Is this really esoteric?
It has been two weeks of sorrow getting Ant to fulfil our product's clustered deployment on WebSphere with pre- and post- deployment customizations. That Ant doesn't support looping was but the beginning of our problems. [When I say 'our' I don't mean to come across as royalty. It just means there were more just myself working on this.] Anyway, all that is a thing of the past. Or so I hope.
I came across this on java.net this morning. Some of you regulars may have already seen this. Martin Fowler has written about DiffDebugging which is just a fancy shmancy name for finding out what recent change caused a regression problem. That something like this has been documented -- against the grain of the Agile manifesto -- doesn't suprise me all that much. What does is that java.net gave it frontpage coverage [if you can call it that. I am never sure anymore.] My question is: doesn't everybody already do this? Isn't that the logical starting point? I don't imagine initiating new team members by letting them in on our team's little secret. That's right son, 'round here we use DiffDebugging. Yes, that will be the day!
I came across this on java.net this morning. Some of you regulars may have already seen this. Martin Fowler has written about DiffDebugging which is just a fancy shmancy name for finding out what recent change caused a regression problem. That something like this has been documented -- against the grain of the Agile manifesto -- doesn't suprise me all that much. What does is that java.net gave it frontpage coverage [if you can call it that. I am never sure anymore.] My question is: doesn't everybody already do this? Isn't that the logical starting point? I don't imagine initiating new team members by letting them in on our team's little secret. That's right son, 'round here we use DiffDebugging. Yes, that will be the day!
1 Comments:
Of course, when you use an iterative & incremental development process, then the gap you have between 'broken' and 'not-broken' is a lot less (hours at most, and with a small set of changes). Thus, going through the set of differences is a lot easier than if the gap is large.
This is a precursor to efficent 'diff-debugging', and is covered in Martin's article.
By Anonymous, at June 15, 2004 at 8:38 PM
Post a Comment | Home | Inference: my personal blog