If I had a penny for every pixel … I’d have around $1088. Which I would take, but really it’s not enough.
just edited to upscale the image
I don’t think I’ve ever explicitly gone four deep. Two is common enough, and three happens on some rare occasions, but four seems like sheer madness.
Dumb question, but when would you need two deep? Is it when you store a pointer as a field in a struct?
If so, isn’t that a massive footgun, because the pointer might go invalid at any point? 🫠
Pointers to arrays or arrays of pointers are common examples.
Your pointers won’t just magically become invalid. You gotta fuck 'em up first.
Ah, you mean “pointers to arrays”, because arrays are themselves just pointers in C/C++. That one still feels like it shouldn’t be needed in practice, because you already got a pointer, why can’t you use that directly? But yeah, I have no practical experience with C/C++.
And you do gotta fuck 'em up, as in
freewhat they’re pointing to before youfreethe struct/array containing the pointers.
But when you do stick them into a struct/array, that often means you want to move them out of the scope with themallocand potentially store them, too.
At that point, surely, it becomes rather difficult for your whole team to know or track down when it’s legal tofreethat.I only know from Rust that if you want to store a pointer/reference in a struct, it makes you specify the lifetime of that struct, which has to be greater or equal to the lifetime of the thing the pointer is pointing to. Hairy stuff. We’ve basically told the Rust newbies on our team to just not store pointers in structs and that’s working rather alright.
Arrays may be implemented as pointers on C, but the distinction is on how they are used, which is why I used the verbiage I did.
What if you need to modify a reference to a pointer, e.g. change the value of a value referencing a certain place in an array? strtol(), for example, uses a pointer to a pointer to a char to indicate the end of the parsed portion of the input string.
Major codebases performing high-level operations on data that’s shared in barely trackable scopes certainly aren’t best implemented in C. It’s still the language of choice for low-level code, especially on embedded systems, where allocations are not taken lightly.
Once your pointer definition looks like a censored swear word you’re doing something awful.
In my entire programming career I’ve used
int **less than a handful of times and I’ve always been borderline about refactoring when I need it.Okay, but what if you’re dealing with a rather high-dimensional tensor? In some kinds of coding it can happen, and you usually don’t want to sacrifice performance when it does.
You can also do increasingly elaborate pointer arithmetic, but that seems worse, not better to me.
*char // I heard it from a friend **char //who heard it from a friend ***char // who heard it from another "You were messing around"Me: building a fluent interface framework…
I already support aWrapperOf<T, T, T, T>
User: Can I have aWrapperOf<T, T, T, T, T>because I’m doing something weird?
Me: *sigh* god-damnit. You’re right but I still hate it.
int
I am a friend.
Is there a code example of this?
I’ve been a four-star programmer a few times. Imagine a blocked symmetric matrix where the rows and columns are indexed by triples (u,v,w). The entries are zero whenever u != u’ or v != v’, and because of symmetry you only store entries with w <= w’. But the range of v depends on the value of u and the range of w on the value of v. So you do
double ****mat = calloc (UMAX, sizeof(*mat)); for (int u = 0; u < UMAX; ++u) { mat[u] = calloc (u + 1, sizeof(**mat)); for (int v = 0; v <= u; ++v) { mat[u][v] = calloc (v + 1, sizeof(***mat)); for (int w = 0; w <= v; ++w) { mat[u][v][w] = calloc (w + 1, sizeof(****mat)); for (int ww = 0; ww <= w; ++ww) mat[u][v][w][ww] = some_function (u, v, w, ww); } } }and weep a little. In reality, this gets a bit optimized by allocating a single chunk of memory and carving that up into the pointer and data arrays, so everything is reasonably close together in memory.
uwu, got it.
My brain hurts
Isn’t that like, how you declare different dimensionally sized arrays? If you don’t care about memory integrity?
It’s been literally decades since I had to deal with code like that, so I may have jumped my stack.
If you do it with fixed-size arrays you can accomplish multi-dimension with just int*. Lots of pointer arithmetic needed though. Probably still faster than n levels of indirection.
I know a friend who can point you to a region in memory where you can insert your exploit. You’re welcome.
