Home | Login | Recent Changes | Search | All Pages | Help
IfEverythingGoesWellThis started out as a letter to JerryWeinberg, where I noticed myself making one classic mistake, and he pointed out another I was making. His replies are interspersed with the original. - ShannonSeverance 2002.07.16 Today I am evaluating proposed changes to the build process we are using. I want to separate the changes into three groups: "Do", "Don't do", and "Worthy of further consideration." To get a rough cut, I'm grading each change on three criteria, "Value", "Risk", and "Ease of implementation." The scale consists of high, medium, low, and don't know. On the index card for "Change to a new version of NAnt"* I marked ease of implementation as low. Then added the note, "if everything goes well." Of course nothing ever goes well, or at least not well enough to satisfy the needs of "if everything goes well" which I think is the same as "if there are absolutely no surprises." To top it off management probably wouldn't hear "Low, if everything goes well," but instead, "Low, blah blah blah." I think I will mark it as unknown. Jerry: It's not clear to me whether "low" means easy implementation or hard implementation. It may confuse them, too. On your three scales, you've made a basic error in human interface design. High Value is good. High risk is bad. High ease of implementation is (unclear). It would be better for your readers if all three scales read the same way. For example, you might change them to: Value, Safety, Cheap (all HIGH are good) or Exposure, Risk, Expense (all HIGH are bad) Changing to a new version of NAnt would rate (I think) Low value, Low safety, and not cheap (potentially). Me: * NAnt is our build tool, which was chosen before I started. Many of the features still aren't implemented in the version we are using. I don't know why anyone would choose a version 0.x version of a software tool to support a build process when very mature tools are available. I guess because the mature tools don't use XML. Jerry: Or, are you saying the the new version would be at least 1.x, and have the features you want, which would have [No, the current version is 0.7.9.0, but does implement some usefull new features - Shannon] High value, Low safety, not cheap (or don't know) - not so clear a decision, but if you decided to do it, it shows clearly that you need to do something to make it safer to install, and curb your expensive difficulties. Me: After writing the above, I just read Reasons from the AYE website. I suppose that answers why someone would want to use a new unstable, pre-release tool, because they wanted to. Jerry: It was a "neat thing." BTW I've adopted the Value, Safety, Cheap scale - ShannonSeverance
Updated: Tuesday, July 16, 2002 |