ByteBuffer: improve and simplify get* implementations #896
Conversation
weissi
added
🔼 needs-minor-version-bump
| return nil | ||
| } | ||
|
|
||
| return ByteBufferView(buffer: self, range: index0 ..< index0+length) | ||
| return ByteBufferView(buffer: self, range: index ..< (index + length)) |
There was a problem hiding this comment.
Can we not use the range-generation function here?
There was a problem hiding this comment.
@Lukasa ByteBufferView uses (and has to use as Collection) absolute ByteBuffer indices so here, we're not interested in range relative to readable bytes but range relative to the whole ByteBuffer
There was a problem hiding this comment.
Ah, yeah, I see that.
This represents a difficulty with the requirement that ByteBufferViews only index readable bytes: specifically, what is readable can change after the ByteBufferView is created, without invalidating the ByteBufferView itself. I don't think this particularly matters much, as this will only happen if you're going outside the standard NIO usage patterns, but it's worth being aware of.
There was a problem hiding this comment.
@Lukasa no, ByteBufferView holds a (private) ByteBuffer which is a value, so the readable bytes can't change.
Motivation: For historic reasons, the get* implementations checked the same indices repeatedly and the checks were also repetitive. Modifications: unify and de-deplicate get* range checks Result: fixes apple#884
Motivation:
For historic reasons, the get* implementations checked the same indices
repeatedly and the checks were also repetitive.
Modifications:
unify and de-deplicate get* range checks
Result: