release blog post for Rust 1.46.0 #675
Conversation
|
I think it's weird how the blog post is labeling this release as a fairly minor release despite const fn being insanely powerful now while before you couldn't do any control flow at all. If you compare this to the 1.45 release blog post, it labels the fairly minor soundness fixes to floating point casts (which affected close to no one) as big changes. Seems a bit off to me. Though I guess it's a smallish release in terms of new std APIs. |
| } | ||
| ``` | ||
|
|
||
| That's it! |
There was a problem hiding this comment.
do you think we should also reiterate how this changes the output? or is that too much?
There was a problem hiding this comment.
I think we should show only the changed output, rather than the code sample. There doesn't appear to be anything different about how this annotation is applied compared to existing annotations, so having a multi-line code diff seems mostly just confusing.
There was a problem hiding this comment.
...though an example that uses panic::Location might be helpful.
There was a problem hiding this comment.
I wanted to emphasize how simple this is to use, but if you wanted to come up with an example using Location, I'm not inherently opposed. I didn't show how the output was changed because we did that when it was added to unwrap.
Co-authored-by: Joshua Nelson <[email protected]>
|
@CryZe maybe I just don't appreciate the magnitude of the change; you're right that that's a big step up, I guess. I'll pump it up a bit :) |
|
@steveklabnik if you're looking for an example of what the new const fn capabilities allow, we really needed this for WinRT-rs to computer type ids of generic types at compile time instead of at runtime. This led to the creation of const-sha1 for const computation of sha1 hashes. Computing something like sha1 in const functions was not possible in const functions until this release. This in turn led to a huge perf boost in WinRT-rs crate. |
| } | ||
| ``` | ||
|
|
||
| That's it! |
There was a problem hiding this comment.
I think we should show only the changed output, rather than the code sample. There doesn't appear to be anything different about how this annotation is applied compared to existing annotations, so having a multi-line code diff seems mostly just confusing.
| } | ||
| ``` | ||
|
|
||
| That's it! |
There was a problem hiding this comment.
...though an example that uses panic::Location might be helpful.
Co-authored-by: Kyle J Strand <[email protected]>
Co-authored-by: Kyle J Strand <[email protected]>
Co-authored-by: Kyle J Strand <[email protected]>
Co-authored-by: Kyle J Strand <[email protected]>
Co-authored-by: Eric Huss <[email protected]>
Co-authored-by: Camelid <[email protected]>
|
Hey folks! thanks for all the feedback; especially around the The only major thing left is the track-caller stuff; I left my justification in the comments, but if @rust-lang/release prefers something else, please just make the changes; I don't feel strongly enough to actually block things. |
r? @rust-lang/core @rust-lang/release