improve stringbuilder::output #1730
Conversation
Pull Request Test Coverage Report for Build 5532Details
💛 - Coveralls |
|
The improvement does not look obvious to me. Can you explain it? |
| pub impl Show for StringBuilder with output(self, logger) { | ||
| logger.write_string( | ||
| self.data.unsafe_to_bytes().to_unchecked_string(offset=0, length=self.len), |
There was a problem hiding this comment.
to_unchecked_string just simply copy bytes into string
There was a problem hiding this comment.
So basically you want avoid the extra copy here. My suggestion is that we should change the semantics of to_uncheck_string since Bytes has become immutable. Then we can still enjoy the batch processing in write_string.
There was a problem hiding this comment.
When using to_uncheck_string with backends that require extra copies, such as the js backend and wasm-gc backend with the js-string-builtin option turned on, use this implementation
| for i = 0; i < self.len; i = i + 2 { | ||
| let low = self.data[i].to_int() | ||
| let high = self.data[i + 1].to_int() | ||
| let code_unit = (high << 8) | low |> Char::from_int |
There was a problem hiding this comment.
this code_unit is the same as String::op_get 's result
There was a problem hiding this comment.
convert two bytes to code_unit, like convert char* into uint16_t* then use pointer read data.
No description provided.