Hello FPC,

I had a problem with strings that are silently converted to '' (nothing) and finally found that this strings pass throught the UTF8Decode function (I'm using Lazarus) and as this strings are ANSI there are some invalid character sequences in some of them. I think this is a problem as the Utf8ToUnicode when find any of this sequences simply discard all the string (I'm using FPC 2.2.2 from 1 week Lazarus snapshot). If this problem has been fixed in more recent FPC releases forget this mail, if not fixed please conside using the attached function instead which take care of wrong sequences and non usable mapping spaces.

As my pascal indentention is not the usual one in FPC you can change it or request for a change. There are some comments about what happends on each step.