> These kind of esthetic arguments miss the point. If I start with the base > system and begin working my way from admin to x11 in the packages. If it > takes me several weeks to get to x11, packages there have grown and dragged > other packages along potentialy all the way back to base. This process has a > strong possibility of never comming to closure. If some file structure can > be made to remain fixed at any arbitrary point in time (say for instance the > release date) then some kind of complete system can be constructed from the > packages in this "static" file structure. At this point the system > administrator can then look toward the bleeding edge of the distribution for > those features or fixes that may enhance the system.
Disk space and limited bandwidth are valid points, and I hadn't thought of the "dragging along" problem. I wonder if the latter wouldn't be a quite rare problem, however, not a "strong possibility," since I guess it is rare for a package update to make its dependencies more stringent (i.e., requiring a more recent version of another package). I might well be wrong, it is only a guess. By the way, are you saying that improved documentation and increased network security is an "esthetic" improvement in Debian? Or else what do you mean?
