RFR: 8341090: Remove support for security manager from JavaFX [v2]

[Kevin Rushforth]

On Fri, 18 Oct 2024 10:13:13 GMT, Johan Vos <jvos at openjdk.org> wrote:

> When you say "all tests pass", what set of tests are you referring to? We have a number of properties that determine which tests are actually executed, and I believe in general it would be good if we could standardize on what set of tests we consider crucial (e.g. not the ones marked as unstable).

Agreed. By "all tests", I meant a full headful test run except unstable tests, meaning that I ran `gradle test` with: `-PFULL_TEST=true -PUSE_ROBOT=true`.

Our typical headful test run is:


gradle --info sdk shims
gradle --info --continue -PTEST_ONLY=true -PFULL_TEST=true -PUSE_ROBOT=true test


I usually run it like that so that I get a fail fast on the compilation (i.e., without the `--continue`) and so that repeat runs, or runs of specific tests don't need to recompile the sdk and shims, but if all you want is a single full test run, you don't need to split (in which case, omit the `TEST_ONLY` flag).

> I'd like to run all tests as well on all supported platforms, but your explanation is very clear and makes sense, hence I don't expect issues.

That would be great. Let me know what you find.

I did a full test run using JDK 23 (our default boot JDK) on the following platforms:

Windows 11
Mac aarch64 (both macOS 13 and macOS 14)
Mac x64 (both macOS 13 and macOS 14)
Ubuntu 22.04

I also did the same full test run setting `JDK_HOME` to a test build of the [jdk-sandbox:jep486](https://github.com/openjdk/jdk-sandbox/tree/jep486) branch on the same platforms, excluding Mac x64 (I didn't have a JDK jep486 build handy on our test machine, and since there is nothing arch-specific about the SM removal, that seemed OK).

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

PR Comment: https://git.openjdk.org/jfx/pull/1595#issuecomment-2423231410