Notice: Undefined index: uid in simple a user view

There is at least one more bug hidden:

base_field: nid
base_table: users

Thank you for providing that report!

I can confirm this bug. It still exists in the 8.x branch.

I have attached a patch to fix the undefined index PHP notice.

Re-rolled the patch.

I also think that

-    // If we want a formatted username, do that.
-    if (!empty($this->options['format_username'])) {
-      return user_format_name($account);
+      if (!empty($this->options['format_username'])) {
+        return user_format_name($account);

is not necessary, so I didn't include that in #5 patch.
(It only change indentation).

That change did more than just indentation, it moved the if statement inside of the other if statement.

Yes that did.. didn't notice that, thanks! Re-rolling now.

Trying again.

Should keep String:checkPlain().

I've updated the issue summary with steps to reproduce, and a beta-phase evaluation. I've confirmed the patch in #13 removes the notice. RTBC I think.

We should have a test for this.

I spent a bit of time trying to get Drupal\user\Tests\Views\HandlerFieldUserNameTest to throw this notice, but something in how that code is being tested bypasses the PHP notice. I'll revisit if I get a chance.

Here's the issue: Since the test view test_views_handler_field_usern_name has certain defaults, the various init() methods in the test are setting the additional_fields parameter, which when set, avoids the PHP notice. Manually toggling these settings during the course of a test doesn't unset those from the field handler.

  /**
   * Overrides \Drupal\views\Plugin\views\field\FieldPluginBase::init().
   */
  public function init(ViewExecutable $view, DisplayPluginBase $display, array &$options = NULL) {
    parent::init($view, $display, $options);

    if (!empty($this->options['link_to_user'])) {
      $this->additional_fields['uid'] = 'uid';
    }
  }

Therefore, the test for this can either change the defaults in the existing test view, or add a new test view. Changing the defaults to false (format_username, overwrite_anonymous, and link_to_user) breaks some current tests, but definitely throws the php notice.

Here's a test. As mentioned above, I needed to rework the default settings of the existing user name field handler view, thus the changes to the current test.

Test looks good.

Applied the patch, edited the people view, checked "Overwrite the value to display for anonymous users" and set "Text to display for anonymous users" to "visitor". Went to admin/people and see "visitor" instead of "admin".

+++ b/core/modules/user/src/Plugin/views/field/Name.php
@@ -81,25 +81,23 @@ public function buildOptionsForm(&$form, FormStateInterface $form_state) {
-      if (!empty($this->options['overwrite_anonymous']) && !$account->id()) {
...
+    if (!empty($this->options['overwrite_anonymous']) && !$account->id()) {

This line....

+++ b/core/modules/user/src/Plugin/views/field/Name.php
@@ -81,25 +81,23 @@ public function buildOptionsForm(&$form, FormStateInterface $form_state) {
+      $account->uid = $this->getValue($values, 'uid');

Needs this to be set.
That leads to the problem in #23.

Updated patch in #20 to account for this.

Well that undid the original fix, so that doesn't help much :) Back to the drawing board.....

Let's try this again....

Given #23 I'm concerned that neither the new tests, nor existing tests, caught that. Perhaps a few more assertions are in order for better coverage around the anonymous username formatting?

Tried something, but it's not working. I don't get it because it looks like the test already logs in a user, but it's supposed to be testing an anonymous user.

Last patch bad. this is what i tried.

Let's make the testbot do it's magic....

In the previous patch, I tried to change the logged in user before executing the view; But, I didn't need that. What WAS needed is to show that overwrite anonymous doesn't affect a regular user's name. Here is a test-only patch against #20 to show the test failure. Plus, an interdiff, and the new patch.

Fixes the issue and the issues in #17, #23 and #28 are addressed.

Nice tests - thanks. Committed 4900170 and pushed to 8.0.x. Thanks!

Thanks for adding the beta evaluation to the issue summary.