I see what you're saying, and again: it's missing the entire baseline and context for that claim. My thesis was that if he had known/considered/used the actual std::list design (which I assume he hadn't, or he wouldn't have proposed nodes that owned siblings), he would not have considered linked lists to be tricky in non-GC'd languages according to whatever his baseline is for that (presumably, tricky enough to blog about it). But somehow you simply extracted my reply with that one quote I replied to, discarded all the rest of the context (his blog post and my comment and all), lowered the baseline for "tricky" (from ≈"tricky enough for him to call it 'tricky' and blog about it" to ≈"possible for the average programmer to write an initially-buggy implementation thereof"), and then ran away with this beautiful straw man to refute. =P Except that (obviously, I had thought) wasn't my claim in the first place...
Re:Why writing a linked list in safe Rust is so damned hard
by mehrdadn on Friday February 23, 2018 @12:24AM
(#16444424)
> No, I am disagreeing with your explicit disagreement with "implementing a data structure like this in a non-garbage collected language without Rust is also quite tricky". [...]
Now that I reread for a third time, I suspect you may've picked up on the "a data structure like this" to mean "a linked list with this exact ownership arrangement", whereas I interpreted it as the looser "a data structure like a linked list". That puts the rest of your comments into more context, and sure, I grant you that it's a suboptimal choice, but I don't think it's what the author meant.
In any case, I think I was rather assuming too much Rust context: taking your suggestion does not resolve the trickiness in Rust (as in, you won't be able to convince the compiler to accept it without using `unsafe` or reference counting). It's a very common meme about Rust that there's no way to write a safe C++-stye linked list with pointers, something independent of the choice of ownership scheme. "Fixing" the design doesn't actually help: it's still way too much of a pointer soup.