RFR: 8340114: Remove outdated SelectVersion() function from the launcher and update the code comments explaining the code flow [v6]

Jaikiran Pai jpai at openjdk.org
Tue Sep 17 11:12:06 UTC 2024


On Mon, 16 Sep 2024 13:55:49 GMT, Jaikiran Pai <jpai at openjdk.org> wrote:

>> Can I please get a review of this change which proposes to remove the (internal) `SelectVersion()` function from the java launcher and also update the code comments in the launcher to match the current implementation?
>> 
>> As noted in https://bugs.openjdk.org/browse/JDK-8340114, the `SelectVersion()` function in the launcher used to be relevant when JRE selection was a feature. That feature has been removed since Java 9 https://bugs.openjdk.org/browse/JDK-8058407. The SelectVersion() in its current form isn't relevant anymore and can be removed.
>> 
>> While at it, it was noticed that the current "flowchart" code comments in the launcher which attempts to explain the flow in the launcher code are outdated. The commit in this PR updates those comments for macosx and unix implementation. The windows variant doesn't have a "flowchart", but it too deserves a high level comment explaining this flow. I haven't updated the windows variant in this PR because that does a few additional things, which I need to review and understand better. I plan to take that up in a future change.
>> 
>> An existing `test/jdk/tools/launcher/MultipleJRERemoved.java/MultipleJRERemoved` test had to be updated due to the changes in this PR. That test was launching `java` (once) with 3 unsupported JRE selection options and was expecting 3 error messages (one each for the unsupported option) for that single launch. With the change in this PR, we don't accumulate and throw all those 3 errors and instead we fail fast for any of these 3 unsupported JRE selection options. The fail fast implementation matches what we do with other similar unsupported options. The test had to be updated to not expect all 3 errors message in a single launch and instead expect to find one of those error messages. Given what this test is for, and the fact that JRE version selection options (rightly) continue to raise an error after this change, I think, an update to that test should be OK.
>> 
>> No new tests have been introduced in this PR and tier testing is currently in progress.
>
> Jaikiran Pai has updated the pull request incrementally with one additional commit since the last revision:
> 
>   comment improvements

Thank you for the review David.

> Okay I had expected (hoped for?) more simplifications but it seems a lot of the complexity of the launch procedure remains.

In general, I've several changes that are lined up for cleaning up the launcher code. Some deprecations, some clean up and some fixes. I plan to propose those changes in the coming days.

This current PR was a small isolated step towards getting the launcher to a state where we can get rid of jar manifest processing within the C code (there are a few additional things needed as a separate PR to help us reach there).

This code and the area is new to me, I welcome any additional inputs that you have in mind in updating/changing or cleaning up this launcher code.

> I think basically this just gets rid of the data model and mJRE stuff and moves the rest out of SelectVersion.

Now that you mention it, one part of the changes in this PR, moves the checks and error reporting for the `-version:`, `-jre-restrict-search` and `-jre-no-restrict-search` options to the `ParseArguments` function. I think we can just get rid of that code and let those options be considered as just another unknown option to the launcher. I can open a CSR linked to this current PR to formalize that change. That would also mean that the `MultipleJRERemoved` test can also be deleted.

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

PR Comment: https://git.openjdk.org/jdk/pull/20997#issuecomment-2355365863


More information about the core-libs-dev mailing list