Last modified: 2013-04-08 11:02:21 UTC
The log comment is not truncate with full bytes. This can leave half bytes inside the database, which are shown as broken characters when looking at the log comment. Please use Language::truncate to truncate the whole full byte at the end of the comment. Thanks.
I assume you mean "characters", not "bytes" since half bytes would be 4 bits and doesn't really make sense, but Unicode characters can be multiple bytes long.
I cannot reproduce this bug by using the string 漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸漸 as a pagemove comment. Can you please provide a test case?
Note, I also used that string with a "." at the start to shift the comment one byte to the right, and it still didn't show a broken character.
Created attachment 8895 [details] add truncate to LogPage::addEntry Have you looked into the log after the move or have you looked into the history of that page? The comment of the new revision is truncated, but not the log entry. Using the following comment shows an broken character at the end: 01234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123ä
Thank you for the patch - Committed in r95456 btw when adding a patch to a bug, please also add the "needs-review" keyword. That way it is less likely to fall through the cracks. Cheers.
