User:Rhelmer:Graphserver Progress

http://graphs-stage.testing/graph-2.html

Bug 601796 - implement new graphserver front-end design

todo

• needs error handling
• review and rethink current JS code
  • look for more places to reuse or adapt existing code
• caching is per-session right now, should be once per session + ifmodifiedsince
• Dashboard
• Custom chart
  • toolbar selects not working - see question #1
    • "Last 7 Days" - see question #2
    • "Running Time" - see question #3
    • "Zoom In/Out" - see question #4
    • "Show Changesets" - almost working, look at how catlee's cruncher does it
      • hardcoded "mozilla-central"
    • "Export CSV" - think graphserver already does this, find the link
    • "Link" - use custom bit.ly URL "mzl.la"
    • "Embed" - see question #5
  • in tooltip on hover/click point
    • hardcoded "mozilla-central"
    • hg rev link should open in new tab
  • hardcoded default zoom for main plot view
  • overview shows all available data - see question #2
    • 90 days
  • incoming links
    • current URL should change when
      • data added/removed
      • range selection modified
    • not sure if they could/should be backwards compatible
  • resizing too small causes overview to go away forever
• Add data
  • "Select Platform" constraint should check branches and tests
  • multi-select not working
  • hardcoded "Add n Data Series" on button
  • hardcoded "Pages" selection
  • adding the same data series twice causes visual glitch, should detect

questions

1. how should these selects be finished and hooked up?
2. should plot and overview show same time range?
3. define, examples?
4. define the zoom in/zoom out buttons (exactly how are overview and plot modified)?
5. define embed button, is this paste-able HTML/JS?
6. why no points on the graph?
7. loading indicator?
8. links - should they always open in new tabs?

suggestions

1. put individual test/branch/platform map in "test run" results
2. tests(!), interactions get really complicated with power-users
3. test with real data, these datasets are too small