1. moved handling of "relative" format from "kDateFormatter::Parse" method (where it wasn't working obviously) to the "kDateFormatter::Format" method
2. restored "DateHelper" class, that was removed in 622 diff by mistake

Please test again. Also update test plan to include testing of added code fragment: http://qa.in-portal.org/D379?vs=927&id=929#toc .

Don't perform unique validation from TEMP to LIVE table, when parent record isn't yet created in TEMP table.

Detected an issue (SQL error), when validation of sub-item happens to before parent item is saved into the database.

In D377#7476, @erik wrote:

Tested, accepted.

In the run_event. php first line must be commented to make test plan possible.

exit_code(1, 'This script needs to be enabled manually !');

Debug code removed.

Usage of \FileHelper::ensureUniqueFilename was expecting specific version of that method only available in In-Portal 5.2.x.

Fixed case, when "output_format" option was ignored, when image wasn't resized.

Opening Front-End gives me these PHP warnings:

Handle cases, when slave is too busy catching up with a master.

In the themes\advanced\in-bulletin\elements\side_boxes\polls.elm.tpl change poll image output code to:

Also since task does optimization only, then here we should be checking that persistent setting retrieval with/without fallback code works as before (not covered) + verify performed SQL count (covered).

In D365#7225, @erik wrote:

Tested, working fine, but needs corrections in the test plan.

1. Some additional table must be in the master DB, not in slave DB.
2. in the tools/run_event.php line #15, which blocks script's test part execution, must be commented:
   
   exit_code(1, 'This script needs to be enabled manually !');

Please retest all from scratch, because I've rewritten test plan.

Sending latest changes (maybe nothing was changed), but just in case.

Changed the code. Need to update test plan.

I haven't noticed, that even though this fix looks standalone it still results in a Fatal Error until fix from D359 is applied. I've updated test plan to reflect that.

Fixed a typo.

Introduce global flag, that allows to disable logging for slow queries for a block of sqls.

1. make sure, that SELECT queries used for slow query logging aren't counted towards total query count limit
2. [bugfix] attemt to get non-debug database query iterator while in debug mode would result in an exception

I've found one use case, that wasn't covered by initial plan.