On Tue, 8 Oct 2024 12:46:42 GMT, Jaikiran Pai <j...@openjdk.org> wrote:
>>> The question is: Would that make any sense in the end?
>>
>> Consider the example of `StringBuffer`, which is a `CharSequence`. Wouldn't
>> something like the following be a logical use of `Reader.of(CharSequence)`,
>> where you create a `Reader` for the underlying sequence (which is not yet
>> populated) and then keep reading through the `Reader` until the underlying
>> sequence's data is finished?
>>
>>
>> final StringBuffer sb = new StringBuffer();
>> try (final Reader reader = Reader.of(sb)) {
>> final ContentGenerator cg = new ContentGenerator(sb);
>> Thread.ofPlatform().start(cg);
>> int numRead = 0;
>> while (numRead != -1) {
>> final char[] content = new char[1024];
>> numRead = reader.read(content); // wait for content
>> }
>> }
>>
>> private class ContentGenerator {
>> final StringBuffer sb;
>> ContentGenerator(StringBuffer sb) {
>> this.sb = sb;
>> }
>>
>> @Override
>> run() {
>> while (true) {
>> String foo = getSomeContent(); // maybe blocking
>> sb.append(foo);
>> }
>> }
>> }
>
>> I cannot image any scenario where such a program would result in useful
>> outcome.
>
> An additional point of reference is the current default implementation of
> CharSequence's `public default IntStream chars()`, which returns an IntStream
> of char values from the sequence and the implementation of which doesn't
> consider the `length()` to be fixed.
For `chars()` or `codePoints()`, I believe calling `length()` or not was a
matter of implementation convenience instead of the assumption that `length()`
can change during calls. Note implementation methods in the anonymous class in
`codePoints()` cache the length in local variables. Maybe they just don't want
the extra field overhead in the objects they construct.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/21371#discussion_r1791830179
