Hi Chris, thanks for the review.
You are right, it looks more consistent if the code would also 'goto cleanupAndReturn' at getaddrinfo failure. I'll change that before pushing. For me this one was driven by reducing diffs of our SAP JVM variant to the master. We did that change already quite some time ago but I haven't been at the team back then so I can't give more insights on the motivation at the time. The code comment tells that in some configurations different results to the getXXbyXX calls could be observed and it lead to issues. So they obviously wanted to align this. Christoph. > -----Original Message----- > From: Chris Hegarty [mailto:chris.hega...@oracle.com] > Sent: Montag, 19. Dezember 2016 20:19 > To: Langer, Christoph <christoph.lan...@sap.com> > Cc: OpenJDK Network Dev list <net-dev@openjdk.java.net> > Subject: Re: RFR (M): 8171077: Use getaddrinfo/getnameinfo in Windows > Inet4AddresImpl native code > > Hi Christoph, > > On 12 Dec 2016, at 10:09, Langer, Christoph <christoph.lan...@sap.com> > wrote: > > > > Hi again, > > > > this is the Windows part. > > > > Bug: https://bugs.openjdk.java.net/browse/JDK-8171077 > > Webrev: http://cr.openjdk.java.net/~clanger/webrevs/8171077.0/ > > I think this looks ok. > > For consistency, should the code ‘goto cleanupAndReturn’ if > getaddrinfo fails ( rather than release + return ) ? > > Out of curiosity were you seeing issues with the older > getXXXbyXXX calls, or is this cleanup/modernisation driven? > > -Chris.
