Community
Participate
Working Groups
Created attachment 237191 [details] Implementation of jslp to ipv6 The patch has the implementation of jslp to ipv6.
Hi Jamal, the patch leaves compiler errors in my workspace: Description Resource Path Location Type The method initMulticastSocket(String) in the type SLPCore is not applicable for the arguments () Activator.java /ch.ethz.iks.slp/src/main/java/ch/ethz/iks/slp/impl line 43 Java Problem Destination cannot be resolved to a variable SLPCore.java /ch.ethz.iks.slp/src/main/java/ch/ethz/iks/slp/impl line 928 Java Problem Destination cannot be resolved to a variable SLPCore.java /ch.ethz.iks.slp/src/main/java/ch/ethz/iks/slp/impl line 878 Java Problem Destination cannot be resolved to a variable SLPCore.java /ch.ethz.iks.slp/src/main/java/ch/ethz/iks/slp/impl line 872 Java Problem The method getMessageType(RequestMessage) is undefined for the type SLPMessage SLPCore.java /ch.ethz.iks.slp/src/main/java/ch/ethz/iks/slp/impl line 866 Java Problem
Created attachment 237220 [details] Implementation of jslp to ipv6 Updated the patch...
Created attachment 237221 [details] Implementation of jslp to ipv6
Created attachment 237223 [details] Implementation of jslp to ipv6
Hi Jamal, thanks for the updated. Please do not forget to sign the CLA [1] which is required before the code can be contributed to Eclipse. Btw. the second copyright holder P Sowjanya has to sign it as well. M. [1] http://www.eclipse.org/legal/CLA.php
Once you have signed the CLA, I can push the review to gerrit. mkuppe@foo:/usr/local/src/ECF/org.eclipse.ecf(master) $ git push review Counting objects: 46, done. Delta compression using up to 4 threads. Compressing objects: 100% (14/14), done. Writing objects: 100% (24/24), 3.53 KiB, done. Total 24 (delta 10), reused 0 (delta 0) remote: Resolving deltas: 100% (10/10) remote: Processing changes: refs: 1, done remote: ---------- remote: Reviewing commit: commit a1a71f87ad181653fc52222b869d46f378fbc072 1383723260 ----sp remote: Authored by: Mohammad Jamal MohiUddin <mjmohiuddin@cdac.in> remote: remote: error: The author does not have a Gerrit account. remote: All authors must either be a commiter on the project, or have a current CLA on file. remote: Please see http://wiki.eclipse.org/CLA remote: To ssh://mkuppe@git.eclipse.org:29418/ecf/org.eclipse.ecf.git ! [remote rejected] HEAD -> refs/for/master (The author must register with Gerrit.) error: failed to push some refs to 'ssh://mkuppe@git.eclipse.org:29418/ecf/org.eclipse.ecf.git'
I have signed the CLA
CQ filed https://dev.eclipse.org/ipzilla/show_bug.cgi?id=7708 FTR: http://www.ietf.org/rfc/rfc3111.txt
OK, we now have a gerrit change reg where we can do the review: https://git.eclipse.org/r/#/c/18618/
With this patch applied, I get: INFO: Total number of tests executed: 82 INFO: Successful tests: 46 INFO: Failed tests: 36
https://build.ecf-project.org/jenkins/job/C-421063-discovery.slp.feature/7/console (unfortunately not publicly available)
The test suites can be found in the ECF git repository [1]. Jamal, correct me if I'm wrong, but you don't know how to use OSGi, do you? [1] http://git.eclipse.org/c/ecf/org.eclipse.ecf.git/tree/tests/bundles
OpenSLP v2.x adds IPv6 support [1]. Unfortunately, none of the common Linux distributions come with OpenSLP 2.x yet that could be used for compatibility testing (Fedora rawhide has 2.x [2], in Debian/Ubuntu openslp is orphaned and thus unlikely to see a 2.x package anytime soon). [1] http://www.openslp.org/download.html [2] https://apps.fedoraproject.org/packages/openslp-server/builds/undefined/packages// [3] http://packages.qa.debian.org/o/openslp-dfsg.html
Created attachment 238155 [details] When running tests with -Dnet.slp.interfaces=[::1], almost all tests fail.
Created attachment 238156 [details] Test for Ipv6 error
I think Error is mainly because when we are sending a packet by setting the socket interface to ::1. This is only for IPv6 but when we send a packet with the interface being set to 127.0.0.1 . There is no error. I dont have much more info on it..
(In reply to Mohammad Jamal MohiUddin from comment #16) > I think Error is mainly because when we are sending a packet by setting the > socket interface to ::1. This is only for IPv6 but when we send a packet > with the interface being set to 127.0.0.1 . There is no error. I dont have > much more info on it.. What operating system are you on? From a quick internet search it might be possible that this problem is Linux only.
Yes.Iam using ubuntu 12.04..
What about the OS which you are using?
(In reply to Mohammad Jamal MohiUddin from comment #19) > What about the OS which you are using? I'm on Linux too, so it the build/test machine.
FWIW I have create a second ipv6 related launch config and the build machine now runs v4 and v6 sequentially (with v6 failing due to aforementioned ::1 problems)
The test failures have nothing to do with Java: mkuppe@foo:~# ping6 ff02::123%lo connect: Network is unreachable mkuppe@foo:~# ping6 ff02::123%eth0 connect: Network is unreachable mkuppe@foo:~# ping6 ff02::123%wlan0 PING ff02::123%wlan0(ff02::123) 56 data bytes
Yes , i also dont agree that it has to do something with java... ping6 ::1%lo also doesnot work
(In reply to Mohammad Jamal MohiUddin from comment #23) > Yes , i also dont agree that it has to do something with java... > ping6 ::1%lo also doesnot work Try instead mkuppe@foo:~ $ ping6 -I lo ::1 PING ::1(::1) from ::1 lo: 56 data bytes 64 bytes from ::1: icmp_seq=1 ttl=64 time=0.052 ms
I'm preparing ECF 3.8.1/Luna for final release, and would like to find out about the status of this effort to add IPv6 support. Markus has indicated in another forum that it's not ready for release. Mohammad...where would you say things are? And what are plans for moving forward? Are things stuck somewhere? Are there things that could be done to help move along? Please let know asap...as the deadlines for ECF freeze are approaching 5/19/2014. Thanks.
There are two problems which Iam facing now 1. If we try to send the packet with source address as loopback address(::1),we will get an error as the OS underlying is implemented according to the RFC 4291 which does not allows to send the ipv6 packet with loopback address you can check the linux or any other OS source code. So any tests which uses ::1 loopback address will fail and which is not the same for IPv4 loopback address 127.0.0.1. 2. Tests which uses the IPv6 linklocal address successfully passed on my system as well as markus, but it fails on the build machine,dont know why ,there may be some network issues on the build machine
