Daniel Alley
Daniel Alley
There's just too much noise with github actions to get good results. It looks like there are swings of up to 13% in the test run. The criteron book recommends...
>As I explained in https://github.com/tafia/quick-xml/issues/638#issuecomment-1693360806, I would not merge it right now, because new dependency increases MSRV and does not give any significant improvement, but you a free to continue...
> Also, investigating MSRV of our primary users also would be a good indicator whether it makes sense to increase it in near future or not. `serde_derive` has an MSRV...
@Mingun As noted in my previous comment I think for most of our users it is possible that our practical MSRV is in fact already 1.56 due to `serde_derive`
So the MSRV is bumped. Are we yes or no on this PR? (if yes, it needs a rebase)
Inlining is also a bit of a tricky form of optimization in general. If you bloat the code size too much, larger programs will get slower rather than faster. Debugging...
And reformat, and fix the new fuzzing failure
About the original purpose - are you serializing the XML out to a buffer first? If so, wouldn't it be easier to just get the length of the data in...
I didn't think of `io::Cursor` specifically but that's kind of what I was wondering with my previous comment. But you're right, `io::Cursor` does do [something very similar](https://doc.rust-lang.org/src/std/io/cursor.rs.html#322). I was thinking...
No worries at all! Thanks for the time & discussion.