D and AI / dub registry and AI

[monkyyy]

On Sunday, 22 February 2026 at 23:58:47 UTC, Julian Fondren wrote:
> On Sunday, 22 February 2026 at 23:37:52 UTC, monkyyy wrote:
>> Maybe atila makes this kind of mistakes, but I think the ai is 
>> failing here; from hyper-safetyism there better criticisms to 
>> make.
>
> Avoiding trivial string injection is level zero safetyism. It'd 
> be like tying your shoes when going rock climbing, instead of 
> not tying them and inviting a tripping hazard for no reason, 
> except that Phobos has made the two options equally convenient, 
> so it's more like you're given a menu of how you'd like your 
> rock climbing adventure to go, and you press one button out of 
> habit instead of the other:
>
> Shoelace setting:
> [ ] maybe step on your own laces and lose your footing and die
> [ ] avoid that (NEW!!!!)
>
> Well, feedback that you dismiss from a human is even more 
> easily dismissed when it comes from an AI.

You have the entire 71 lines of context; its not one line in the 
void, just factually the strings are in question all have some 
processing on them.

These airnt for safety, and I know exactly how my error handling 
preferences differs from the norms. But, mangling is mangling.

https://github.com/crazymonkyyy/Google-Fonts/blob/7aa2b3c0238d7c56f95546642cdb5601926c5ee9/source/googlefonts.d#L24
https://github.com/crazymonkyyy/Google-Fonts/blob/7aa2b3c0238d7c56f95546642cdb5601926c5ee9/source/googlefonts.d#L30

> `void main(string[] args) {exe(["curl", args[1], "--output", 
> args[2]]).writeln;` overwriting system files

is not a fair criticism. Im quite sure of this, even if I 
imagined this was where safety belongs(I really dont); a string 
without spaces and having a path prefix and a `.tff` suffix can 
only over ride `.tff` files and modifying system  fonts for... 
reasons? would take alot of `/../`

If you want this to have an extra safety trade off you could scan 
get any non-alpha around line 28 and return 
"findfont(fallbackfont)" before any curl is called