?e=Op19dZ

turns out that afaict trying to start all over again to clean up and
reboot and remake something that you already have in unreal is kind of
really hard. the graph of relationships of actors you've made and
blueprints you've created and all the interactions thereof make it
hard to migrate smaller subsets of things over to a new clean project
a little bit at a time.

i've also heard that game devs tend to never ever try to migrate a
project to a newer unreal, too risky, often doesn't work, or has bugs
you only find later, etc.

the underlying architecture of all this seems so bad. a big ball of
mud object graph tar baby nightmare. i don't expect unity to be any
different, from what little i've done in unity i feel like they have
the same problem of having a giant object model with inter-references
and probably no good tools ui ux to manage and migrate all that stuff.

no way to really do much software engineering on it. dogs playing chess.

(besides all the badness that exists just while trying to slap
together a game in those paradigms in the first place, so full of
horribleness.)

this is of course true of many things in software development. games
have a lot of dynamic state so it seems like they are nigh worst case
situations. i really wish somebody would come up with a better
paradigm and make a great affordable cross platform game engine.