Big Name Company Testing Strategy: Let The User Find The Problems
We thought we were loosing it when we tried to use the latest MSDE 2005, until we saw a lot of posts about the same problem. Fortunately for us, we had a Plan B. Unfortunately for Microsoft, their database product got tossed.
Enough Is Enough
It’s embarrassing to say how much time we spent trying to get MSDE 2005 to install correctly. Regardless of what we tried, the install program never could find sqlncli.msi, and terminated without putting in the prior stuff that went in without a hitch.
The log files were less than useful, and it wasn’t clear why the logs were made in the first place: the designers never looked at them and tried to abstract meaning. It is really annoying when clairvoyance seems to be a presumed user requirement. And that the user (i.e., us and many others) have to figure out the problem that was built in to the product from the beginning.
We didn’t have time to fix this Microsoft problem. Whether the problem was really in the downloaded Microsoft product, or we don’t know what we’re doing, became immaterial: We have customers to satisfy and commitments to meet. We were fortunate enough to have a Plan B that allowed us to use another RDBMS product quickly. All of our database designs consist of two parts: Separate logical and physical drivers. All we need do is change the latter and recompile to switch databases. Like Staples, that was easy.
Perhaps Microsoft has really lost its way. The first tipoff was the web page where MSDE could be downloaded, something more akin to what a Berlin cabaret in the 1920s may have looked like. It was certainly easy to imagine all the programmers with goatees and wearing dark leather turtle necks.
Larry Was Right After All
Some time ago, Larry Ellison (of Oracle fame) publicly satirized the career path of the Microsoft employee(s) who work on Word’s red squiggly underline. At the time I thought his comments were unfair; I still do, but they’ve proven to be accurate.
A Modest Proposal
Microsoft has a lot of cash, and it’s unlikely that it can employ it to get good returns on software. There are just too many smaller and nimbler companies that make better software than Microsoft can ever hope to. Perhaps Microsoft should exit the software business and use its large cash holdings to become an investment company. Sell the campus in Redmond, get rid of their worldwide overhead. Nothing too aggressive, diversifying its investments over a large group of index funds, anything that doesn’t take too much critical analysis or require an army of analysts.
Tags: abusing customers, Microsoft, software testing, sqlncli.msi, useless log files