Filter criteria for views, option "Content is updated" - causes a SQL syntax error

Do you mean the filter named "Content: Has new content"?

Thank you for translating. I tried that filter on a view of nodes and did not see the error you reported.

Can you post the database version you are running and any other possibly relevant details in the issue summary? And was this on the current Drupal 8.0.x HEAD from git?

Perhaps someone else can reproduce the issue.

I moved this to the views_ui module, which is the Views management module.

views_ui is wrong, this is certainly views which causes the problem.

@fomenkoandrey Do you mind providing an export of the view

A simplified view to reproduce the bug.

The error disappeared after attaching comment field to Article content type.

should fail

combined 15 and 16

It would be great if someone could describe in the issue summary or better via a comment in the code why this ensureTable call could fail ...
Needs work as we need tests.

added a note

+++ b/core/modules/history/src/Plugin/views/filter/HistoryUserTimestamp.php
@@ -82,9 +82,12 @@ public function query() {
+      // The table is not available if none of the content types has attached
+      // comment field.

Interesting, so we already have sort of a test coverage, great!

+++ b/core/modules/history/src/Plugin/views/filter/HistoryUserTimestamp.php
@@ -82,9 +82,12 @@ public function query() {
+      // The table is not available if none of the content types has attached
+      // comment field.

This should read "...has an attached comment field." or "... has attached comment fields.", whichever is more correct. Then, it can go right back to RTBC.

Back to RTBC per #25.

@fomenkoandrey, the patch is not commited yet.

Its still green

@Chi
Do you think this has to be committed before 8.0.0? For me this is a normal bugfix, but let's see whether we can convince the core committers

@dawehner, I was unaware that usual bugfixes are not allowed in release candidates. I do not understand the reason behind this but that won't be a big problem since D8 release is coming soon.

+++ b/core/modules/history/src/Tests/Views/HistoryTimestampTest.php
@@ -24,7 +24,7 @@ class HistoryTimestampTest extends ViewTestBase {
-  public static $modules = array('history', 'node');
+  public static $modules = array('history', 'node', 'comment');

So this is what provides the test coverage, but I have no indication as to why it's necessary for the comment module to be added in this test.

Can we either add an explicit test or at least a comment as to how it provides coverage? I'd prefer the former probably to decouple the coverage.

@effulgentsia and I agreed with both the points in the summary in terms of RC. This is indeed a bug that can be fixed in a patch release. Since it causes a fatal, though, that seems like an OK reason to make the change during RC since there is no disruption.

Should fail.

It looks for me that moduleExists('comment') is not needed here since we are checking comment_entity_statistics table.

Well,

Committed/pushed to 8.0.x, thanks!