> Excellent points!!!  Especially about the constant upgrade path.
> 
> Consider this:
> 1.  Assume Java is backward compatible.  I think it mostly is,
> and the parts that are not, the compiler warns you about using
> a depricated api.
> 2.  Java 1,0 programs still work in a 1.2 JVM, right?
> 3.  Sun supports their OS's for 5 years after they are declared
> at end of life.  This announcement was made for Solaris 2.6
> about a year ago.  Microsoft hasn't supported any product version
> for five years yet.

Support for Win95 ended just last November.  1995-2001 is more than 5 years. 
Win98 support is supposed to die in 2003 (July IIRC).

see:

http://slashdot.org/article.pl?sid=01/12/06/2123216&mode=thread

and:

http://www.microsoft.com/windows/lifecycle.asp

Then again, RH seems to have a similar lifecycle for their distro as far as rpms
go.  Of course you can roll your own rpms, you can't roll your own M$
fixes/upgrades.

> With these considerations, figure out the total cost of ownership
> for the project, from birth to death.  Use Microsoft's track
> record.  You know they will force you to upgrade OS at least
> every 3 years.  Also consider the price of hardware.  Sun boxes
> and AIX boxes are more expensive than Intel-based Compaqs or IBMs.
> 
> Have fun!
> 
> George
> 
> Sundar Narayanasamy wrote:
> >
> > Joseph,
> >
> > This is what I infer by reading .NET, though I might be biased as I have
> > successfully worked/implemented quite a few J2EE projects.
> >
> >                 1. .NET is not just a framework on how to develop your
> > application, you have to buy middle layer support that would enable you
> > to run .NET apps. Of course these middle layer products are released by
> > M$ and you pay a hefty price for that.
> >                     Even J2EE needs J2EE complaint servers to support
> > their framework, but you can get better than commercial grade free
> > software like JBoss, Enhydra totally free.
> >
> >                 2. Even when you buy M$ software, lot of times you have
> > to train your employees and/or pay lot of money for support contract
> > with M$.( I am implementing MS SMS for our company and had few questions
> > about certain things, but when I asked the questions in their News
> > Group, one of the M$ representative politely asked me take advantage of
> > their current specials on training sessions-- i.e $3000 for one day
> > session in Las Vegas)
> >
> >                 3. .NET implements M$ way of developing applications
> > that are tightly integrated to one another(though they claim otherwise).
> > It is not based on MVC model, which any object oriented programmer would
> > swear by.
> >
> >                4. I haven't yet read field case studies on successful
> > enterprise level .NET implementations outside Microsoft; whereas J2EE
> > has many to go by.
> >
> >                5. And since M$ always ties their software releases with
> > Operating System and Servers, you have to upgrade your hardware/software
> > regularly to get continuous support from M$ and their vendors. My
> > philosophy is - if it works why fix it. ( We have some old Venix
> > systems, which we still use actively; they work!)
> >
> >                 etc..
> >
> > Sundar
> >
> > Joseph Gledhill wrote:
> >
> > >
> > >I need some legitimate reasons not to go with .NET as a development
> > >platform.  Any comments would be appreciated.
> > >
> > >thanks,
> > >
> > >Joseph