I think this is broken in SST. Here's why: One of the optimizations of the record size calculation is to assume a completely "full" first record if it determines the string will definately spill over to another continue record.
Since a string can't always be split perfectly along a record boundry I'm not sure how this could possibily work. I think it's worked okay so far because most of the time a record will split somewhere in the middle which makes this assumption okay. Andy, was this your one or Marc's? I might just rewrite it if thats okay. BTW, what do you think of the changes I've made to SST so far? It's somewhat different from how it was. Regards, Glen Stampoultzis (TriNexus Pty Ltd) Fixed:+61 3 9753-6850 Mob:+61 (0)402 835 458 ICQ: 62722370 EMail: [EMAIL PROTECTED] URL's: http://jakarta.apache.org/poi, http://www.krysalis.org -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
