Software Deployment

As with Mark Twain, the rumors of windows installer's death are greatly exaggerated. I don't know that windows installer was ever a lively art, given the dearth of publications on the subject. It has always been a niche. If you're in it long enough to realize that windows installer is a relational database you're considered a "guru." [:D] Doubt it's going away - we're up to - what - v 4.5 now? And MSFT is committed to windows installer being their standard way to install stuff on Windows. Also with hard drives getting bigger and cheaper I don't really see the logic in going to all virtualization or "cloud computing" or citrix or "web apps" or <insert latest replacement for Windows client computing here> especially when the replacements are not as feature rich as their "thick client" brethren... or have performance problems, or both...and the management tools exist to do deployments of even large software packages.......if anybody can point out a flaw in my reasoning I'd be glad to admit I'm wrong, but I've been hearing about the death of the thick client for some time (and for purposes of this thread I think we can equate windows installer with the standard windows thick client) and it ain't happened yet.

As long as Application Virtualization can't deliver everything Windows Installer can it won't get the adoption it that Windows Installer has. I am not only talking operating system components but interfaces that don't work with App-V such as COM+. I was hoping COM+ would die in Vista but it looks like it still lives in Windows 7 so we will still have a community of applications that will require regular installation for full functionality.

I know Microsoft, ThinApp, and Xenocode to name a few want to get vendors on board with releasing applications in a virtual format but I think the key success factors aren't there. What made Windows Installer useable as a standard (in my opinion) was the install base (part of the OS), the openness of the standard and the facility to let customers customize installers without modifying the original package (the transform). Microsoft built and supported the engine that facilitated software installation but kept the platform open enough for vendors and hobbyists to develop products that build, manage and deploy packages. Maybe if App-V one day becomes part of the OS I could see it being widely adopted by software vendors as a native package format for their software to be delivered in but right now the application virtualization platforms are too closed and the market is too competitive for vendors to open up their platforms. In order for ISVs to be on board in numbers a solution needs to come with the OS or someone needs to win with an embedded client solution such as ThinApp.

Good subject.

I agree with all of the above, while app virtualistion has many benifits it's still relatively new.

We've trialled App-V, Thinapp and Citrix App Streaming. All have their benefits and drawbacks - App-V probably the most powerful and attractive product but also the most expensive since you have to buy into MS Software Assurance agreement.

We had a chat with virtualisation consultancy backalong any they said that at best you'll get 80% of your apps vitualised with App-V and around 60% with Citrix. The rest is good old MSI. Thinapp isn't really there yet from a management perspective.

Whatever you go for, it still required a skillset of a traditional packager and scripting capabilities are a must to deliver fixes as part of the vitrualised apps. It's just a new art to learn - just as niche as windows installer [:)]

Regards,
Rob.

Not so much "dying" as "moving (very) slowly in a different direction". I'd agree that packaging will be a required skill for a good few years yet.

Interesting topic and interesting wise ideas :)

One thing is for sure, I'd bettter get myself equipped with App-V, Sequencing and Virtualization otherwise I risk losing my job...lol

I don't know how many of you are involved with the Build / Release world but I think this is an area where Packaging products are particularly weak.

Using WiX is quite an art (much like I used to feel with Windows installer on the whole).

The only WiX editor WiX aware is quite buggy and misses some of the refinements offered by Wise / IS.

Perhaps this is a space for some work / new tool development.

For those of you who are deep into WI, packaging and not in the release space you should check it out, it offers a world another really interesting aspect of deployment and some insight as to why many developers have such unusual techniques (to say the least)

i agree learning the ways to build msi's isnt easy.
But maybe it helps when the support from microsoft wil be like a good intergration of wix with visual studio.
Even give out a express edition with wix intergrated in it so more people wel be interested in it and the quality of setups wil improve.






don't know how many of you are involved with the Build / Release world but I think this is an area where Packaging products are particularly weak.

Using WiX is quite an art (much like I used to feel with Windows installer on the whole).

The only WiX editor WiX aware is quite buggy and misses some of the refinements offered by Wise / IS.

Perhaps this is a space for some work / new tool development.

For those of you who are deep into WI, packaging and not in the release space you should check it out, it offers a world another really interesting aspect of deployment and some insight as to why many developers have such unusual techniques (to say the least)