Jump to content

Meeting Notes/2016-02-24 Android Phlogiston

From mediawiki.org

Follow up from last meeting

[edit]

Joel to create histogram of completed, pointed Android stories (T124246)

[edit]

Reviewed current histogram. Is it telling us anything

Dmitry: what would this tell a product owner?

Max: Could use it to assign average value to unpointed tasks, so that we can use them for point-based forecasting, and then figure out how long it would take to burn down the backlog.

Dmitry: in our meetings, it feels ilke 5 is discouraged, better to break things down into smaller points.

Max: We could just use count forecasting, since everything is roughly the same size. Would therefore recommend we don't assign a default point value, and instead rely on count for metrics.

Joel: no big points in histogram, suggesting you do small breakdown only. When you do eg a 2 month project, that ends up being 80 tasks, do you have the list of 80 at the start, or do you discover them week by week?

Dmitry: discover week by week, though we do have epics.

Joel: Judging from the histogram, the epics must be 0-point. If you want more accurate forecasts, probably need to make high-point epics.

Dmitry: not useful to forecast the entire backlog; it has hundreds of items.

[edit]

Dmitry: We do use t-shirt sizing. It would be helpful to have historical data about how many points/stories ultimately comprised old epics.

Will have the data at the end of this quarter. How do we get it for old data?

Max: would have to do retroactive parenting - assign tasks retrospectively to epics.

Joel: Suggestion: Do a planning session for upcoming features, in which you break out all of the epics and give them all a single value like 20 or 40 (i.e., don't bother to measure big vs small epics, count them all the same). Then, spend Q4 practicing shaving and exploding. This will provide a pilot of doing point-based forecasting, and also will generate useful historical data (about # of tasks per epic) during Q4.

Joel to report on # of tasks done per dev per week

[edit]

don't have this in data but can estimate.

Dmitry to close Done stories and continue doing so, rather than wait for end of Sprint.

[edit]

Have been doing that.

Dmitry to assign existing tasks to Sync Saved Pages and Analytics... milestones via "Blocks" relationships

[edit]

Done. More tasks coming in, getting assigned.

In Phlogiston, they only have 2-5 tasks. can see a few more in Phabricator, up to 10. Not sure if this indicates a problem with Phlogiston data or not.

Dmitry to do work breakdown for Gather Collections

[edit]

no discussion.

Next Step

[edit]
  1. Max/Dmitry Plan a work breakdown session for Q4.
  2. Max: Do historical checking for Q3 t-shirt sizing to see how many points were assigned.
  3. Joel/Max to check if Phlogiston category counts are accurate.