RFR: 8360037: Refactor ImageReader in preparation for Valhalla support [v13]
Alan Bateman
alanb at openjdk.org
Fri Jul 25 17:12:00 UTC 2025
On Fri, 25 Jul 2025 16:58:49 GMT, David Beaumont <duke at openjdk.org> wrote:
>> src/java.base/share/classes/jdk/internal/jimage/ImageReader.java line 138:
>>
>>> 136: * @param name a JRT file system name (path) of the form
>>> 137: * {@code "/modules/<module>/...} or {@code "/packages/<package>/...}.
>>> 138: * @return a node representing a resource, directory or symbolic link.
>>
>> This is the jimage reader rather than the jrt file system so it's very confusing to see comments about "JRT file system" in this file.
>
> This class is *almost* only here for building JrtFileSystem. Everything else is marginal usage. Hence, admitting that in these docs felt reasonable to me. I think it would be bad if someone reading the docs in this class didn't know that there's a strong reason for the namespace unification. I can reframe it to talk-around the fact this is all ultimately for file-system like semantics without naming JrtFileSystem, but they are extremely closely coupled in terms of usage.
It should be possible to move jrtfs to its own module, say jdk.jrtfs. It does not need to be in the base module. My comments are only to ensure that we keep the architectural boundaries clean.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/26054#discussion_r2231635708
More information about the core-libs-dev
mailing list