[Qgis-developer] Rethinking the testing and release procedure of QGIS

[Sandro Santilli]

On Fri, Jul 08, 2011 at 05:10:45PM +0200, Marco Hugentobler wrote:

> > branch
> > write tests
> > work
> > request for merge
> > integrate
> 
> For me, this workflow is fine in case of new features. For smaller changes or 
> quick bugfixes, it seems unhandy to always write a unit test for it. 

I think tests for bugfixes are really important as they become a warranty
that the bug you'll be spending time on won't come back in the future.
I don't have a long file of qgis contributions, but the few (single?)
bug I contributed a patch for appeared a few weeks later (the line
orientation detection code). It's been _very_ frustrating.

If providing a test is unhandy it may mean writing a test is too hard 
and should be made simpler. Writing a test for a bug should really be
the first stage of bug handling, it's very like asking the reporter
"can you reproduce the bug?". Only translating that reproducability
in code, which then becomes _the_ test for that bug, linking to ticket.

--strk; 

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html