On Wed, 20 Oct 2021 00:04:48 GMT, Jonathan Gibbons <[email protected]> wrote:
>> test/langtools/jdk/javadoc/lib/javadoc/tester/JavadocTester.java line 953:
>>
>>> 951: * @param file the file
>>> 952: */
>>> 953: public OutputChecker(String file) {
>>
>> Given JDK-8274172, consider using java.nio.file.Path instead of String here.
>> Using Path for files and String for target strings reduces potential for
>> confusion.
>
> I'll investigate, but IIRC there is broad precedent for using `String file`
> to simplify the call site, avoiding boilerplate `Path.of`.
I looked at this, including just adding a path-based overload.
In general, I am a big proponent of using more specific types instead of
strings, but the prevailing pattern in `JavadocTester` is to use strings for
"paths relative to the most recent output directory". Internally, `Path` is
used more specifically for paths that have been resolved relative to the output
directory, etc. so switching to `Path` in the public API would be less
convenient for use sites, and would lead to ambiguities that would need to be
addressed in the implementation.
TL:DR; this might be worth additional consideration down the road, but not as
part of this work.
-------------
PR: https://git.openjdk.java.net/jdk/pull/5743
