Regarding the IPV6 issue in JBPAPP-2941, there was a change in JBNAME-25 to support an alternate syntax using '@' as the host/port separator: "jnp://[3ffe:ffff:100:f101::1]@1099" Does this not work for the EAP usage?

Is it really necessary to introduce a new notation? Under the assumption that the address passed in as a Context.PROVIDER_URL is well formed, its easy to separate the host from the (optional) port. With a few extra lines of code to do this, the original failures no longer appear when running the testsuite. In fact, the whole of the EAP 4.2.GA_CP testsuite now seems to run clean against IPv6, with the fixes to JNDI, aside from the clustering tests where there seems to be a problem with explicit IPv6 addresses and the use of cookies (https://jira.jboss.org/jira/browse/JBPAPP-3008). David M. Lloyd wrote: > On 10/26/2009 06:39 PM, Scott Stark wrote: > >> Regarding the IPV6 issue in JBPAPP-2941, there was a change in JBNAME-25 >> to support an alternate syntax using '@' as the host/port separator: >> >> "jnp://[3ffe:ffff:100:f101::1]@1099" >> >> Does this not work for the EAP usage? >> > > A couple problems with this fix - first, it uses InetSocketAddress as a > hash key, which contains InetAddress, which can trigger DNS lookups on > equals/hashCode; you might get away with this if you only store addresses > which you know to be fully resolved, but it's still a bit iffy if you ask > me. I don't know if this was among the fixes that Jason made for the > InetAddress-as-key situation. > > Second, this isn't really RFC compliant at all and will cause URI parsing > to crap out. Is there a problem with using the RFC syntax? I couldn't > find any discussion in the JIRA but I might just be blind > > - DML > _______________________________________________ > jboss-development mailing list > jboss-development(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/jboss-development > _______________________________________________ jboss-development mailing list jboss-development(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-development

I'm fine with updating the logic as indicated in the patch to look for the last ':'. The JBPAPP-2941 issue is marked against EAP 5, but the patch is against naming in 4.3. The naming project used in EAP 5 is coming from the standalone naming release. Does this need to be fixed for EAP? If so, I'll integrate the change in https://jira.jboss.org/jira/browse/JBNAME-39. I don't suppose it hurts to keep the parsing to strip leading and trailing [] even though they don't need to be used?

David M. Lloyd wrote: > That one looks to me like "not a bug", since the [] are only for URI usage > and are not really valid for IPv6 addresses in general. I'd say the > nonstandard use of [] and @ is the unnecessary notation, though I may be > missing something else here. > > All I can say about the cookie thing is what RFC 2965 says: "Host name (HN) > means either the host domain name (HDN) or the numeric Internet Protocol > (IP) address of a host. The fully qualified domain name is preferred; use > of numeric IP addresses is strongly discouraged." It doesn't seem to > mention IPv6 at all... > > Tracking back the cookie exception to the underlying service, and requiring > that service to be configured with a proper domain name, seems like the > logical step to me. > > (I'll copy this into the ticket...) > > - DML > > On 10/26/2009 08:05 PM, Richard Achmatowicz wrote: > >> Is it really necessary to introduce a new notation? Under the assumption >> that the address >> passed in as a Context.PROVIDER_URL is well formed, its easy to separate >> the host from the >> (optional) port. With a few extra lines of code to do this, the original >> failures no longer appear >> when running the testsuite. >> >> In fact, the whole of the EAP 4.2.GA_CP testsuite now seems to run clean >> against IPv6, with the fixes >> to JNDI, aside from the clustering tests where there seems to be a >> problem with explicit IPv6 addresses >> and the use of cookies (https://jira.jboss.org/jira/browse/JBPAPP-3008). >> >> David M. Lloyd wrote: >> >>> On 10/26/2009 06:39 PM, Scott Stark wrote: >>> >>> >>>> Regarding the IPV6 issue in JBPAPP-2941, there was a change in JBNAME-25 >>>> to support an alternate syntax using '@' as the host/port separator: >>>> >>>> "jnp://[3ffe:ffff:100:f101::1]@1099" >>>> >>>> Does this not work for the EAP usage? >>>> >>>> >>> A couple problems with this fix - first, it uses InetSocketAddress as a >>> hash key, which contains InetAddress, which can trigger DNS lookups on >>> equals/hashCode; you might get away with this if you only store addresses >>> which you know to be fully resolved, but it's still a bit iffy if you ask >>> me. I don't know if this was among the fixes that Jason made for the >>> InetAddress-as-key situation. >>> >>> Second, this isn't really RFC compliant at all and will cause URI parsing >>> to crap out. Is there a problem with using the RFC syntax? I couldn't >>> find any discussion in the JIRA but I might just be blind >>> >>> - DML >>> _______________________________________________ >>> jboss-development mailing list >>> jboss-development(a)lists.jboss.org >>> https://lists.jboss.org/mailman/listinfo/jboss-development >>> >>> >> _______________________________________________ >> jboss-development mailing list >> jboss-development(a)lists.jboss.org >> https://lists.jboss.org/mailman/listinfo/jboss-development >> > _______________________________________________ > jboss-development mailing list > jboss-development(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/jboss-development > ------------------------------------------------------------------------ _______________________________________________ jboss-development mailing list jboss-development(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/jboss-development