(sigh) Without wishing to promote any war of words, the important issues here are: (1) Linux is very, very much like Apple's operating system. A tiny job to output linux versions of software if a modern Mac version is already done. (2) Linux is not every end-user's choice for desktop operating systems, but it is big in servers (e.g. web servers) for various reasons including reliability/uptime, and the same businesses that want reliable servers will probably want reliable business software on them. That is the reasoning behind a lot of companies making linux (and other unix-family) versions of their software... it makes $$$ and sense. (3) When I worked at a university I administered a lot of PCs as well as Linux, Solaris, VAX terminals and servers, and the Total Cost of Ownership of PC's running Windows was much worse than the "thin client" approach of Linux and Solaris, when taking absolutely everything into account (and the linux workstations cost much less in $ to buy and operate than HP's own PC's running Windows that they were bragging about at the time). The main point here is that what costs admin time is where people muck up their workstation (doing "reasonable" things like adding printer drivers even), whereas thin clients (and Linux doesn't have a monopoly on this of course) have always been a safer option, more resilient and cheaper to manage. (4) Many people are using Linux at home or at work without knowing it. It lends itself very well to "appliances" where you just connect it up and use it, with no PhD in computing required. There are already starting to be database application appliances even - not just database servers, but applications using a database that people buy as a package - a little bit like the old idea of buying a package of mainframe hardware and software from IBM and not needing to know the grubby details of what goes on inside, but now its 1/10000th the cost of the 1970's bundles. It used to be thought that the thin client approach only made sense if there are many workstations, but it still makes good business sense with two or three computers because it divides the all-important server from the most virus-prone area of the organisation and it is cheap to do now in an environment where the workstation has to be doing several things, not just being a thin client X11 workstation (e.g. you can use cygwin under Windows to act as a client* to a linux server that runs the application, and still run MS Office/whatever). *I use "client" in the conventional sense, rather than the meaning specific to X11. Given the way viruses (let alone crashes due to DLL incompatibilities or user "finger trouble") are eating into uptime, the professional business approach has to be to maintain workstations that act as thin clients - that store nothing valuable and can be swapped out for another working PC in a minute or less - and use applications servers that are kept safe from viruses, safe from fiddling, with RAIDed disks, etc. That means spending the best part of a day recompiling MYOB for linux and then running it in a good client-server setup, as I described, for a while to see first-hand (should that be "first-eye"?) what a robust system can be achieved.
