Redmine

Thank you for posting the patch.
But I think it would be better not to assume that the value of REPLIES_PER_PAGE constant is always 25. The test in your patch may fail if #26033 is implemented.

This patch is great work but sometimes did not work well on my environment.

• MySQL 5.6.42 (ClearDB)
• Ruby 2.5.5p157
• Redmine 4.0.3 + reply_display_order_with_pre_page.patch

• count of the topic children is larger than per_page_option, and
• some children has more than one attachment files

In which case, r=xx parameter may not be expanded correctly due to incorrect paging.

On my environment, @replies.pluck(:id) generated an array containing duplicate IDs.
I could not identify the root cause, but I found a line includes(:author, :attachments, {:board => :project}) cause this problem.
Therefore, I corrected this problem as follows to avoid this problem:

--- a/app/controllers/messages_controller.rb
+++ b/app/controllers/messages_controller.rb
@@ -34,7 +34,6 @@ class MessagesController < ApplicationController
     # Find the page of the requested reply
     replies_order = User.current.wants_comments_in_reverse_order? ? 'DESC' : 'ASC'
     @replies = @topic.children.
-      includes(:author, :attachments, {:board => :project}).
       reorder("#{Message.table_name}.created_on #{replies_order}, #{Message.table_name}.id #{replies_order}")

     if params[:r] && page.nil?
@@ -45,6 +44,7 @@ class MessagesController < ApplicationController
     @reply_count = @replies.count
     @reply_pages = Paginator.new @reply_count, per_page_option, page
     @replies = @replies.
+      includes(:author, :attachments, {:board => :project}).
       limit(@reply_pages.per_page).
       offset(@reply_pages.offset).
       to_a

Popoki Tom (@cat_in_136) wrote:

I improved Minoru-Maeda-san's patch. My patch includes:

• This issue #26030 itself;
• display considering "per_page" (#26033);
• place "reply" block considering the display order (like #31438);
• fix regression that I reported in #26030#note-6;
• fix regression where redirection after deleting a message raises an error; and
• fix test codes to pass rails test -n /Message.*Test/.

Thanks for updating the patch. Can you check the content of test/integration/messages_test.rb? It seems the class class MessagesTest is duplicated.

Popoki Tom (@cat_in_136) wrote:

Opps, Thank you for checking my patch. I've removed the duplicated MessageTest.

My patch has still a some degrade bug where r=xx parameter may not be expanded correctly. This patch should not be merged.