[haiku-qa] Re: Haiku Alpha 4 testing and SBTM
- From: Ryan Leavengood <leavengood@xxxxxxxxx>
- To: haiku-qa@xxxxxxxxxxxxx
- Date: Wed, 8 Aug 2012 00:53:43 -0400
On Tue, Aug 7, 2012 at 11:28 PM, James Goss <jjgoss@xxxxxxxxxxxxx> wrote:
>
> Are there more "established" tools like Session Tester - it doesn't look like
> it gets a lot of love, the last posts on its forum are from early 2010. This
> makes me wonder if it is stable enough.
While I tend to agree with you about not using out of date software,
in this case it may just be that the tool does the job it needs to do
and didn't need more features :)
> I had a crazy idea (remember, my user name *is* krazykat - mostly due to my
> Siamese helper cat) that maybe we could get some extra mileage out of all
> this work and build something totally new, like integral system checks for
> Haiku that include user runnable tests to check for stability, proper
> function, and what not. I wonder if the development team would be hip to
> this? Or does this seem like a lame idea. (I'll blame the Siamese ...)
I can't speak for the whole team, but I'd imagine this is something we
would welcome. Obviously we need to think about it more and figure out
exactly what that means.
Just so you know, we do have some (now probably out-of-date) unit
tests, which I have a task in my bug list about updating. Though that
is more of a development tool and for QA we obviously are thinking
about higher-level functional tests.
I said before in the forum that I would like to see just how automated
we can make functional tests. BeOS and now Haiku is very scriptable,
and I think that could even be extended. So for example I could see
doing things like sending an application messages, mouse messages
and/or keyboard messages, and then either asking for its state, or
dumping it to a BMessage (called Archiving) and then examining its
state. That was my very rough plan for experimenting with automation.
In the simpler case we could probably still drive the applications
with scripting through various test scenarios, and the behaviors could
be checked by a person. All this probably could be wrapped in a Mac OS
X Automator like tool. We would then have both an automator for users
and a test tool.
> At any rate, let's keep talking - this is more fun than Sun Microsystems,
> their R&D was very bureaucratic. And Intel chip teams now have huge QA
> groups, like 10 times the size they were when I last tested a chip there.
> Small is better.
If anything you will find the Haiku community small and not very bureaucratic :)
Not that we want to be running around chaotically either, but I hope
doing QA for Haiku can be as fun for you guys as developing for it is
for me.
--
Regards,
Ryan
Other related posts:
