Search engine strangeness

[gabrielxz]

I noticed another strangeness on my wiki today. It may have started when I updated to 10.0.3, but I can't be sure. Some searches aren't showing full results. It seems that titles of pages are being left out a lot of the time (but not always). After reading around the forums, I rebuilt my index. When that didn't work, I stopped the wiki, deleted the index folder, rebuilt the index, and then restarted the wiki. That didn't change anything. Any ideas? I'm on a VM installation.

[tisso]

I have the same problem here. It really looks like some Pages are completely left out. I noticed that the search-results show all attached pdf-files but only a few pages are indexed.

i just noticed following entries in deki-api.log

2010-10-29 14:00:51,451 [DispatchThread #9] ERROR MindTouch.LuceneService.UpdateDelayQueue - dispatch of 'http://default/Administration/Bestellbedarf' encountered an error
System.IO.InvalidDataException: message body with content type 'application/xhtml+xml; charset=utf-8' is not well-formed xml
at MindTouch.Dream.DreamMessage.MakeDocument () [0x00000]
at MindTouch.Dream.DreamMessage.ToDocument () [0x00000]
at MindTouch.LuceneService.LuceneService+<OnQueueExpire>d__27.MoveNext () [0x00000]
at MindTouch.Tasking.Coroutine.Continue () [0x00000]
--- End of exception stack trace ---
at MindTouch.LuceneService.LuceneService.OnQueueExpir e(UpdateRecord data, Result result)
--- End of coroutine stack trace ---

looks like lucene has a problem with some pages in utf-8, i think it's because of the german "umlauts" on those Pages, i'll try to reproduce and post the Results.

update:

It is like i suspected. All pages without special characters are correctly indexed. As soon as i add an äöü in the content, lucene gets an error described above.

This is a major issue.

[Guerric]

We have identified this issue and are in the process of packaging the 10.0.3a release. Stay tuned.

[derblitz]

Hello Guerric,

i have the same problem.
The searchfunction is very important for me.

Please post till the Update is available.

Thanks.

Simon

[Guerric]

10.0.3a has been released to resolve this issue:
http://blog.developer.mindtouch.com/2010/11/02/critical-update-mindtouch-10-0-3a/

[gabrielxz]

This fixed my problem completely. Thanks.

[derblitz]

When can we get the DEB Package?

[derblitz]

Is there a manual way to update?

[crb]

apt-get it.

It's called 'mindtouch' now.

If you wait until 10.04 you will be able to install mindtouch and have it replace the existing 'dekiwiki' package.