NixOS is an ingenious construct, and, despite the fact it is less than a year old, it includes enough packages and versions that you can easily put it through its paces.
However, the solutions to long-standing problems that NixOS provides seem questionable in a couple of ways. To start with, abandoning the long-established file hierarchy standards of Unix seems a drastic step. Perhaps Conary's solution of adding version control and specific package names for separate versions of the same application is more economical than such an extreme alterationl? If nothing else, a system that is so obviously non-compliant with the Linux Standards Base is unlikely to be adopted by any major distribution, regardless of how useful it may be.
In addition, on a practical level, NixOS depends on unlimited hard drive space. Unlike with other package management systems -- as imperfect as they sometimes are with cleaning up after themselves -- nothing gets removed unless you specifically make the effort to remove it. In the short term, this arrangement is not a problem, but over a three or four year lifespan of a machine, it could create a serious shortage of hard drive space. On a network, administrators would almost surely want to run it with set limits on the storage allowed for each account to prevent a handful of users from gobbling up all available space. Long-time users, too, accustomed to jealously preserving hard drive space, might find the idea of keeping everything an extremely inelegant solution.
Still, the point of an academic distribution like NixOS is not necessarily to gain widespread popularity. Rather, the purpose is to experiment and offer alternatives in areas that developers and users often take for granted. From that perspective, NixOS is an intriguing success, and well worth investigating.
This article was first published on LinuxPlanet.com.