VSTS 2005 offers nothing but Visio in the way of tools to the Software Architect. There are plenty of tools for the System Architect, Developer, Tester, and Project manager, but banks everything on DSLs and Software Factories for the Software Architect. Both will add some value in the future, but the only tool we have to work with now is UML 1.4 in Visio AGAIN, which they say they have no plans to upgrade. The MSF agile templates so far are horrible, and our process we are putting together is a Product Line Engineering (PLE) process that uses off shore development, so the process's ceremony is very high. We need heavy documentation to produce an effective communication channel between our Marketing/Management team, Architecture team, and our Development Team. We are currently widdling down the RUP to UP, and then building it up to a PLE process on our own. So we are creating 70% of the artifact templates and guidelines by hand.
I put this out a few weeks ago after evaluating their DSL toolkits:
I think every yahoo in the world who likes to re-invent the wheel will be putting out their own designers. This now ticks me off as much as the MSF did when it claimed to be a process framework when it first came out. The only thing that has accomplished is making a huge mess, and then coming back full circle when they decided to give it some instances of structure, and they implement XP, UP, and RUP. The same processes that were around when they started their mess.
We have worked far too hard to get where we are with industry standards for software development. Why is it the electronic engineering field has been able to stabilize? I doubt they would have if they would have engineers defining their own schematic language on every project across the world.
Imagine being a consultant and every new gig you get you walk into you need to learn a new modeling language. I think DSLs can add value if they are used correctly, but UML already does when used correctly.
The only part of it I like is the template engine, but this isn’t going to be worth the mess this makes.
The only hope we have is if the experts in the field like Mr. Booch, Mr. Gamma, Mr. Martin, Mr. Ambler, Mr. Jacobson, Mr. Gomaa, Mr. Clements, Mrs. Northrop, Mr. Fowler, Mr. Evans… etc. (You know who you are) step in and take the lead on this, to get this under control as quickly as possible, and guide us in the right direction on how to use this correctly. Then and only then will this possibly stand a chance of just not making a huge mess.
We need someone at MS to produce a MSF template mirroring the UP (the only process I have seen be truly successful in a non-agile environment), and if they are banking on Product Line Engineering (PLE) as there core process for Software Factories, we need an MSF process Mirroring what SEI combined with the ITEA projects says PLE is. There is already a company working on getting a MSF RUP template to work within the VSTS environment.
Along with these processes that are intended to document the architecture (software architecture, not system architecture), they need to provide the modeling tools and process templates that make implementing them possible.
MS needs to get on the ball as to what Software Architecture is and what we need to be successful, so far VSTS 2005 is greatly lacking.
links: digg this del.icio.us technorati reddit
I've posted a reply to these comments on my blog
(http://blogs.msdn.com/jackgr/archive/2006/01/06/510162.aspx).
