tag:blogger.com,1999:blog-36118108.post6454428922652005755..comments2023-12-06T00:17:28.519-08:00Comments on Creative Chaos: Test Estimation - VIMatthewhttp://www.blogger.com/profile/05956714498778698672noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-36118108.post-46603289179107370412010-09-21T12:13:43.280-07:002010-09-21T12:13:43.280-07:00So...iterate, learn, and be smart (or whatever ord...So...iterate, learn, and be smart (or whatever order is most appropriate) :}Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-36118108.post-90231097368376896092010-09-21T11:50:52.403-07:002010-09-21T11:50:52.403-07:00Logic tells us that we need /some/ padding, right?...Logic tells us that we need /some/ padding, right?<br /><br />The buffer percentage is going to come from experience and intuition - how much longer does it take us to actually test the software than the sum of the tasks?<br /><br />That will be different by team. If you team has no experience with this sort of thing at all, well, 35% if a place to start.<br /><br />Likewise, we'll say a 'cycle' is whatever we need need to do get some coverage of the app. If we find a ton of bugs in module A and none in module B, and all the fixes are made to A and 'we promise' say the devs, that none of the changes impact module B "at all." - well, cycle two will be very different than cycle one.<br /><br />But it's probably a fair first-order approximation to say that cycle two will probably take about as long as cycle one did, no?<br /><br />As for what I'm "getting at" - well I'm getting quite a few things. The general pattern is to lay out an estimation system, break it, and improve on it with the next one. I hope that's an interesting style of essay where we can at least learn a little - and be a little entertaining along the way.<br /><br />I hope you agree.Matthewhttps://www.blogger.com/profile/05956714498778698672noreply@blogger.comtag:blogger.com,1999:blog-36118108.post-51268377629674174612010-09-21T07:27:20.838-07:002010-09-21T07:27:20.838-07:00So each cycle time should be larger than ideal - p...<i>So each cycle time should be larger than ideal - perhaps by 30 to 40%.<br /></i><br />Why 30-40%? Do you use a formula or is that STSP (standard test schedule padding) based on your experience? If it's your experience, how do you know it will work for others?<br /><br />You are also saying (I think) that each test cycle is equal - do you have to run the same tests in each cycle? If you learn something on the first test pass, you may want to run more tests on subsequent passes. Conversely, you may learn something differently and run fewer tests - or different tests? Or is there an unwritten rule that says "thou shalt only add as many tests as thour removes?<br /><br />Still trying to figure out what you're getting at - should I just wait for you to finish, or can I make one of those "choose your own ending" books?Anonymousnoreply@blogger.com