Hi, i think my first sugesstion was false. While it seems that we can fix the symptom by changing the final URI-creation in relativize through reparsing we need to be sure that the relativize doesn't change to intension of the original url (which is an absolute path without an shema).
I think the problem is in the Parser (while it can parse the URI "/a:b" ) it misses to encode the forbidden punction into "/a%3Ab". After relativize it is only a URI with the path a:b. But to string and reparsing makes it a URI with a schema of "a" and an schemaspecific-part of "b" I would like to add another low_mask and high_mask to manage all forbidden chars and put this mask into the encode method. So we can encode all forbidden chars for every part of the URI differently. What do you think? Is there someone who would sponsor this? -- Sebastian Am 07.09.2015 um 16:30 schrieb Sebastian Sickelmann: > Hi, > > i want to have a closer look to JDK-8022748[1]. > > I make experimented a little and for me it seemed that the relativize > method is making the trouble. > Has someone had a look at it, already? Or is it a academic (relativize > on "." ) case that is described > in the bug-report? Elsewise I would investicate the case and try to find > a solution to it. > > -- Sebastian > > [1] https://bugs.openjdk.java.net/browse/JDK-8022748 > >
