OK I found the problem: 3.x can't handle the U+FFFF character (we replace it on indexing), while trunk can. So I think we just have to fix randomFixedByteLengthUnicodeString to never use that char. I'll commit...
Mike McCandless http://blog.mikemccandless.com On Sun, Jul 31, 2011 at 10:33 PM, Robert Muir <[email protected]> wrote: > On Sun, Jul 31, 2011 at 10:24 PM, Chris Hostetter > <[email protected]> wrote: >> >> : fix test to not create invalid unicode >> >> I'm confused ... when/why does randomFixedByteLengthUnicodeString not >> return valid unicode? >> > > I'm confused too, but all of the issues are with replacement chars for > invalid unicode: > > Since the intent of this test is to test thread safety, not to test > unicode enc/dec back and forth, I switched it to ascii until the test > grows up (e.g. trunk, which now uses full unicode range correctly, > maybe I backported this wrong before) > > -- > lucidimagination.com > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
