On Tue, 2004-03-30 at 19:40 -0800, Stas Bekman wrote:
Philippe M. Chiasson wrote:
On Tue, 2004-03-30 at 15:25 -0500, Geoffrey Young wrote:
Something like this will most likely work for everyone (ipv4/v6) with the only difference is that the test servers are not listening to localhost anymore and could be reached from oustide machines (do we care?)
What happens in the following situation:
One machine, two NICs, two IPs 10.0.0.1 and 10.0.0.2.
If I say 'Listen *:8529' it'll try to listen on both IPs.
What happens if 10.0.0.2:8529 is already taken and 10.0.0.1:8529 is available? Will it fail to start, saying that the port is busy?
That's what it'll do, yes. The test framework will detect this early and report that :
port 8529 is in use, cannot determine server pid to shutdown
I guess it's OK, we can always change that in the future, if we encounter problems with this approach.
I actually find that a nice feature, as it means you can use t/TEST -start to start up your test config then, say, hit it from a browser on a different box. as it is now, you have to hand-edit the generated httpd.conf for that (or at least I did)
Well, I agree with you, but other people might argue that it is somewhat of a security risk to start up a server and _not_ listen strictly on localhost.
But, yes, I do think that Joe's suggestion is the best one and will keep tests passing until LWP becomes IPv6 aware.
Untill we have a better solution (proper detection mechanism), by all means go with it, Philippe.
You mean the one with Listen 0.0.0.0:port ?
Yes, please.
-- __________________________________________________________________ Stas Bekman JAm_pH ------> Just Another mod_perl Hacker http://stason.org/ mod_perl Guide ---> http://perl.apache.org mailto:[EMAIL PROTECTED] http://use.perl.org http://apacheweek.com http://modperlbook.org http://apache.org http://ticketmaster.com
