Skip to content
Snippets Groups Projects
  1. Apr 28, 2016
    • Jared Hancock's avatar
      cli: deploy: Automatically deploy rewritten files · 93402825
      Jared Hancock authored
      With the advent of the MANIFEST file, files which are rewritten when
      deployed, such as those which have the GIT hash in the query string to force
      browser reloading, were no longer deployed unless the content of those files
      changed.
      
      This patch adds a flag to the MANIFEST file to indicate that the file was
      rewritten when it was deployed. This allows the file to be deployed and
      rewritten again when the deployment is run, but also adds an indication
      to the console output to distinguish deploying changes as opposed to
      deploying rewrites.
      93402825
    • Jared Hancock's avatar
      b7529965
  2. Apr 25, 2016
  3. Apr 24, 2016
  4. Apr 19, 2016
    • Jared Hancock's avatar
    • Jared Hancock's avatar
      orm: Propagate LEFT joins in join paths · 6be61133
      Jared Hancock authored
      If something like members__staff is considered leaving the Team model,
      and the `members` relationship is nullable, and the `staff` relationship is
      not, in the context of the compiled SQL statement, the second join should
      also be considered nullable (LEFT join), because otherwise inconsistent
      results would be returned from the query.
      
      In other words, if a count is considered as an annotation to the Team model
      instances, Teams with zero members should still be considered as valid teams
      and should be selected with such an annotation. Before this patch, however,
      the join between TeamMember and Staff would have been an inner join instead
      of a LEFT join, which could skew the database results.
      6be61133
    • Jared Hancock's avatar
      oops: Fix max of a single attachment · d1a9161f
      Jared Hancock authored
      d1a9161f
    • Jared Hancock's avatar
      forms: Fix lag updating the CDATA table · 33455cae
      Jared Hancock authored
      This fixes an issue where the CDATA table is updated with the OLD version of
      an answer after it is updated. That is, after an answer is updated, the
      CDATA table is also updated to reflect the change; however, the old content
      of the answer value is used when updating the CDATA value.
      33455cae
  5. Apr 14, 2016
  6. Apr 13, 2016
    • Jared Hancock's avatar
      orm: Don't null out value on assignment · 93a34c43
      Jared Hancock authored
      This fixes an issue where the PK of ThreadEntry will be NULLed in the
      following code
      
      ```php
      $entry = ThreadEntry::create(array(...));
      $entry->save();
      $entry->email_info = new ThreadEntryEmailInfo(array(
          'mid' => 'xyzAbc',
      ));
      ```
      
      In the above code, the $entry->__set('email_info', <ThreadEntryEmailInfo>)
      would be invoked. The ThreadEntryEmailInfo object is new, and so will cause
      the local part of the relationship (`id` in this case) to
      become null further down in the `__set` method.
      
      This issue is fixed in this commit by removing the NULL assignment to the
      new object. This was added in 11322766,
      however, it is unclear why the null assignment is performed.
      93a34c43
    • Jared Hancock's avatar
      files: Attempt to standardize thread entry attaching · 00a1371d
      Jared Hancock authored
      This commit attempts to remove some of the confusing and redundant code to
      attach files to thread entries and replace it with a single code base. It
      also attempts to remove and error where a single attachment might be
      attached to a new thread entry multiple times.
      
      Lastly, it removes the insert followed by an update for emails with inline
      images. This should improve performance processing emails as only one trip
      to the database is now necessary for thread entries with inline images.
      00a1371d
  7. Apr 08, 2016
  8. Apr 07, 2016
    • Peter Rotich's avatar
      closed-by: Credit the agent opening a ticket · 8598e999
      Peter Rotich authored
      This PR addresses a bug where an agent-opened ticket doesn't set closed by
      or mistakenly set it to auto-assigned agent when a status of closed state is
      selected.
      
      * Always credit the agent opening the ticket with close credit on closure
      * Only do ticket auto-assignment IF the resulting status is in open state
      8598e999
  9. Apr 01, 2016
  10. Mar 30, 2016
  11. Mar 29, 2016
  12. Mar 28, 2016
  13. Mar 26, 2016
  14. Mar 25, 2016
Loading