Running the community project from the Dev’Garten was a fascinating experience both in terms of seeing how tools like Team Foundation Server can be adopted in a relatively short space of time by a team of developers that have never seen it and how things need to be structured to increase chances of success.
I’ve learnt a lot of lessons, but before I get into that I wanted take this opportunity to thank the team leaders and developers that helped focus the efforts of the other team members that worked on the project in the Dev’Garten. So – thank-you (in no particular order):
Without these guys involved in the project we wouldn’t have had a failure to launch. Because of these guys we managed to get hundreds of work items completed and thousands of lines of code written. We even managed to get some testing done.
Frank Arrigo has a copy of the graphs and stats on his locknote slide deck so I will see it I can get a copy and post the relevant bits here.
The best thing about doing something at an annual event like Tech.Ed is that you might get the chance to do it again using the lessons learnt the first time. The Smith Family community project was no exception. Here are a couple of things that I learned:
- User account provisioning took too long; this one was a time killer. Some of our team members like Grant, Tatham, and Andrew Spent a long time creating accounts and giving them the level of access required in TFS. Next year I would either like to have everyones accounts pre-created, or perhaps even better is having some kind of scanning device which scans their event pass and creates the account in Active Directory and automatically gives them contributor access to the Team Project.
- We needed VPN access on day one; we had a few troubles getting external access to the Dev’Garten going. In the end we decided not to use ISA and instead just set up RRAS. This allowed developers in the Dev’Garten to get access to the Internet to look up documentation, and the team leaders to get access from wherever they happened to be located. Next year I’d like to get this going earlier so we could get more people coming in from networks like CommNet.
- We needed a scenario for each of the owners; one of the issues we had was people having visibility of how close we were to being done. This was my fault, I thought that I could own the scenarios and keep track of it, but I should have assigned one or two scenarios to each of the team leaders, that way they can help focus others efforts in those areas.
- Scheduling; its really hard for team leaders. They tend to be the ones who have sessions to present at the conference so we need to be consider the impact that has on delivering the project – I think that most people managed quite well, except for perhaps me, and I was lucky because Grant Holliday and Chris Burrows basically took change of getting the build up and running with TFS Integrator.
- Stats stats stats; Andrew Coates managed to swing a large LCD display for us on day two of the event. This was useful as it allowed the team leaders to take a quick look at how we were moving the line. Next year I’d like to get this setup on day one, and configure the warehouse updates to happen much more regularly.
There were also a few things that I thought were done exceptionally well. Those included getting the skeleton of the project fleshed out quickly including the data-model and data access layer. This made it possible for people to get data out of the system quick and we had very few questions about the database – it was just “there”.
I am completely psyched about next year, and while we still have a bit of polishing work to do on the UPlift project I think that this is a great idea for future Tech.Ed’s.