Thomas' commits for JBAS-6436 looked a bit scary and there is probably not good justification for this type of changes on Branch_5_0 while we are trying to fix VFS and release 5.0.1. Communicating with Remy would also be a good idea, although the whole thing was on the forums: http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4205438 However, after the initial checking and some corrective fixes the testsuite passed. I don't know about TCK though because we didn't run it immediately after that. The problems started after testing with the latest VFS snapshots. I don't know, maybe we should apply Ales' proposed fix and see how it goes before rolling back anything. Ales Justin wrote: >> Thomas, >> >> I am looking at revision 83709 and 83774, which likely are causing the >> VFS problems that are reported by Ales. > > Actually it's not Thomas's code. > It's this hack-ish line: > > if (warName.endsWith("/") == false || warName.endsWith("!/") == true) > // Hack for jar urls being exposed > > Since current URL is neither. > - it ends with '/' > - it has no jar ending (in vfs we don't need that) > >> In addition to being very risky >> and done without even contacting me to talk about them, the actual code >> changes are obsfucated inside a ton of formatting changes (because you >> do not seem to like my formatting. Please never do that, at least put >> the formatting changes inside a commit, done prior to the actual code >> changes. > > We don't have an exact formating rules? > >> My opinion right now is that these two revisions should be reverted, >> 5.0.1 is not the right place for this sort of risky refactoring (should >> be planned for 5.1, carefully). > > Afaik Thomas did this with care - line by line. > So I don't agree with you about the revert. > This should've been done long ago. > > OK, whatever you decide wrt revert or not, > I need that line fixed. > > Will you do it or should I? >