On 6 Dec 2012, at 14:14, Igor Tandetnik wrote: > Your code assumes, in several places, that strings passed to collation > function are NUL-terminated. They don't have to be - that's why lengths are > also passed. I think this may be causing the problem you are seeing: when the > string comes from a literal (as in x < '' ) it just may happen to be > NUL-terminated, but when it comes straight from the database, it may not be, > and you are cheerfully reading garbage past the end of buffer.
Thanks for your analysis and helpful comments. I'll need to take care to make a NUL-terminated copy of each source string, as inet_pton doesn't take a count argument. > Your test program, of course, always happens to pass NUL-terminated strings. Doh! 8-) /Niall _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users