by jiml8 » Oct 15th, '13, 18:04
The article DOES make some good points, and fragmentation in Linux is probably the single biggest reason it is not a much larger presence on the desktop. Not only is it fragmented, but there are no standards to the UI such that a program developed for one UI will work across all UIs. This makes the test matrix for any developer who wishes to make a commercial product that runs on Linux into a freakin' nightmare.
I have a product in the marketplace that I started work on 22 years ago. This is a vertical niche program that I initially deployed it in my office where it ran a business of mine and was tested by my employees. In 2000, I started selling it on the internet. It started life on the Amiga, and by the time I started selling it, it was a Windows-only package. It sold fairly well.
Fast-forward to the present. My product is totally obsolete; it will not run on 64 bit Windows and 32 bit Windows is dying. Not quite dead yet, but dying. My choices are to abandon the product or to rewrite it.
Well, it is a good product. I have rewritten it using a completely new environment, and in this new environment it is fully modern and runs on both 64 bit (also 32 bit of course...I develop it in Windows 2000) Windows and Linux. Presently it is in beta testing (I still need a couple of beta sites...I no longer run that business that was so key in testing this thing before) but my current release plans are Windows-only.
Why? Because the Linux version only runs on redhat-descended distros using Gnome, which is where I have tested it. It probably does not run on all of those. The wxWidgets package (which is central to this new package) behaves differently on Gnome and on KDE, and there are some problems using it with KDE. I have loaded my package on some other distros and encountered problems. To make it run on all distros, even all major distros...I simply don't have the resources and it would not be profitable anyway. To provide config instructions for every distro out there...well...my end-user base tends to be clueless anyway and they want to install and use. If they need a degree in computer science in order to configure it, they won't like it and will harass me with support requests.
I do intend to release an Android version that will play in a client/server role with the desktop version. But Android is a big market and more or less uniform, once I get past the display resolution issues.
But, that's my developer's perspective of desktop Linux. My perspective is by no means unique; I won't develop for-profit user packages for Linux, and that's why.