On Wed, Jun 8, 2011 at 08:00, <t.bro...@gmail.com> wrote: > On 2011/06/08 14:48:12, tbroyer wrote: > >> On 2011/06/07 16:34:24, xtof wrote: >> > Thomas, here's a patch that adds a test case to repro the bug >> > > FYI, with openjdk-6-sdk on Ubuntu 11.04, the test passes, but the >> testFromTrustedString_withInvalidUrl fails to throw the expected >> IllegalArgumentException; which means the "surrogate-pair detection >> > regexp" > >> doesn't work as expected. >> Strangeā¦ >> I think I'm going to rewrite it the "scan each char" way. >> > > Oops! Silly me! Forgot to uncomment the check. Now the *_withInvalidUrl > passes, but so does the testFromTrustedString w/ LONG_DATA_URL. > > (but then, isn't the testFromTrustedString_withInvalidUrl also failing > for you with the check commented as in r10285? or is "new URI" throwing > a URISyntaxException then?) > > The tests pass for me at r10285. I was a little surprised that they pass with the checks disabled but didn't have time to think about it. It might mean that a test is passing vacuously? Hmm, you're probably right that new URI is failing.
> http://gwt-code-reviews.appspot.com/1443813/ > -- http://groups.google.com/group/Google-Web-Toolkit-Contributors