New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[SUREFIRE-1909] Replace --add-exports with --add-opens #461
base: master
Are you sure you want to change the base?
Conversation
Signed-off-by: Nils Renaud <renaud.nils@gmail.com>
Hi @NilsRenaud , I have approved the CI. Now it is running. If it would pass successfully, we should obtain a new distinct integration test from you related to your use case. Would you have some spare time to write one? Thx |
@NilsRenaud |
Pls name the title of the PR, Jira and commit same. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Lgtm
Thanks
@NilsRenaud |
@NilsRenaud |
Signed-off-by: Nils Renaud <renaud.nils@gmail.com>
Signed-off-by: Nils Renaud <renaud.nils@gmail.com>
@Tibor17 :
Yes, I've updated the PR with a new integration test using this new option (not enabled by default as advised by @sormuras
Oops. I saw you message a bit too late. But I kept these changes in a separate commit for now, feel free to comment my code with your advices
I'll squash my commits in a single one with a correct name after your reviews, OK ?
Nop, it's not related to multi modules, I've created a dedicated test with a single module tested by a non modularised tests
It does not work without the open keyword. Since the tests are running in a module, I'm not sure it uses the same underlying ForkConfiguration though, but I've not investigated. I've added an integration test and a parameter usable directly from the XML file in the last version of this PR. Let me know what you think about it. |
Why we should not stick only to the directive to
|
I tend to agree with you @Tibor17, but @sormuras said that this drop in replacement could "break other use cases". @sormuras, could you give us an example ? |
"Could" as in might. I have no concrete example at hand. My assumption was that it affects user test code and their expectations regarding modular boundaries. The proposed change replaces one directive with another directive. Is the new directive the one users want(ed)? |
@NilsRenaud
|
I'm sorry @Tibor17 , I'm not sure to understand your point. There is currently 2 strategies working with this PR :
Is that what you meant ? |
@NilsRenaud But you can additionally use the module descriptor in |
@NilsRenaud |
In this example, my feature has no impact as you can see in the arg file used to run the tests :
And playing with the example a bit further : if I remove the "open" in the test's As reminder, here is the output when tests are not modularised :
I don't think so, as stated above when the tests are modularised themselves we are using their module description and we're not interfering. In case of non-modularised tests the compiled code will be patched into the "main" module, then exported/opened. I don't think there is more control to offer to the user : they can reuse the package names or not, it's fine, everything will be packed together anyway. |
@NilsRenaud |
I don't see any use case where we would prefer I still think having a dedicated parameter instead of an hard coded value makes sense. |
@NilsRenaud |
PR created : #471 |
Let's keep this survivor open during next few versions. |
This PR follows the discussion on the associated Jira issue : https://issues.apache.org/jira/browse/SUREFIRE-1909