Looks like I found the guilty manifest, please see attached. It is from "concurrent-1.3.4-patched.jar" that is part of confluence 2.7.1 distribution. You may notice that this manifest was created via gcc (?) and follows best practices for all strings in C language - ends with '\0'. So we get a null-terminated manifest here. :-/
Can we avoid throwing exception in such cases? Regards, Alexei 2008/3/28, Alexei Fedotov <[EMAIL PROTECTED]>: > Alexei, > Could you please dump "buf" buffer from InitManifest constructor to > check if it really contains a null character? If it does, we should > think how to ignore it smoothly. > > Thank you in advance. > > > On Fri, Mar 28, 2008 at 3:15 PM, Alexei Zakharov > <[EMAIL PROTECTED]> wrote: > > I'd like to note here that I don't remember any problems with > > manifests loading during confluence startup a couple of months ago > > (however, there were other problems). It looks like manifest parsing > > routines were significantly updated since then. > > > > Regards, > > Alexei > > > > 2008/3/28, Alexei Fedotov <[EMAIL PROTECTED]>: > > > > > > > Alexei's problem (introduced by a patch from [1]) is fixed in my code. > > > He faced another one though. > > > > > > [1] http://issues.apache.org/jira/browse/HARMONY-5473 > > > > > > > > > > > > > > > On Fri, Mar 28, 2008 at 2:41 PM, Tim Ellison <[EMAIL PROTECTED]> wrote: > > > > Tim Ellison wrote: > > > > > I just committed code in this area (at r641928) so it may indeed > be a > > > > > regression. Looking... > > > > > > > > Sorry Alexei, just noticed that your version under test was r640865. > > > > > > > > Given the recent changes is it possible for you to retry on a later > > > > version >r641928 since the failing code has been replaced. > > > > > > > > Thanks, > > > > Tim > > > > > > > > > > > > > > > > > > > > > > > -- > > > With best regards, > > > > > > Alexei > > > > > > > > > > -- > With best regards, > > Alexei >
