wiki:SprintPlanning20120112

We had a sprint planning session at the face to face meeting at the ISC office in Redwood City.

In attendance (in no particular order) were:

Stephen
Michal
Jelte
Haikou
Cathy
Jeremy
Shane
Larissa
Joao
Jinmei

Past Sprint

#1546 regression due to socket creator

should be merged before the release

#1518 was an investigation, is done

About 50 closed

Goals

  • Socket creator
    SocketCreator tickets (just) done
  • bugs
    All but 1 done
  • DDNS if available
    Only 1 done

Q: How was the sprint?
Michal: Kind of lonely.
Jinmei: It was unusual, but not so bad considering we completed the socket creator.

#1452 was estimated as 5, ended up being 50.52
Jinmei: Was handling low-level stuff, so easy to cause portability issues
Jinmei: Also required some design-level stuff, which caused it to take more time. This is generally the case. Should be conservative about estimation if they involve low-level interfaces or design things.

#1462 Should have been easy, but was harder because it was a new developer. We need to take that into account because we have more new developers.
The whole workflow is not clear, so it takes longer.

Statistics...
#1399 was done but not reviewed
#1400 still being worked
Do we include these?

#1447 Jinmei proposes to carry over and have 1 additional person look at it, will look at this in an ad-hoc design review meeting on 2012-01-17

Next Sprint

Proposed:

  • socket creator followups
    still missing some tickets (#805 documents this)
    Jinmei proposes create the tickets in this sprint, then discuss it for the next sprint.
  • profiling
    Some already done. Should we do it properly?
    NXDOMAIN, realistic tests (not querybench), realistic data

Tickets:

  • document how to do profiling
  • create a set of benchmarks (for publishing, and profiling)
    document what we are comparing
    include scripts, documentation into git
  • changing exception to assert()
  • experiment with converting tolower() on start rather than during check
  • experiment with disabling name compression (after exception to assert())
  • experiment with multi-core hack
  • #1431 (closest enclosure proof)
  • look at BIND 9 NSEC3 code, look at how to break down logic into our tasks
  • parsing functionality? #1549, #1551 (discuss tomorrow)

Defect tickets:

Other actions:

  • real data for profiling (F-root)
  • real data for profiling (TLD)
  • ready benchmark scripts

Note: for performance tickets, we'll run the branch on our performance baseline before merging

We skipped estimates. Doing a short sprint.

Last modified 6 years ago Last modified on Jan 12, 2012, 5:17:16 PM