Thursday, 12 June 2008

Distributing fixes and upgrades

This post if a follow-up/ response to The Lost Benefits of Handling Serious Errors post.

Whenever a problem is corrected in a software product, I assess whether a new release is necessary.

If a release is necessary, how many people will benefit? When I'm talking about benefit here, I really mean "How many people will come across the error and be inconvenienced enough to report the problem to me?"

There's no easy way to make this assessment. Things I consider are:-

1. How long was the software package in circulation before the error was spotted?
2. How serious is the error?
3. How close is the next planned upgrade?

If a package has been in the field/ in use for quite a while, I am likely to only make a new release to the customer reporting the problem. Too many releases can be off-putting for people.

One thing I do is "push" releases. I email all registered customers to tell them a new release is available for download. I don't wait for them to check for updates.

by ML

No comments:

Post a Comment