Umm, well yes, that's true, I am asking you to distribute the binaries, but you know, to be fair, it is your project, plus you already do host them using archive.apache.org <http://archive.apache.org/>, it's just slow.
I'm a bit confused why you would expect me to host them instead of you. I'm also a bit confused how I managed to offend you there, I really wasn't trying to upset anyone, just asking for help. > On 4 Apr 2016, at 17:36, sebb <[email protected]> wrote: > > On 4 April 2016 at 17:29, Oliver Lloyd <[email protected]> wrote: >>> Why do you want to use the older versions? >> >> It's not so much that i want to use older versions. The problem is if you >> don't let people specify the version then they will get whatever is either: >> a) current or b) I give them. In both cases whenever these things change >> then next time the script is run (when they next test) then they will >> suddenly have a different version of jmeter which could, conceivably, result >> in different results or even problems running their scripts. > > That should not be a problem if you host the binaries. > >>> Why will it break? >> >> If I host the binary files myself then the risk is a new version of jmeter >> will be released that is not available in my repo (because I haven't copied >> it over) which will mean the script, in its current form, would fail were >> someone to specify the new version. Sure, I could check for a 404 response >> and return a useful message but this really starts to over complicate things >> and is quite a step backwards. Plus I'd essentially be managing my own >> distribution server for jmeter which I'd rather not do. > > Sorry to be so blunt, but instead you are effectively asking someone > else to manage the distributions server .. > >> Bintray does seem to be made for exactly this problem. It's also free for >> open source; do you think it is possible to set something up? > > This is completely out of scope for the JMeter project. > >> >> >>> On 4 Apr 2016, at 17:03, sebb <[email protected]> wrote: >>> >>> On 4 April 2016 at 15:28, Oliver Lloyd <[email protected]> wrote: >>>> Thanks Vladimir, I can pull out the host from this using: >>>> >>>> preferred=$(curl -s 'http://www.apache.org/dyn/closer.cgi?as_json=1' | >>>> grep "preferred" | cut -d ':' -f3 | cut -d'/' -f3) >>> >>> No need, the URL I gave is equivalent. >>> >>>> But there's another problem: The mirrors only seem to host the latest >>>> version of JMeter. >>> >>> Yes, that is intentional. >>> >>>> Older versions, as far as I can tell, only live on archive.apache.org >>>> <http://archive.apache.org/> >>> >>> Yes. >>> >>> or www.apache.org <http://www.apache.org/>. >>> >>> Huh? No. >>> >>>> I know that these servers are not meant to serve binaries like this but I >>>> keep ending coming back to using them. >>>> >>>> I thought about trying to host older versions myself but it will break >>>> each time a new release is made. >>> >>> Why will it break? >>> >>> And why do you want to use the older versions? >>> >>>> Another solution I considered was only allowing the latest version to be >>>> used but that would mean an uncontrolled upgrade for everyone each time a >>>> release is made, which isn't correct. >>>> >>>> >>>>> On 4 Apr 2016, at 13:34, Vladimir Sitnikov <[email protected]> >>>>> wrote: >>>>> >>>>> curl 'http://www.apache.org/dyn/closer.cgi?as_json=1' >>>>> gives something like >>>>> { >>>>> "backup": [ "http://www-eu.apache.org/dist/";, >>>>> "http://www-us.apache.org/dist/"; ], >>>>> "cca2": "ru", >>>>> "http": [ "http://apache-mirror.rbc.ru/pub/apache/"; ], >>>>> "path_info": "", >>>>> "preferred": "http://apache-mirror.rbc.ru/pub/apache/"; >>>>> } >>>>> >>>>> Then you grep for preferred somehow, then construct the proper URL. >>>>> >>>>> Does that work for you? >>>>> Vladimir >>>>> >>>> >>> >> >
