I guess I was wrong in expecting a user agent header would always be
avaible. So WebClientInfo should accept null, and then in init() do
nothing.
Another thing I'm wondering is whether this is a quirk, or whether you
can expect a client that doesn't send a user-agent header to never
send it. Cause if is a quirk, and does send it with a next request, we
might want to try to reconstruct the WebClientInfo object. That's a
bit more complicated to achieve but doable. We just need to decide if
that's the best way to go.
Eelco
On 4/13/07, Martijn Dashorst <[EMAIL PROTECTED]> wrote:
On 4/13/07, Al Maw <[EMAIL PROTECTED]> wrote:
> It's rare, but sometimes you get real requests that don't have this
> header. I could have sworn I opened a bug for this a while back.
Just found several in our logs. They seem to come from a 'watch dog' service.
> Yes, IMHO this is broken - it causes failures for a small minority case
> that's really not obvious. At the very least, the WebClientInfo
> constructor should throw a checked ClientInfoNotAvailableException or
> similar.
What goes wrong when the User Agent is not set, or set to "unknown"
("n/a", "", "-") when the header is not submitted?
Martijn
--
Learn Wicket at ApacheCon Europe: http://apachecon.com
Join the wicket community at irc.freenode.net: ##wicket
Wicket 1.2.5 will keep your server alive. Download Wicket now!
http://wicketframework.org
