Log in

No account? Create an account

Previous Entry | Next Entry

I Feel So Special

weather: cloudy
outside: 6°C
mood: giddy
I'm running a private build of </a></b></a>semagic because I reported the bugs and I get to try out the new build with the fixes.

I really don't know what the big deal is. I get pre-releases of all kinds of big name commercial software all the time. You'd think I'd be pretty blasée about that kind of thing by now.

Anyway, I got a fix build. The original problems were fixed, but halo regression showed other areas that still showed the same symptoms. I got a second fix build and halo regression cases all passed.

Look for </a></b></a>semagic soon. =)

*          *          *
Ladies and Gentlemen,



Jan. 11th, 2004 07:50 am (UTC)
halo regression
What's halo regression?
Jan. 11th, 2004 12:40 pm (UTC)
Re: halo regression
It means "testing around the bug", like a halo =) Ideally, you would do a full regression of all possible test cases for every bug fix because as soon as you touch the code, there's a potential (however slim) of breaking something else. But that's not likely to happen because it takes too much time and sometimes, it's obvious that touching one part really has very little impact on another area.

Halo testing could be done for similar instances, it could be multi-feature scenarios that involve the bug, it could be retesting the entire feature that the bug belongs in. Or if you know that the code changes will impact other features (if your changes were in a core module that's used throughout your app.), you do testing in each of the areas that you know it might affect.

So, say one particular font has trouble being displayed. You would, for sure, retest the font that was originally logged.

You could then see if you could save the document, reopen and get the same font back. You could try to do Print-Preview and see if the font shows up.

You could try another font in the same class or type. Or do a full regression on all the Font feature test cases to be sure that it didn't kill anything else.

It's less of a formal process than BVTs or Regression — it _could_ be formalized, but usually there's no time to do that. It's more random and the hope is that if you at least do something, most of the time you'll hit on side effects or other problems that the fix caused.

Does that help? =)
Jan. 11th, 2004 09:20 pm (UTC)
Re: halo regression
A little; thanks.

I think I'd need to know more about software engineering to appreciate all of what you said.

Which is kind of sad if you think about it, since I'm a software developer.


The Bride of the First House

Latest Month

March 2015