RFR: 8358533: Improve performance of java.io.Reader.readAllLines [v3]

[Brian Burkhalter]

On Mon, 23 Jun 2025 21:16:30 GMT, Stuart Marks <smarks at openjdk.org> wrote:

>> `Scanner` seems to scan for even more characters: https://github.com/openjdk/jdk/blob/c4fb00a7be51c7a05a29d3d57d787feb5c698ddf/src/java.base/share/classes/java/util/Scanner.java#L490
>> 
>> Would it make sense to resemble this? Would it make sense to simply use `Scanner` directly? 🤔
>
> The `readAllLines` method has a specification of line terminators that agrees with that of `BufferedReader::readLine` and `String::lines` and so we don't want to change it to be different.
> 
> Unfortunately `Scanner` doesn't seem to have a specification of what it considers to be a line. Also unfortunately, its notion of line separators isn't the same as the regex pattern `\R`.

> It might be worth adding a test of unconventional sequences or \r and \n, including \r\r and \n\n, \r.

I think that the test changes in [d5abfa4](https://github.com/openjdk/jdk/pull/25863/commits/d5abfa450cb3fcd604560833038735e41952bce9
) cover this.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/25863#discussion_r2164681227