How old is too old to support?

Question

I'm writing a new project, and I have a choice between using a library that only exists in OSX 10.5 and later (We're on 10.6 now), but makes my life much easier, and using a library from earlier versions, but I have to a lot more of the work myself.

How does one make this decision? How do you balance new/better technology vs customers on old systems?

ETA: does anyone know of a site that compares market share by precentage of a specific OS? Since this is a consumer product, if only 2% of mac users are still on 10.4, that sort of makes my life easy. Similarly, if 25% are still on 10.4... (I know, it's almost guaranteed to be somewhere between...)

Answer 1

Ask your clients - how many are on older versions of the OS?

Can you afford to lose them?

Edit: (following comment)

If you don't know what your target audience is using, you have a problem. You need to get an idea of the magnitude of how many potential customers you will not be able to serve if you go with your new library.

Having said that, shipping is a feature, so if you get the product out much quicker, you can always refactor the code to use the old libraries if you think it will gain many sales.

Answer 2

In general you should base your decisions like that around the interests of your paying customers. You should present the issues to them and the risks involved in each alternative and let them make the decision.

Answer 3

Depending upon your particular application and requirements, I would personally ship this as a major update (i.e. version 2 compared to version 1) and explicitly state that a minimum of OSX 10.5 is required.

You could still support your previous version with bug fixes, just not new features that depend on library X.

Answer 4

As others have said, this really boils down to whether you can afford to lose customers who aren't on 10.5 yet. That said, lots of companies seem to support the two most recent versions of OS X in their new major releases, although older versions are often available for people with older systems.

Answer 5

Another way to think about it is that if someone is on 10.4, then they likely haven't been an active upgrader / software purchaser for the last 3 years. So the likelihood that they will want to spend money on your software is low.

Additionally, if they really want your software, they'll upgrade to 10.5 or 10.6 and gain loads of other advantages at the same time. While that OS upgrade won't be free, it will come with so many other advantages to the customer, they might not mind.

It's also important to consider how much time and effort it will take to develop your software. If these newer libraries mean that you ship the product months earlier, or with better features, that will also pay off.

Answer 6

If software ownership is stable and software vendor is not pushing too hard in phasing out their own obsolete software, then there are no reasons to not support.

The problem is much worse, when vendor is passively aggressive or committed the phasing out: dead download links, dead 3rd party companies, who made the hardware/drivers/compilers/libraries, unobtainable documentation, incompatible media/installer to recover/reinstall the product.

My example: pre-2000 vs 2005, it is nearly impossible to reconstruct say.. the build process of 1 mln lines of 100% saved and mothballed Visual Studio 6.0 projects from year 1999-2001, obtain all 3rd party libraries from the era, prepare proper SDK, platform itself, all patches, make results binary identical. No way.

But it pretty much works for Studio 2005.

Answer 7

You need to talk to both sales and support, and let them judge what the impact will be.