Re:UTF-7: a ghost from the time before UTF-8
by valesco on Saturday December 01, 2018 @08:36AM
(#18576717)
Good job on improving the system! However it seems like using 'Référence CSS' would have been more correct grammatically and wouldn't have caused the problem.
Re:UTF-7: a ghost from the time before UTF-8
by ubernostrum on Saturday December 01, 2018 @12:02PM
(#18577610)
The authors/editors for different languages were the ones who chose the styling of the tags. Don't know why they put it that way, but they did. Our job was just to make it work.
https://bugzilla.mozilla.org/show_bug.cgi?id=776048
The tl;dr is that MDN is a multilingual site, and supports categorization in a few ways, including by tagging. An example of the bug: the English MDN's CSS reference tagged articles with "CSS Reference". The French MDN's CSS reference tagged them "CSS Référence". Sometimes, the French tag appeared on English articles.
The source of the issue was MySQL's utf8_general_ci collation, which did not see "e" and "é" as distinct, so it was a toss-up as to which tag you'd get back from the database. The solution was to build and install a custom collation to teach MySQL to see accented and unaccented characters as distinct.