Posted by
Matthew Hardy - February 2010
pike/7.8.352/lib/modules/Sql.pmod/sql_util.pmod:172: Sql.sql_util.MySQLUnicodeWrapper()->fetch_row()

Truncated UTF-8 sequence at end of string.
Unknown program: utf8_to_string("thestring which includes or ")

Looks like a random show stopper, after a number of retries the error goes away. ie. a reconnect all is ok again.

Any ideas or other using unicode connections to MySQL?
Posted by
Martin Stjernholm - February 2010
Is it random for the same response string? What is the string?
Posted by
Matthew Hardy - February 2010

Basically a name like Anders Jrgensen

It seems to be random in that a reconnect generally brings things back to  a good result.

In this case it is coming from the Roxen SQL tags module. <emit...>

It may be that the connection is reverting to Latin1 without realising the connection needed to be unicode. We've seen it before with the Roxen sql tag.

Oh - we think it is only an issue on the 7.8.352 - apparently there are fixes that cover the issue in 7.8.385
But I guess it'll still affect the shipped roxen version... but the mysql client does not support unicode - so maybe it is impossible to happen for everyone but users of non-roxen Pikes...
Posted by
Arne Goedeke - February 2010
I think its the issue that has already been solved in 7.8.37something. There was a discussion back then on the pike-devel mailing list. Installing a more recent version should be enough.
Posted by
Matthew Hardy - February 2010
Pike 7.8.409 installs fine on OS X
Will report back if any more unicode issues, but none are expected ;)
Search this thread: