Thursday, October 15, 2015

About Bugs

Been a few weeks since I last checked in.  Onboarding with Oracle has been like drinking from a data firehose, so I've been a bit pressed for time...

As I write this, I'm sitting out on my backyard patio right around sundown.  It's been unseasonable warm here in Utah of late, so the flying bugs are out in abundance.  While they're a bit irritating, they're not the type of bugs I have on my mind this evening.  I'm thinking more about software bugs.

I design a software bug as a flaw that causes said software to perform differently than designed or intended.  Simple definition for my simple mind, I suppose.

One of the eye-opening insights in having an insider's perspective at Oracle has been looking at software bugs.  Not the volume of bugs so much as the type of bugs.  If I apply my simple definition of a software bug, over half the logged bugs are not really bugs at all.  The software is working as designed or intended, but we're not seeing the expected result.  Could be any one of a number of reasons:  applying the software incorrectly due to lack of knowledge, desire for features not provided, violation of business rules, data quality flaws in converted or interfaced data, errors in writing data...the list goes on and on.

Wading through this data, I see a couple of trends.  First, it's pretty obvious that enterprise software vendors could do a better job of educating and enabling customers and partners on how their software works...especially from a systems engineering perspective.  Second, the industry needs better tools for evaluating data quality prior to converting or interfacing data between data sources...a symptom of the old "garbage in, garbage out" rule.

If we could improve in this area, think of the reduced workload for application developers.  Keep in mind that most enterprise software application development teams are dealing with at least four application versions simultaneously:  the previously released version(s) in the field, the latest release in the field, the next release being built, and the design work for the release after the version currently being built.  Anything we can do to alleviate the bug resolution workload allows vendors to apply that extra bandwidth in ways that will shorten release cycle times...something everybody wants.

I'm sure there are more trends to add to the list.  You have a contribution to make?  The comments await.

1 comment:

James Morrow said...

One of the things I see frequently is related less to a "bug" (by your definition) than "bad design". Frequently it's "the person designing/developing the software doesn't actually *use* the software..."