I have provided an .Net component to modify the thesaurus files for Sql
Server FTS. After writing the thesaurus file, I re-start the MSFTESQL
Service and execute a query that will invoke the thesaurus file.

Instead of returning a resultset, I get an event log entry as follows:

Unicode byte order mark (0xFEFF) missing at the beginning of thesaurus file
C:\Program Files\Microsoft SQL
Server\MSSQL.1\MSSQL\FTData\tsenu.xml.Component: MicrosoftIndexer

I've tried all sorts of encoding when writing this file but can't seem to
replicate the format required by MSFTESQL Service. If I modify the thesaurus
file directly without using my api, I do not receive the error in the event
log and my search returns the relevant resultset.

Does anyone have any ideas on how I can create this unicode byte order mark?

Many thanks

Craig

For those interested, I managed to get the byte order marker problem working.

In my API, I was encoding the xml files as BigEndian which is 0xFEFF,
however after some digging around in TextPad, I saw that an un-touch (by my
API) file was encoded as unicode. So I switched the encoding and disco, it
works!

If anyone is interested in using my API, let me know.

Happy me



"Craig" wrote:

Hi Craig,
Sure. The XML Thesaurus Files in SQL 2005 (and un-supported by MS in SQL
2000) have always been a under-utilized and little know feature that for
Yukon may have good benefits as populating the files with useful
production-level replacement text is a bit trick and time-consuming. Email
me at jt-kane at Comcast dot net or just post a comment on my blog.

John
--
SQL Full Text Search Blog
http://spaces.msn.com/members/jtkane/


"Craig" <Craig (AT) discussions (DOT) microsoft.com> wrote