The hide errors function has actually a purpose - on a finished product one might want to hide error message and instead replace them with their own error handling system
When I was thinking how to write what I wrote that I already thought
someone would come up with that. And that is exactly why I included that "local error handling" paragraph.
The problem is that that "don't show errors" mode is often engaged quite early in the development-phase (instead of allmost at the end), resulting in the earlier-mentioned messs.
I was even considering to suggest a "reset to default at loading" feature (forcing the person creating the game to re-evaluate any errors that might pop-up), but this was about
removing features, not a request for more stuff to implement.

I don't think it's a good idea to recommend a deprecated (and undocumented) function
I do not either. But I can't help but wonder : why should that command go ? It has no negative side-effects or is just plain ugly. And yes, that made me mischievous and play a bit of the advocate of the devil.

To be honest, I have not used it for ... wel, directly after I switched to v6.x I guess.
Although its funny to see that this "should
go" command has been depricated but survived 3 versions and about 6 years now.
In this regard (using undocumented commands ) I must say I was a bit baffeled to see the "action_...()" commands prominently displayed in GM v8.0s "show build-in functions".

Maybe they should be removed there (mike, looking at you).

Registry - WILL go. It's a massive security hole that needs nuked as quickly as possible.
<snip star-treks facepalm picture>
I agree with you here. But when its put that way ("WILL go") I know when to shut up.
And luckily we have some capable programmers here which will probably replace that killed functionality with a simple DLL, eventually embedded in an extension-package. In short: you probably won't even notice it was gone.

(And yes mike, this is a (somewhat) silent protest, as well as showing you the futility of your decision).