Thread

Last month I upgraded my site from EE 1.x to EE 2.2.2. I’ve since discovered that large sections of text in my wiki module were deleted during the upgrade. Specifically, in articles that contained foreign characters, such as the Swedish letter å, everything after the foreign symbol (including the symbol itself) in the article was deleted.

Luckily I made a backup of my database before the upgrade. So I’m able to recover the missing text. So I’m posting this more as a bug report than as a request for help. Just thought you guys should be aware of the issue.

Hi Alex,

This is gonna be a hard one to troubleshoot but lets try and gather some info first.

Do you remember the what the collation of your database was set to?
Typically in an EE1.x database it would be latin1_swedish_ci and in EE2.x it would be utf8_general_ci

Did you have any add-ons running in EE1.x?

Can you attach a text file to your reply so we can test an upgrade with some of your content

Unfortunately I have no idea what the collation was. If it helps I could send my entire database backup for you guys to play around with, though it’s around 200mb.

I did have add-ons running in EE 1.x, including an add-on for the wiki module (a table of contents add-on). But again, I can’t really remember what the other add-ons were.

I’ve attached one of the articles that got partially deleted during the upgrade. Halfway through there’s a french word, infâme. Everything after that word got deleted.

Hi, Alex. That one entry should be enough to test against. I’ll add that to a wiki entry and convert this afternoon.

Were you using utf-8 or some other collation?

Hi Sue,

I assume I would have been using whatever the standard setting for the wiki module was. It’s not something I remember altering.

Hi, Alex. One other quick question - what collation is being used for that particular table? That may make a difference. Utf-8, Latin1?

I know you said you didn’t know, but if you still have the database dump, you can see that towards the end of each table structure. Look for CHARSET=something

I dropped the ball on this and failed to respond promptly. Sorry. But I recently learned that the problem with text being deleted during the upgrade process was actually far more widespread than I realized, and it wasn’t only limited to the wiki module.

I’ve been receiving emails from people complaining about why their comments were deleted from my site. At first I was confused until I realized the deletions were related to the upgrade problem.

Specifically, if any comments had smart quotes or foreign symbols in them, they got deleted. Or, that is, everything after the symbol which the upgrade program didn’t like got deleted.

So anyway, I figured out what charset I was using. It was utf8. Hope that helps.

As for restoring everything that got deleted, I doubt that’s possible. I did make a backup before I upgraded. So for anything that’s really important, I can find it and restore it. But it would take too long to do that for everything that disappeared. So much of it will have to remain disappeared.

Alex, I hate that you have suffered a loss of data. Providing your encoding before the upgrade will help us track down any problems with the updater.

Is there anything else we can do for you?