Am 17.02.2015 um 12:55 schrieb Lukas Krejci
<lkrejci(a)redhat.com>:
Do you envision these being generated on the feed-side or server-side?
No idea :)
Just because we have a 32char string that has a name of universally unique
does not guarantee that it is unique. Each client/feed could name its toplevel
resource "main....32" :-) so it may make sense to have some collision detection
in place if feeds send what they want.
I also did not imply we *have to* change to numbers.
Fact is though that in many places the randomly long ids waste space
and bandwidth.
Especially when using a cache with those long ids as key and a payload
of a single byte.