David Timms <dtimms(a)iinet.net.au> wrote:
On 14/03/16 13:12, Nico Kadel-Garcia wrote:
> If it goes stale quickly, I'd consider it useless. Git was not
> designed to support magical dancing URL's, that kind of instability
> takes actual planning and work to inflict on the software repository
> and on its users..
I think it's quite reasonable; they have the complete content stored as
git diffs, they then do-not need to store a zip archive after every
commit ever made. Instead, just create one when someone asks for it...
Sure, creating a file when someone asks for it is fine, but that's not
what they do, according to your description. Your problem is that you
have to first create the file with one request, and then ask for it
with another request.
There wouldn't be a problem if it were all done with a single URL. The
server could perfectly well create the file on the fly and send it as
the response to the first request.
If it's done with HTTP redirection, where the first GET request creates
the file and returns a 303 See Other response with the URI to the file,
then that's unnecessarily complex but should still work with any
reasonably competent download utility. Then you only need to find out
the correct first URI and use that in your spec file.
browser, then that's not reasonable. That's being difficult on purpose.