Jaroslav, Nope. I'd replaced calls to writeShort and writeByte to one writeUTF.
I'll reformat comments to avoid any confusions. -Dmitry On 2013-06-04 13:57, Jaroslav Bachorik wrote: > On Tue 04 Jun 2013 11:50:41 AM CEST, Dmitry Samersoff wrote: >> Staffan, >> >> Two bytes len prefix is a part of JDP protocol specification, so I would >> prefer to write it explicitly. >> >> But changed to *writeUTF()* as recommended. > > Wouldn't the length be written twice now? *writeUTF()* also writes the > string length. > > -JB- > >> >> see: http://cr.openjdk.java.net/~dsamersoff/8015604/webrev.02/ >> >> -Dmitry >> >> >> On 2013-06-04 11:48, Staffan Larsen wrote: >>> Looks good. >>> >>> Even better would have been to use DataOutputStream.writeUTF(). >>> >>> /Staffan >>> >>> On 3 jun 2013, at 23:33, Dmitry Samersoff <dmitry.samers...@oracle.com> >>> wrote: >>> >>>> Hi Everybody, >>>> >>>> The problem: >>>> >>>> Code uses string length rather than byte array length to write length >>>> field of JDP packet entry. So if an entry contains non-ASCII characters >>>> code produces invalid packet. >>>> >>>> >>>> http://cr.openjdk.java.net/~dsamersoff/8015604/webrev.01/ >>>> >>>> -Dmitry >>>> >>>> >>>> -- >>>> Dmitry Samersoff >>>> Oracle Java development team, Saint Petersburg, Russia >>>> * I would love to change the world, but they won't give me the sources. >>> >> >> > > -- Dmitry Samersoff Oracle Java development team, Saint Petersburg, Russia * I would love to change the world, but they won't give me the sources.
