The Collage of Devankur Thakur

"It ain't braggin' if you can do it."-Dizzy Dean, 1930's baseball player

  Home  |   Contact  |   Syndication    |   Login
  72 Posts | 83 Stories | 711 Comments | 952 Trackbacks

News

Article Categories

Archives

Post Categories

Image Galleries

Blogs to Read..!

My Other Blog..!

Software Testing Related Links

The product Version I have been working is nearing release and the work has been going at break neck speed, especially with bug fixing and its validation schedule taking a real upward curve. Getting builds thrice a week and for quite a few builds, we have primarily been doing fix validations. Already through 3 regression cycles (cycles are generally pretty long), and am amazed to find how bugs creep in some areas even when so unattached to the areas of fix. Regression makes sense really, it’s not always easy to ascertain the areas affected by a particular fix and a full regression can act as a cure for that. A small test team, thrice a weekly build, a build having 7/8 fixes, what’s the best way to deal with a situation like that, when if you have to do regression on a particular build say on bld05 and then the next build you pick up is bld09(due to elapse of time for the whole duration of regression) with all practicality taken into consideration!!

posted on Friday, April 08, 2005 5:20 PM

Feedback

# re: Multiple Builds and Testing 4/14/2005 2:15 PM Saravana
Hi,

I have been in the similar kind of situation earlier. There is one obvious cure, which is nothing but automation. But, even we did not have the automation. One thing we have done is, we use to validate the complete test suite, once in 3 builds, and in other builds, we just used to perform the regression tests. I really do not have more idea on this. Somebody can add their inputs.

# re: Multiple Builds and Testing 5/3/2005 3:31 PM Suresh
I prefer to have a Basic Checklist. This would first help us to analyze the features which are fixed and which are broken.
Once you validate the result then you can decide on whether to continue with full regression testing or going for impacted modules

# re: Multiple Builds and Testing 5/4/2005 6:19 AM Devankur
Well, yes, I agree to both of you here, prioritizing and identifying potential areas for regression is one of the solutions. Automation can be really handy here. But when it comes down to team size and the time allotted, we couldn't go for full automation, only the smoke/sanity part is automated which is helping us in BVT. Anyway, we are real close to release now and hopefully, things would be fine :-).

Post Feedback

Title:
Name:
Email: (never displayed)
Url:
Comments: 
Please add 6 and 6 and type the answer here: