[linux-audio-user] Distros and Desktops

Just noticing that the discussion of 'audio' distros tends to primarily revolve around the binary star system of Planet CCRMA/DeMudi. I know there are people on this list succesfully using Gentoo with unpatched kernels and getting outstanding performance w/re to latency. We have found Arch Linux to be an excellent choice for audio use, and feel it has many advantages over the 'big 2' - namely streamlined size and the fact that they stay very cutting edge with their kernel and base packages while offering a good selection of audio apps. (We are using an unpatched kernel, and even, heretically, doing everything as root - with no issues.) Different distros are optimized for different purposes, and it appears that RH/Fedora, Debian and SuSe are focused on the enterprise/server market, where stability is paramount, and performance is measured in terms of server optimized parameters. CCRMA and DeMudi being tied to RH/Fedora and Debian might prove to be a weakness in the long run - (or their greatest strength, depending on your perspective.) Distros which are either comparatively streamlined (like Arch), or highly customizable (like Gentoo) have a lot of advantages in that they are small, easy to install and stay much more up-to-date, allowing one to take advantage of the most recent kernel improvements - which lately, have a great deal to do with preemption/latency, scheduling and memory management - all crucial for audio apps. Perhaps the need for specially patched audio kernels is nearing an end, and a well-configured vanilla kernel can perform well within the limits required by audio producers. I think that it is the use of large, graphics-intensive, multipurpose distros which has brought about much of the need for kernel-patching, disk-tuning, and requirements for high-powered hardware. In audio production, functionality and usability is what matters more than appearance (imo). Sometimes that usability necessitates gui features that increase the complexity of an app, but I think this should always be weighed against the possible performance degradation that may, as a consequence, be introduced. I'm of the smaller-and-faster-is-better school, since that is consistent with my experience. While a pretty desktop or pretty app may have aesthetic appeal, I'll take a plain version any day if it is faster and allows me to be more productive. To my mind, the ideal, dedicated audio distribution would have the absolute bare minimum of base programs required to run the audio apps one needs, and the audio apps should be light on graphical features other than the ones required for functionality. The graphics server should be very fast and lightweight, using a minimum of resources, and similarly for the desktop/WM. I would like to see an audio distro that behaves much like an embedded system - i.e. the OS being specifically dedicated to optimally interfacing with the DAW hardware, and the (normally) small number of audio apps being used. As in everything else, the more choices available the better - everyone having different needs and preferences, and I understand why some desire to create Windows-look-alike DEs to demonstrate that Linux can do it too, and attract current Windows users to try Linux. I would just like to be an advocate for scaled-down, audio-optimized distros with an emphasis on simplicity and performance achieved through a minimum of small, fast programs and well-configured, up-to-date kernels. (Am working on this. :) ) I am even wondering if there might be a place for a special, audio-optimized file system - AFS - whose journaling, write behavior, etc. is specifically optimized for audio recording. We are having very good results with XFS, and it may already adequately serve audio requirements, but I still wonder if there could be an even better FS designed for audio? -Maluvia