3685: zio duration uses java.time.Duration #4034
Conversation
There was a problem hiding this comment.
This looks really good! Great to see so few changes in the rest of the code base.
I think at this point there are only a few user facing differences:
- Because
Finiteis not a subtype ofDurationanymore but just an extractor, when pattern matching users need to docase Finite(duration) => ???rather thanduration: Finite => ???. I think this is unavoidable with the new encoding and relatively minor. - Because
Durationis no longer our data type, we have to add methods like>with extension methods. That means if you just have aDurationinstance you can't call>on it without doing something to bring the implicit syntax into scope. So right now the user needs to doimport zio.duration._. We could potentially either move this into the ZIO package object so it is in scope withimport zio._or use the trick where we give the implicit conversion the name of the type (implicit def Duration(duration: Duration): DurationOps = new DurationOps) so it is in scope when the user doesimport zio.Durationbut unfortunately I don't think any of these are ideal as a lot of times you will just have a Duration value from some combinator you are working with.
One other thing that could be good to do here is define an Ordering[Duration]. We had Duration extend Ordered[Duration] before but now we can't do that since it isn't our own data type.
|
|
||
| def apply(amount: Long, unit: TimeUnit): Duration = fromNanos(unit.toNanos(amount)) | ||
|
|
||
| def fromNanos(nanos: Long): Duration = nanos.nanos |
There was a problem hiding this comment.
| def fromNanos(nanos: Long): Duration = nanos.nanos | |
| def fromNanos(nanos: Long): Duration = Finite(nanos) |
There was a problem hiding this comment.
I did not use Finite because if we pass a Long.MaxValue it will result in an Infinite.
There was a problem hiding this comment.
Also all constructors (.nanos, .millis, ...) build Duration.Zero if given negative values.
Co-authored-by: Adam Fraser <[email protected]>
Co-authored-by: Adam Fraser <[email protected]>
Co-authored-by: Adam Fraser <[email protected]>
Co-authored-by: Adam Fraser <[email protected]>
Co-authored-by: Adam Fraser <[email protected]>
Co-authored-by: Adam Fraser <[email protected]>
|
We could consider using a super lightweight "newtype" encoding to present our Albeit, I think it's common to do |
|
@adrianfilip Can you resolve conflicts so we can merge? |
|
Yes. |
|
Done |
|
💪 |
Used the suggestions from:
There were minor test changes where tests relied on types.
Also because duration is now a java.time.Duration Duration.fromNanos(Long.MaxValue) will equal Duration.Infinity.