[Opensim-dev] (proposal) QA, Release and other tought

[Dahlia Trimble]

I just finished a full regression test of my new meshing code. It involves
visually inspecting and measuring several meshes in a 3d display environment
(currently blender) and collision testing over 400 prims in my test region.
I'd love to have an automated way to do this, but at this point I'm at a
loss how to do it with my current toolset. I guess if physical object
collisions could generate accurate position data, I could set up scripts to
rez physical objects, let them collide, record the results and compare them
with expected results, but so far I haven't found the time to set something
like that up, nor would I trust it considering the inaccuracies I've found
in SL's collisions.  For now, a full regression takes 3-4 hours of my time,
but fortunately an abbreviated test will suffice more often than a full one.

BTW, the test passed! :D  I'll be committing the code as soon as I can clean
up the interface to ODE.

On Tue, Aug 26, 2008 at 6:39 PM, Mike Mazur <mmazur at gmail.com> wrote:

> Hi,
>
> On Tue, 26 Aug 2008 09:24:21 -0400
> Sean Dague <sdague at gmail.com> wrote:
>
> > Honestly, I think any formal manual testing is mostly wasted effort at
> > this point.  If these things were part of something automated, that
> > would be a different story.
>
> I disagree. If we can get people dedicated to run through some manual
> tests at regularly scheduled intervals, I'm sure they can provide
> benefits. For instance, it can regularly be confirmed which releases
> behave as expected when the tests pass, which is better than the
> current "Hey guys, what recent SVN revision was stable so we can tag
> 0.5.10?"
>
> The key lies in having somebody around who is willing to do this. In
> fact, they can get started right away, take a revision they like on a
> Friday, run through the tests and post the results on the Wiki. If
> after a while they realize nobody cares they'll stop and we'll just end
> up where are now.
>
> Having said all that, I'm not going to do it :)
>
> On Tue, 26 Aug 2008 17:58:24 +0200
> Lc <lcc1967 at gmail.com> wrote:
>
> > yes but no automatic test will find any inventory or assets issues or
> > other wearable prims lost in the metagrid...
>
> There are some tasks for which automated tests are difficult to design.
> Things like "Does the avatar sit properly on this prim?"
>
> Also, a quick suggestion to your test plan -- you need to remember the
> different set ups that people can have: two regions on one machine, two
> regions on separate machines, etc. Most tests will need to be performed
> on all combinations of setups.
>
> Mike
> _______________________________________________
> Opensim-dev mailing list
> Opensim-dev at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/opensim-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20080826/befaf443/attachment-0001.html>