Kernel Projects weekly meeting minutes

Michael Vorburger <vorburger@...>


here are the minutes of the weekly Kernel Projects meeting from today:


#opendaylight-meeting: kernel-projects

Meeting started by vorburger at 16:56:56 UTC (full logs).

Meeting summary

  1. Attendees please #info in (vorburger, 16:57:07)
    1. vorburger skitt brendan jmorvay Martin Sandberg (vorburger, 17:02:48)

  2. Agenda Bashing (vorburger, 17:02:59)
    1. Daya raises Martin from E. hit some NETCONF issues in ONAP using ODL Oxygen (vorburger, 17:03:59)
    2. jmorvay saw JIRA from Martin, it's about TLS support in netconf clustered implementation (vorburger, 17:05:07)
    3. jmorvay thinks porting non-clustered to clustered should be possible, he'll have a look (vorburger, 17:05:17)
    4. (jmorvay, 17:07:09)

  3. skitt Ongoing version bumps (vorburger, 17:07:51)
    1. need a new 4.0.9 but has an issue which rovarga is looking into helping to fix (vorburger, 17:09:02)
    2. current situation: all projects have merged bump patches and build fine, but distribution is broken (vorburger, 17:10:36)

  4. Tell based? (vorburger, 17:11:39)
    1. Daya is asking about where we are with regards to switching based Tell based (vorburger, 17:11:55)
    2. rovarga says that tpantelis was looked at the CSIT for Tell, and things that it looked like it was pretty much working (vorburger, 17:12:33)
    3. the "CSIT" here refers to Controller CSIT, not full leaf project (like netvirt). Days says jamoluhrsen said in the netvirt meeting he'll run test (vorburger, 17:13:42)
    4. rovarga suggests starting a new thread on a list like controller-dev to track this (vorburger, 17:14:14)
    5. Daya points to an existing JIRA in netvirt which tracks current status (vorburger, 17:14:43)
    6. (vorburger, 17:15:56)

  5. Can the RESTCONF draft implementation be dropped? (vorburger, 17:17:38)
    1. jmorvay raises the idea of dropping the code of the draft implementation (vorburger, 17:18:14)
    2. main effort seems to be adjusting the CSIT tests, not removing the Java code (vorburger, 17:18:49)
    3. step 1: duplicate netconf.git CSIT jobs such that we have one set talking bierman02, the other talking RFC8040 (rovarga, 17:24:32)
    4. step 2. declare bierman02 endpoint "not recommonded" or deprecated (i.e. use RFC8040 if at all possible) (rovarga, 17:25:17)
    5. 3. migrate all of CSIT to use RFC8040 (which *will* find bugs) (rovarga, 17:25:33)
    6. step 4. deprecate bierman02 by not installing it by default (rovarga, 17:26:16)
    7. step 5. remove the code (rovarga, 17:26:29)
    8. steps 2 and 4 need to happen on release, i.e. must be documented and release noted (rovarga, 17:27:45)

  6. Sodium planning (vorburger, 17:29:04)
    1. (rovarga, 17:29:21)
    2. skitt mentions some work in planning odlparent, mostly clean up and version bumps with breakages (vorburger, 17:30:38)
    3. rovarga skitt discuss the need for breaking version changes .. "gain vs. pain", incl. e.g. the Jersey version bump (vorburger, 17:31:16)
    4. Nobody is jumping up to tackle the Jersey version ump (vorburger, 17:33:05)
    5. yangtools and mdal has nothing major planned (vorburger, 17:33:30)
    6. controller as MRI? - let's discuss this in a future weekly meeting when we have Tom (vorburger, 17:37:29)
    7. yangtools does not have major changes ready, the cost/benefit is simply not there to spend full-time on developing and stabilizing things for downstream consumption (rovarga, 17:37:43)
    8. (jmorvay, 17:38:17)
    9. mdsal has some stuff which would warrant a major release (rovarga, 17:38:34)
    10. I've opened the issue to track RFC 8040 in netconf CSIT (jmorvay, 17:38:48)
    11. 1) remove DataObject.getImplementedInterface() which was forgotten when we have eliminated getKey() and getAugmentation() (rovarga, 17:40:40)
    12. this is low-impact on downstreams, as it affects only generated code and controller/netconf tests (rovarga, 17:41:51)
    13. IFF we go for Java 11, that would warrant a new major odlparent (vorburger, 17:41:58)
    14. 2) change union generation, limited fallout in downstreams (rovarga, 17:42:10)
    15. 3) change 'list foo { key bar }' with ordered-by=system from List<Foo> to Map<FooKey, Foo> (rovarga, 17:44:13)
    16. this is very useful to downstreams, but will cause major compilation breakage (rovarga, 17:45:19)
    17. hence mdsal will probably shoot for a major release containing 1) but little else (rovarga, 17:45:37)
    18. odlparent may need a major release contingent of whether the TSC decides to make Sodium a Java 11-only release (because of source=8 -> 11 change) (rovarga, 17:46:13)
Meeting ended at 17:49:17 UTC (full logs).

People present (lines said)

  1. vorburger (34)
  2. rovarga (17)
  3. odl_meetbot (4)
  4. jmorvay (4)
Generated by MeetBot 0.1.4.

Michael Vorburger, Red Hat
vorburger@... | IRC: vorburger @freenode | ~ =