Not long after my post about the newest rev of OpenSolaris, a programmer friend called me to dissent. He'd tried OpSol, too, found it sorely lacking, and from his comments I found a criticism that now applies to Linux as well.

Since my friend's a programmer, he sat down with OpenSolaris from a programmer's point of view -- and most of what he saw missing were things that a programmer would have needed. Not just asked for, needed, or otherwise found it very difficult to cobble together from scratch. The word he used most often was broken, which is not something you want to hear coming out of a programmer's mouth.

Biggest gripe for him: the lack of dev tools included by default. He had to go dig up the Sun Studio C++ compiler, the Boost libraries and put all of those together by hand, just so he could then sit down to the task of compiling the software he was going to write. And, of course, he was disgusted by the lack of AC'97 audio, and lousy codec support in general.

All of this talk made me think about one of the chief difficulties in putting together a distribution. When you set up a distro and its attendant software repositories, you have to assume some degree of responsibility for everything that goes in there. Every single app has to be vetted and policed to make sure it works with the distribution. That includes everything up to and including the big, staple applications like OpenOffice and Firefox.

That might in turn go a long way towards explaining why, when you pick up a given distribution, some of the biggest and most important programs in it are a revision or two behind the times. Not because the maintainers are lazy, but because there's a finite number of such people and they only have X hours in the day (unpaid hours, sometimes) to crossbar-test what they can and make sure it works.

A Linux/Solaris/BSD distribution is a whole ecosystem, and that's one of the problems with the state of the whole thing right now. It's got too little in common, by design, with other distributions. The sheer weight of the project, the burden of maintaining things that by all rights should have been made at least halfway automatic ages ago, falls on too few shoulders.

The end result is what I have come with a heavy heart to realize is one of Linux's (and by extension, Solaris's and BSD's) ongoing tragedies: The price of diversity is eternal incompleteness. Since it's still next to impossible to provide a single binary package that is distribution-neutral, and compiling from source is still not as automatic a process as you'd like to think it is (Gentoo notwithstanding), the end result is that if you pick a distro, you're stuck with what's in it in more ways than one.

I was hoping Solaris would be different, but it seems they, too, are simply waiting for someone else to come along and do all the heavy lifting. Again.

InformationWeek Analytics has published an independent analysis of the current state of open source adoption. Download the report here (registration required).

Follow me and the rest of InformationWeek on Twitter.