Paul B. Gallagher wrote:
Rufus wrote:
Paul B. Gallagher wrote:
JeffM wrote:
Paul B. Gallagher wrote:
Inviting end users who are incapable of coding or testing
is an empty promise.
Describe more fully "incapable of testing".
What I mean is that most end users can demo a program, play around for a
bit, and generally satisfy themselves that "it works" for their favorite
tasks. They might accidentally stumble upon a problem. But they won't
perform rigorous, systematic testing such as navigating to each and
every option on a menu. So if you have a feature that's rarely used, or
if it doesn't elicit interest or curiosity because of its menu location,
name, or description, it won't be tested. In this scenario, you need a
very large cohort of testers with very diverse interests (ways of using
the program).
I test avionics software professionally, even though I don't code any of
it - it helps that I'm a pilot, and share a pilots "user" experience.
That's about all that's required, and what you describe is likely
sufficient because it highlights specifically what is important to the
user; anything that the user doesn't *use* isn't even a consideration as
far as an end-user is concerned.
...
Well, as I'm sure you know, there's "don't use" and there's "don't use
much" and there's "don't use unless my life is on the line." So it's
important that things like parachutes work even if you hardly ever think
about them, much less touch them. If you lose an engine or your rudder
fails or your fuselage pops its top and starts spraying passengers and
flight attendants into the ocean, you have to be able to fly the plane
safely anyway (and walk away from the landing, too). So all those things
have to be tested.
Yes. But we do retain and maintain functions where the hardware has
gone out of inventory...which is a waste, IMO...but we do it because our
product has to delivered to meet specification, and removing code often
costs just about as much as putting new code in.
And it's not all the same code in all of the boxes - some is high code,
some is assembly...and it all has to play together.
Of course, I understand no lives are at stake here. But there are
parallels.
Yes. It's the process parallels that I try to point out. And end quality.
And I bet your process is probably much more rigorous and thorough than
the average software beta tester's.
Actually, it's very similar...but far more organized and priority
driven. And it's driven by interface functionality, operational
requirement, and not much else. We have hardware set aside in labs, and
we use actual airplanes.
One point where my process is near exact is in the new "accelerated
release" numbering scheme that the SM team as adopted - that's almost
straight out of our process book and doesn't bother me one bit. Even
the numbering scheme is familiar to me.
--
- Rufus
_______________________________________________
support-seamonkey mailing list
[email protected]
https://lists.mozilla.org/listinfo/support-seamonkey
