The fix should be available for testing on the next 5.2 milestone release which will be documented here shortly https://confluence.atlassian.com/x/i4NkFQ

I had noticed odd behaviour with search when I startede using confluence but put it down to inexperience. We are now about to deploy a space which relies on search - and the trial users have hit this and dont like it.

Is there any likelyhood that this gets fixed......preferably soon

kevin

PLEASE fix this, this is really a problem for german customers.

By the way there is one more problem when Searching in German Language:

German connects all nouns without spaces. For example instead of "Polizei Auto Rad Naben Schrauben Schlüssel" we write "Polizeiautoradnabenschraubenschlüssel". This might seem funny for foreigner but it is very typical.

If I search "schrauben" in confluence, I will not find the word, and I cannot type schrauben because Confluence does not allow asterix and the beginning of a word. This needs to be fixed, too, when talking about wild cards.

Thanks to Atlassian for your help on this,
Henning

I noticed the problem is even worse in German working environments:
In German language nouns always start with an upper case letter, rather than smaller case like in English. So people of course tent to type in upper case letters.

If there is a page named "Kontierung" and you search for "Kont*" (correct spelling!) you will NOT find the page, but when you for "kont*" then you will find the page.

So the search is case sensitive, and expects the wrong case! Please fix quickly, because this is hart do explain to our users and not intuitive. Smaller case or upper case should not matter at all when searching.

Thanks,
Henning

Hi Jurriaan van Reijsen,

as you may see, the patch was provided from me ... therefore I can answer a few of your questions.

Due to this issue, nothing has changed since then. That means - as you have already stated - the wildcard search is still case sensitive. And I cannot tell you why. I also don't know if the core implementation changed and if the patch is still working or not.
But you can proceed with the patch using an IDE of your choice, adding the Confluence source code to it and apply the patch using the IDE. Then you can also see and fix the changed lines of code or conflicts. Afterwards you have to compile Confluence your own and deploy the war file.

Hope this helps ...

Hi,

I noticed that the previous comment dates back 2+ years. Has anything been changed/updated since then? We are currently on Confluence 4.3.3 and notice that wildcard searches are still case sensitive. Is the provided diff-tng-confluence-3.1-source-wildcard-case file still able to fix cae sensitivity for wildcard searches? And how is that file implemented in our instance? I cannot find any documentation on how to proceed with the file. Is there a reason why this fix was not implemented as default in newer versions of Confluence?

Thanks in advance for clarifying the current state of wildcard searches.

Best regards,
Jurriaan van Reijsen
The Courseware Company - Dutch Atlassian Expert

There are, however, some fields in our index that are stored with their case intact.

Can you supply a list containing this fields?

Enhanced patch diff-confluence-3.4.1-enhanced-tngpatched.txt excludes a specified list of search fields. At the moment just "content-name-untokenized" which I found on http://confluence.atlassian.com/display/DOC/Confluence+Search+Syntax.

you may be right as we didn't think about this issue that deep
would be great if you would fix that until next release

This has been reclassified as a bug. The patch provided does solve the issue, but will create another issue. Currently, Confluence stores most fields as lower case in the search index, and for normal searches, converts the search terms to lower case. For wild card searches, search terms are not converted to lower case, which causes wild card searches with upper case characters to return no results. There are, however, some fields in our index that are stored with their case intact. When searching with wild cards specifically on these fields, the case needs to be preserved (which this patch does not do).