ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Wannheden, Knut" <>
Subject RE: OpenVMS changes (was Re: FileUtils#normalize(File))
Date Thu, 31 Jul 2003 15:43:09 GMT

> > On the downside a FileUtils#toVMSPath(File) would push the OS
> > awareness responsibility up a level, in which case
> > Commandline.Argument.setFile(File) would have to check for OpenVMS
> > and call this method.
> I agree, Commandline.Argument shouldn't need to be aware of this.
> Maybe the better place for toOSPath would still be Execute, even if it
> would start stretching the responsibility of that class.

Yes, Execute is rather large now: 1000+ lines.  I have implemented a

> > I was thinking about the FileUtils#getSetLastModified(), which using
> > reflection returns the File#setLastModified(long) method.  Alas, I
> > found that exactly this method doesn't have any effect on OpenVMS
> Well at least it won't make a difference whether you invoke it via
> reflection or not.

No, not really.  It just looks ugly.

> One question I asked some time last week in a post with a subject that
> didn't catch your interest.  Does the OpenVMS VM translate the exit
> code?  I.e. if your Java application exits with System.exit(0), what
> will the return code be?

No, it doesn't.  The $STATUS symbol will be set to the return code and
$SEVERITY to the return code modulo 8.

Although I found that if I run Java inside the bash shell of GNV (GNU is not
VMS :-)) then the return code is rewritten by GNV!  This is probably because
GNV thinks you're running an image following VMS standards.  So there even
return codes are mapped to 2 and odd ones to 0.

But Java shouldn't be subject to this, even if it's running as a child
process.  So again, the VM doesn't tanslate the return code.  Also note that
the VM will return with exit code 0 on normal termination (no explicit
System.exit() call).

I guess the reason for this is that they are in a little bit of a dilemma.
They both have to conform to Java (Unix) and VMS standards at the same time.

> I've changed <java> to use Execute#isFailure in the forked case, but
> that would be wrong if the VM doesn't translate the exit code.

Then it is wrong :-(



  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message