Tracking issue for `clone_from_slice` stabilization

[aturon]

The clone_from_slice method clones a slice into another, using the clone_from method, which in principle can be more efficient than using clone.

It's not clear whether this method, or clone_from, are seeing any use at all.

[bluss]

This method is our one API that offers "copy from x to y", i.e. a kind of copy_nonoverlapping, although so general it handles everything that's Clone. This is requested semi-often. The need is relatively larger since your .zip() codegen is nonoptimal, so writing the loop manually ends up worse.

Using clone_from() vs clone() is a paranthesis.

[gsingh93]

I use this method. I have a large static array (a file header) and a slice (the bytes of the file), and it seems like this is the only way to copy the bytes from the slice to the array without looping.

[bluss]

@gsingh93 Yeah, there is one more way, see How to “memcpy” bytes in stable rust

[gsingh93]

The subteam report this week said this feature was in the stabilization phase, so I'm nominating it for the next beta.

[aturon]

Nominating for discussion for 1.6. Discuss post.

cc @briansmith

[alexcrichton]

🔔 This issue is now entering its cycle-long final comment period for stabilization 🔔

The libs team was a little up in the air about this function, but we're willing to consider stabilizing it. Some thoughts we had were:

• The name is a little odd and there may perhaps be better alternatives. @sfackler notes that this is "the memcpy" but the name doesn't quite imply that.
• The style of the API is somewhat unique and not always found everywhere else, so it may be good to ensure it's nice and principled as well.

[briansmith]

I noticed that the rand::Rng uses the word fill to refer to changing the value of every existing member of a slice. Also note that #27744 is related to this in that it would make sense to have the function that replaces all the values in a collection with a slice's value to have a similar name to the function that appends the values of a slice to a collection. Note that there's already a function named extend that is a more general form of the function that appends a slice to a vector.

So, I propose that:

• clone_from_slice be renamed to fill_from_slice.
• clone_from_value (the new function I proposed in the discussion to replace set_memory) be named fill.
• push_all be renamed extend_from_slice.

[alexcrichton]

That seems like sound reasoning to me, I may go even further to see if we could drop the _from part of the method names and just have fill_slice plus extend_slice, a little less wordy to type!

[briansmith]

I may go even further to see if we could drop the _from part of the method names and just have fill_slice plus extend_slice, a little less wordy to type!

Without _from, it would be ambiguous which slice gets filled. I think the _from is useful for making that more clear. Also, seems consistent with the naming of clone_from.