On Fri, Dec 5, 2014 at 6:00 PM, Alan Bateman <alan.bate...@oracle.com> wrote: > On 05/12/2014 16:50, Volker Simonis wrote: >> >> Hi Chris, >> >> thanks for the fast response. >> >> I saw that code in ImageBuilder, but it looked overly complicated to >> me. What about cleaning that up as well: >> >> http://cr.openjdk.java.net/~simonis/webrevs/8066766.v2/ >> >> I've just checked that on Solaris 'jspawnhelper' still has the right >> execution bits set after the change. > > setExecutable doesn't throw an exception when it fails so this is why it's > not used here. >
But why do we need an exception if setting the executable flags fails on 'jspawnhelper' and don't need on if it fails on the executables. We'll actually never notice that 'jspawnhelper' isn't executable if we can't execute 'java', right? Moreover the old solution depends on the fact that PosixFilePermission are available. I understand that the 'jspawnhelper' is probably only used on systems which run on a posix files system, but who knows. On the other side using setExecutable() on all executables as proposed by Chris seems dangarous, because Windows probably doesn't supports PosixFilePermission so we will always get exceptions. > It's probably not a big deal here of course, esp. as this is just a > temporary tool for the build that we hope to replace soon. > So do you agree with the clean-up in my second webrev or should I just change the way how the executables are treated (like in my first webrev)? Thank you and best regards, Volker > -Alan
