Advertisement
this is more an XP question, but I am curious to find out whether people have been able to implement the 2-week release cycle prescribed by XP -- or even 3-week cycle.
Follow up question: do the product managers (customers) actually revise/adjust priorities at the beginning of the cycle.
My experience has been that it is hard to make a self-consistent set of features fit within an arbitrary boundary of 2- or 3- weeks.
In addition, Product Managers tend to change priorities "whenever they want".
The good news however is that we run large regression tests every weekend, so this limits the entropy of the whole process.
Net-net: Given the high rate of change on the product requirements side, and the fairly poor scheduling skills of this current team, I make sure that we have concrete deliverables, fully tested every 4 weeks at the most.
Follow up question: do the product managers (customers) actually revise/adjust priorities at the beginning of the cycle.
My experience has been that it is hard to make a self-consistent set of features fit within an arbitrary boundary of 2- or 3- weeks.
In addition, Product Managers tend to change priorities "whenever they want".
The good news however is that we run large regression tests every weekend, so this limits the entropy of the whole process.
Net-net: Given the high rate of change on the product requirements side, and the fairly poor scheduling skills of this current team, I make sure that we have concrete deliverables, fully tested every 4 weeks at the most.
Advertisement
Advertisement
-
Re: 2-week release cycle: real-life or not?
Sat, August 23, 2003 - 8:58 PMI've actually been doing weekly releases on my last two projects, but honestly the frequency of releases just seems to make things worse.
The more visibility people have into your process, and the more they can see things coming together, the higher the rate of churn from the product managers, as they think that they can change whatever they want and it will probably only add a week to the schedule, since, of course, all features are equal effort.
This may be that I am working with a really extreme company, and that my product managers and upper level management are on something of a suicide path, so take it with a grain of salt.
My takeaway is that what matters here is the audience for the releases. In my experience, the end user is just overwhelmed with new features and bugs more than once every few months. There is guidance and feedback that comes from product management if they have insight into the process, and it definitely raises their comfort levels as they see progress. However, you need to have a strong mechanism in place to really put a muzzle on them or things can quickly spiral out of control. -
-
Re: 2-week release cycle: real-life or not?
Wed, October 8, 2003 - 8:19 AMThe answer to this is The Planning Game. Without it one can't really embrace change. See xprogramming.com/xpmag/wha...m#planning -
-
Re: 2-week release cycle: real-life or not?
Tue, January 27, 2004 - 9:14 AMAnd I think the summary of my gripe is "real world buyin", which is at the heart of having the planning game work. I love XP, and would love to have things work out where all levels of the organization are willing to buy in and see it through. However, the fundamental issue I've seen time and time again is that upper management, especially upper layers of product management, are typically not clued enough to understand or not willing to listen.
The danger is having people with lots of power that aren't on the same wavelength, and they quickly stop paying lip service to the development process when cash starts getting short.
That is not to say I don't think it can't work, I just think that expecting it to work smoothly in a cash-constrained startup where not everyone is truly onboard (which is unfortunately frequent) doesn't really happen.
-
-
-
Re: 2-week release cycle: real-life or not?
Mon, January 26, 2004 - 6:08 PMI've never been able to implement XP 100%, but have created modified versions that seem to meet the needs of the business that I support. I've implemented 3 week cycles that operated within a 5 week window. The first release is 5 weeks. Week one is a combination of user requirements and group desings between the developers and the users, also during week one test script are written. Week 2 and 3 are pure development weeks with development testing starting and ending with the 3rd week. The application is handed over to the user group which tests for 2 weeks, during which time the week one cycle starts over again for the developers / user with requirmenets. The users end up getting new functionality every 3 weeks after the initial 5 weeks.
This has worked well for a few large projects developing production web applications using ASP. The only draw back is the flexablity the user has. Changes to the list of features can occur at any time and it seems like the list continues to grow to a point where the application is never finished. However, the user gets the prioritize the list of features and if some of the features never get into the application it's only the features that the user isn't worried about. -
-
Re: 2-week release cycle: real-life or not?
Mon, March 22, 2004 - 1:19 AMThis sounds really reasonable. I have been working on my own, and trying to implement a 2-3 week release cycle. I am still missing the testing aspect though, and that creates a lot of problems after every release. Thanks for your post though, I am getting a better picture of what needs to happen. I'll get there eventually.
-
