track and present status of individual tests (was: test infrastructure ideas appreciated ...)

Bjoern Michaelsen bjoern.michaelsen at canonical.com
Thu Jun 18 02:57:22 PDT 2015


Hi,

On Wed, Jun 03, 2015 at 02:33:23PM +0100, Michael Meeks wrote:
> 	Constructive thoughts appreciated in reply here.

This has some synergies with the dashboard proposal[1] but it is sufficiently
distict to be included as an ideas on its own:

We lack a good representation of test status. tinderbox.libreoffice.org is only
very basic and overloads one view trying to show way too much at one:
- e.g. http://tinderbox.libreoffice.org/MASTER/status.html shows:
    - when something broke
    - where (on which machine) something broke
      (which is already creates a ~useless table a fullscreen wide on
      my 30" 21:9 screen at default zoom levels -- quite an "achievement")
  however it doesnt show:
    - _what_ (which test or module) broke

The "what" is the most important information for a developer checking if he
broke something, followed by the "when", while the "on which machine" is
less relevant in most cases.

If we really want to value tests we should be able to present a view on our
automated testing that shows what broke first -- and then allows ivestigating
the when and where from there. Take the "Test Statistics Grid" at the end of:

 https://jenkins.qa.ubuntu.com/view/vivid/view/AutoPkgTest/

as an starting point, showing e.g. the status and the stability of each and every
CppunitTest_ JunitTest_ and PythonTest_ individually.

Best,

Bjoern


[1] http://nabble.documentfoundation.org/TDF-Grant-Request-Proposal-LibreOffice-project-dashboard-quot-All-about-LibreOffice-quot-td4151652.html


More information about the LibreOffice mailing list