Add flush to many builders #8876
Conversation
| @@ -244,7 +245,7 @@ module YAML | |||
| end | |||
|
|
|||
| # Writes YAML into the given `IO`. A `YAML::Builder` is yielded to the block. | |||
| def self.build(io : IO) | |||
| def self.build(io : IO) : Nil | |||
There was a problem hiding this comment.
No, because YAML::Builder already does that.
There was a problem hiding this comment.
Why does YAML::Builder flush but XML::Builder doesn't? Shouldn't we make this behave identical?
JSON::Builder#flush delegates to io.flush which is another different behaviour.
There was a problem hiding this comment.
I noticed that too. I didn't want to start making changes all over the place. That's why I added flush when needed.
Also, maybe this PR is not needed...
| @@ -49,9 +49,10 @@ class YAML::Builder | |||
|
|
|||
| # Creates a `YAML::Builder` that will write to the given `IO`, | |||
| # invokes the block and closes the builder. | |||
| def self.new(io : IO) | |||
| def self.new(io : IO) : Nil | |||
There was a problem hiding this comment.
General question: Should this method be named build? It's not a constructor and doesn't return an instance of YAML::Builder which I'd usually expect from a .new method.
There was a problem hiding this comment.
Yeah, I was surprised by this name too. We could probably rename it, but I'd like to avoid doing that in this PR.
|
I love the branch name, btw 🤣 |
Possible regression from crystal-lang/crystal#8876
|
Hmm, I believe I've found out one possible regression: calling |
|
Related #8712. |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
Fixes #8875
Sorry that there are no specs but it's a bit cumbersome to test. Plus it's really easy to see what's going on.