[TYPO3-english] Re: Re: Re: 6.2.0 Install Tool Update Wizard: MySQL-Error

[Thomas Skierlo]

Hi Xavier,

I's good to hear that you agree collation-wise. Currently the core, as well as Gridelements and my personal extensions show no issues with utf8_unicode_ci. News extension, which is currently disabled in my setup, was showing no artifacts too. I think, I have issues with static_info_tables (Stanislav is taking care of), and possibly indexed_search (an upgraded version with utf8_general_ci works, a brand new install with utf8_unicode_ci silently fails).

My intention to get 6.2.0 running on my dev-system primary was to get my own extensions 6.2-ready. Part of it depends on Gridelements, part on the Rte. Some features of Rte depend on static_info_tables, so this environment must work for me, together with some must-haves, like RealUrl. I'm struggling now for 3 weeks, and can't get my work done.

Currently I'm facing two types of issues. "Real" 6.2 compatibility issues (like usage of loadTCA or other depreciated stuff) and some fatals that might be related to compatibility AND/OR collation. The first usually are easy to fix (unless they are not as cryptical as "Usage of feInterface is no longer part of the TYPO3 CMS Core.") Problems of the latter type effect the whole system, so suddenly install-tool checks fail for extensions, which are running smoothly and which were approved earlier.

What must happen now? Most important is to share the information which made the core "utf8_unicode_ci" ready with all extension developers and some pro-active actions in case of mission-critical extensions (incl. system extensions).

To get the "white noise" out of this critical migration phase it might be a good idea to 

1) Define current state of 6.2.0 as "utf8_general_ci" ready (utf8_unicode_ci pending, currently "experimental"), with a clear statement in favor of utf8_unicode_ci for future sub-releases (like 6.2.2).

2) After logs have calmed down, let's say, around October, there might be a conversion script to migrate existing "general" installations to "unicode". This must be done by a read SQL-pro, since it is an extremely critical action.

Maybe this would get some steam out of our pipes, reducing two battlefields to only one.

Regards,

Thomas