The only piece of software that installed 100% without problems here was Office 2003. Virtually all educational software (primary) gives us problems. Those that come with msi's promise so much and rarely deliver, and those we have to package (on Ranger 5) won't behave once rolled out, usually because they've been written for W9x and expect access to all areas of the C drive, whereas in an XP Pro environment with clients fairly well locked down, the program falls over because it can't write to a part of the C drive (all pupils' files are saved to a server share).
What I don't understand is the software writer's concept of network installation. To me, if a product is sold as can be installed on a network, it should mean that you can make a software install GPO, link the GPO to the client OU and off you go. What most educational software's idea of network installation is you can put it on a server, then walk round all of your clients and install it on the client from the server. I can't see that as any benefit over taking the CD round to every client.
I'd really appreciate some sort of league table of software that can show clearly who has made the effort to make it install from the server with a GPO (surely the only sane way of doing it these days with so many clients in schools) and who hasn't. Then we vote with our pockets and don't buy the stuff that won't, despite how good it might be. The downside may be that only MS software would get top ratings, as it installs so easily... However maybe it will give writers of desirable stand-alone software the incentive to get it to roll out effortlessly from a server without having to walk to every client machine.
Although we try to have everything rolled out using a GPO or as a second option package builder in Ranger 5, I do have to have some software as part of the Ghost image, as it's impossible to roll out any other way. However, I feel that Ghost imaging is an inelegant way to add or remove software; using GPOs is far more elegant.
As with everything, you get what you pay for, and I wouldn't mind if a piece of £50 software was a pain to roll out, but a lot of these companies are jumping on the 'let's have a maintenance charge or a licensing fee and get a steady income' bandwagon (as with SIMS) and not coming up with the goods. You can say what you like about MS but when you consider what you do get for your money compared to the likes of SIMS or some software manufacturers, it's a bargain.
To get back on thread, we're currently bashing our heads against Coco3, as it will roll out reasonably OK but four of the simulations it claims are not installed (even though they are there in the same directory as the other five that it can show). Digital Blue for the QX5 microscopes just won't work, the software for the Canon scanners that we have everywhere won't permit us to build a package as it has some sort of licence protection on it, Catch Up Literacy CD1 is a crock (although CD2 went out with only minor tweaking), and ACTIVprimaryMV installed to a distribution point (using setup.exe /a) as if it would roll out but it never appears on the client machine, and tries to install every time the client is started.
I wonder how many thousands of hours are wasted every year across schools trying to install ropey software? Of course, for our BSF colleagues in high schools this will soon all be done for them - if it's anything like the PFI model (and it will be, or worse) then it'll simply be 'This software won't install on the network so you can't use it.' Problem solved!