Could the issue be I am behind a proxy?
However, I will not be able to test otherwise before tonight. Just to summarize with my Chrome or Firexfox: * corrupt: http://www.apache.org/dyn/closer.lua/qpid/java/6.1.1/qpid-java-6.1.1.tar.gz * works: http://www.apache.org/dyn/closer.cgi?filename=qpid/java/6.1.1/qpid-java-6.1.1.tar.gz&action=download I asked other colleagues and they have the same issue. Regards, Adel ________________________________ From: Justin Ross <[email protected]> Sent: Monday, March 20, 2017 2:47:55 PM To: [email protected] Subject: Re: Corrupt artifacts on Qpid release web page On Mon, Mar 20, 2017 at 6:26 AM, Rob Godfrey <[email protected]> wrote: > On 20 March 2017 at 14:23, Adel Boutros <[email protected]> wrote: > > > Hello Rob, > > > > > > I just saw your mail and I confirm what you propose works correctly. I > > opened your link via chrome and it downloaded the correct package this > time. > > > > > > Please note this is the case for all the artifacts proposed on the > website. > > > > > > So are you going to fix them all? > > > > I'm not sure what the intent of the page is, whether the idea is to allow > people the (interactive) choice to download from one of several alternative > mirrors, or whether it should give the impression of just directly > downloading from the mirror that Apache deems the most appropriate. > > Anyone with more involvement in the website want to comment here? > I don't have a strong preference. I think the behavior we have is the one Qpid started out with way back. For some, it may be attractive to select a mirror. For me, I'd let the CGI script do it. Data points: httpd gives you a mirrored download directly, but it's using its own CGI script. Lucene goes to the mirror list. ActiveMQ gives you the download directly. I still don't understand what combination of OS and browser versions is giving Adel the wrong content type for the mirror list page. It doesn't reproduce for me.
