On 12/26/2020 3:07 PM, Mark Sapiro wrote:
As far as utf8mb4 vs. utf8 is concerned. There will be issues with the hyperkitty_email table subject and content columns in particular and maybe others if email messages contain 4-byte utf-8 unicodes. However, as I read the MySQL docs, for columns like VARCHAR(255) and VARCHAR(510), the difference affects only the number of characters that can be stored in the column and not the actual column width. I.e., a VARCHAR(255) column is actually up to 256 bytes wide (1 byte for length and up to 255 butes of data) and can thus hold up to 85 3-byte utf-8 encodings or up to 63 4-byte utf-8 encodings, but since any string normally contains only a few if any (probably none in the case of the mailinglist table) the actual limit is much closer to 85.
Earlier I had attempted changing the encoding and collation and manually adding the new column, and it failed with the same error. This fits with Mark's explanation of how the column size and encoding go together.
I just ran an "optimize table mailinglist;" and then the update ran clean.
Thanks for the eyeballs today, no idea why that would have worked but now we're up and running.
-- Joel Lord