Features / Usability

Posted by hman 24 May 2022 21:16 GMT-0000 posts: 210

Looking into my backup, things get even stranger. The following is true only for tiki_output. In my last backup from March, tiki_output was the only table, as far as I can tell, that did not have any collation...

CREATE TABLE `tiki_output` (
  `entityId` varchar(160) NOT NULL DEFAULT '',
  `objectType` varchar(32) NOT NULL DEFAULT '',
  `outputType` varchar(32) NOT NULL DEFAULT '',
  `version` int(8) NOT NULL DEFAULT '0',
  `outputId` int(11) NOT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8;

Other tables do have utf8_unicode_ci... And tiki_output in my backup of the other Tiki ist defined as

CREATE TABLE `tiki_output` (
  `entityId` varchar(160) COLLATE utf8_unicode_ci NOT NULL DEFAULT '',
  `objectType` varchar(32) COLLATE utf8_unicode_ci NOT NULL DEFAULT '',
  `outputType` varchar(32) COLLATE utf8_unicode_ci NOT NULL DEFAULT '',
  `version` int(8) NOT NULL DEFAULT '0',
  `outputId` int(11) NOT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci;

That backup is just one week older, so it's comparable.

This raises the question: What changed the collation of tiki_output after March? Is this some Tiki feature to "repair" things collationwise? If so, why isn't the correct collation applied?