(LinuxWorld) -- Many readers sent e-mail commenting on my article A strategic comparison of Windows vs. Unix. In this page, I hope to summarize and respond to the more interesting points raised.

Issue 1: Pricing

Several readers pointed out the Windows software costs quoted do not reflect academic or other negotiated discounts.

True, but neither do the Unix costs. For both sides prices were taken from the applicable Web sites on or about September 19, 2001. I don't usually pay retail and hope you don't either, but the real-world discounts available pretty much net out between the two architectures.

Issue 2: Citrix thin-clients

Several readers suggested that a comparison between the Unix architecture with smart displays and the centralized Windows architecture with Citrix-style thin clients might be valuable. Others missed the point that we were comparing Windows client-server to Unix with smart displays and complained that comparing Windows fat clients to Unix smart displays was unfair.

The suggestion is very interesting and ties in well with the concern that articles like this have to draw unrealistic "pure play" scenarios to produce generalizable conclusions. Is the Unix architecture a better choice for business? Yes, but you don't see this in the real world because almost all environments are mixed and you can't easily separate costs deriving from management failures from costs deriving from technology issues. An analysis of transition processes in a realistic environment would almost certainly involve use of Windows thin clients as part of the transition. I need to think about how to cost this out. Suggestions are appreciated!

Issue 3: SPARCstation IPX won't run Solaris 8

My mistake. It will run 2.7 unmodified but getting 2.8 to boot would take upgraded ROM code.

Issue 4: Software not available for Unix

Several readers commented that the Unix architecture could not be implemented as a pure play because their users demand software functionality not available outside the Windows environment.

This issue comes up in two forms. People see:

1. Some functionality as simply not available.
2. The difficulty of getting people to use non-Microsoft products as creating a de facto requirement for those Microsoft products.

I don't think there is any Microsoft functionality that isn't matched or surpassed by Unix software except for functionality, like BASIC macros in Microsoft Word documents, that no sane security manager would allow people to use in a business context. In the specific case of the scheduling component within Microsoft messaging, for example, there are at least three Unix alternative directions, including:

1. License an equivalent product like IEMS (from International Messaging Associates) for Linux, BSD, or Solaris.
2. Use a package like iPlanet or a combination built around Apache to custom fit a solution to your needs.
3. License an alternative solution, such as Lotus Domino or GroupWise.

In addition, you can always adopt one of the ways of running Windows software under Unix -- including putting up to eight Intel PC boards inside a Sun server -- and delivering it via smart displays.

The user expectation issue is much harder to deal with.

Can you really get several hundred, or several thousand, users to switch from Microsoft Word to something like swriter? The glib answer is that the article wasn't about transitions but the more serious one is that you generally can, and quite easily too because Word is a product few users actually like. If your organization doesn't depend on getting conversions to and from various Word formats exactly right, your users will be quite happy with StarWriter or any other Unix word processor.

Unfortunately, this is not true for Microsoft Excel. Up to about version 5.0 this was a lightweight, powerful, and easy-to-use tool that even now retains its position as the star of the Microsoft Office group. There are Unix products that do the same things but not with the same look and feel. Users may reasonably be reluctant to give up the skills they've acquired in the use of this product.

There are several good solutions:

1. Run Citrix on Solaris with an attached Windows 2000 Server instance to distribute Excel, or other Microsoft, services to uses of smart displays.
2. Use an embedded processor running something like NT 4.0.
3. Give a few people with real business needs for Excel their own PCs.
4. Attach a few Macintoshes (with OS X, Macs are Unix machines) with Microsoft Office to your network.