The second question Philip asked in a previous blog entry was:

"How long can we expect to see the MSI as we know it or when will DSI (with XML) blow it away?  Of course you can't answer this officially, but what do you think?"

I've been struggling with this blog entry for a couple days now.  I've sensed that questions like Philip's have been floating around my blog for a while but I guess I was just waiting for someone to ask it.  The problem, as Philip notes, is that I really can't answer the question.  It isn't because I don't want to.  The reason isn't even that insidious.  It all comes down to the simple fact that I don't know.  I really don't what the future will hold.

Fundamentally, decisions by people with titles much larger than my lowly "Software Design Engineer" title have not been made.  More to the point, they don't consult with me about what they are currently thinking.  Heck, I'm not even sure they spend much time (if any!) thinking about Philip's question.  However, I think it is an interesting question and I do spend some time thinking about the question (even if Senior Vice Presidents do not).  I've come to two conclusions I'm willing to share.

1.  Unless Microsoft policy changes radically, the Windows Installer is going to be well supported for the next seven years at a very minimum.  Why seven years?  Well, Microsoft Office System 2003 just shipped and that version of Office (like the two before it) requires the Windows Installer to install.  Last I checked, Microsoft promises to support technology for at least seven years and sometimes more (for example, the Windows9x code base recently got a life extension).  So, I think you can be very comfortable betting on the Windows Installer for your installation needs for the next few years.

2.  It isn't clear to me that Dynamic Systems Initiative (DSI) will "blow away" the Windows Installer (MSI).  The Windows Installer is inherently single machine focused where System Definition Model (SDM) (the core of DSI, IMHO) is inherently distributed.  Depending on the problem you are solving one technology may be more appropriate than the other.  On top of that, we've had some hallway conversations about how to get SDM to interoperate properly with MSI.  I don't know where those conversations will end up but that is just one of many interesting ideas we're tossing around.

Finally, I'm going to link over to Chris Sells interesting blog entry because I think his opening paragraph provides a really pragmatic way to approach all this information about unreleased technology.

Philip, I hope somewhere in there that I've answered your question.