QA (public) » openQA Project (public) » openQA Tests (public) » qe-yam

You got the idea, yes, I would prefer to perform the installation and check the partitioning on it post installation that corresponds with the expected default.
I'm not really in favor of changing to command line to do some actions like to check that the encryptioned was opened, checking this via the planned deleted partitioning should be sufficient.

First step is done, successfully handled the dialog to decrypt the device or skip the decryption. VRs on my local are succesful so far.
No the difficult part: How to check the resulting action....

glad you are on track, you can send a draft PR so we can pair-up/code review early code.
as I mentioned in last daily, we should ensure that with the new UI things as similar or adapt them, agama 12 will reach B2 in few days and currently you can use development job group.

After a painful battle with locators it seems that those work for checking the destructive actions:

private readonly decryptDestructiveActionsText = () =>
this.page.locator("::-p-text(Check the 31 planned actions)");

private readonly skipDestructiveActionsText = () =>
this.page.locator("::-p-text(Check the 16 planned actions)");

VRs on the local VM pass now in both cases.

Thanks for opening PR as suggested in daily the other day, so can do some early review, could you resolve conflicts there to start the process:
https://github.com/jknphy/agama-integration-test-webpack/pull/84

I'm wondering if the case for skip is so fundamental, we might simplify the task here only checking for the opening.
Leo is creating a test suite sles_extended where we could just add it. Most likely we would not be able to adress all the yes/no option in the installer. wdyt?

I now changed the code so that we have a parameter --decrypt-password which works on my local VM.

To enhance the checks for destructive actions I need some more knowledge about all that locator and selector stuff.

One step forward. The storage page now got this:

  private readonly decryptDestructiveActionsText = () => this.page.locator("::-p-text(Check)");
  private readonly destructiveActionText = (name: string) =>
    this.page.locator(`::-p-text(${name})`);

and

  async verifyDecryptDestructiveActions() {
    await this.decryptDestructiveActionsText().click();
    await this.destructiveActionText("Delete btrfs on").wait();
    await this.destructiveActionText("Delete logical volume lv ").wait();
    await this.destructiveActionText("Delete logical volume lv0").wait();
  }

Tests on local VM passed and when adding a typo to the texts to check for the test fails as expected, so it seems to work.
Commit pushed.

Thanks for the update. I try to avoid to do code review when I see conflicts in the PR, because comments might be more confusing that helpful as I would referring to outdated code. Please take care of that each time you push code.
I commented in the PR relative to your previous comment, other than that it is unrevieweable atm.

Rebased and solved all conflicts, but still troubles.
The main problem now is PR#89
which introduces

function ensureProductConfigured() {
  it("should start configuring the product", async function () {
    await new ConfiguringProductPage(page).wait();
  });
  it("should finish configuring the product", async function () {
    await new SidebarPage(page).waitOverviewVisible(40000);
  });
}

And the problem here is that after starting configure the product now the dialog with the decryption options appear, so we will never pass the "should finish configuring the product" part becasue we're stuck in that unhandled dialog. That of course only kicks in when the existing hard drive is encrypted.

Please address: https://openqa.suse.de/tests/17160082#comments