Why write a thousand comments when I can write one post? My recent post, Are Gnome and Ubuntu ruining the Linux Desktop? has garnered a lot of interesting comments, including some with useful information, some thoughtful commentary, the scratchings of a couple of wet hens and one rabid Canadian that I had to delete from the internet.
This post is a response to some of that commentary.
First, let me say that while I was asking the question “are Gnome and Ubuntu ruining the Linux Desktop” … and it is a valid question, I have to say that I like Ubuntu A LOT and I also like Gnome. Ubuntu, running Gnome, is my distribution of choice.
But that is not what my post was about. What I’m asking is this: As a mainstream distro goes more and more mainstream and time goes buy with the accretion of this and that new feature, is it not likely that eventually the system will become weighted down and too much oriented to eye-candy?
One issue that came up is the use of, and the wisdom of the use of, libraries. I thought libraries were a good thing. I thought libraries were how well designed code gets re-used, and how core functionality can interface with software on one hand through the libraries’ API while continued work on efficiency and bug-squishing could happen on the other side. Am I wrong about that? Is this really a problem:
Developers don’t realize that using a library / libraries is a trap that you cannot get out of. Once in, you’re stuck with the methods and every bit of (un)logic thought of. Therefore lxde will become the new Gnome.
I’m not necessarily disagreeing here with Monzo, I’m just not entirely sure.
The “Unix philosophy” vs. not issue has also come up, and I want to throw an idea out here: I was just reading something by a well known famous computer guy who noted that Emacs is implemented the exact opposite way of what one would expect for the Unix philosophy. Instead of making a bunch of little tools that work together, Emacs is almost like an entire operating system thinly disguised as a text editor.
Well, it seems to me that that can work as well. Having a high level meta-system, an application that can directly access the shell (and more) and that a reasonably sophisticated user can adapt (maybe in a way that won’t run on your 10 year old laptop ….. but still) as one way of making the CPU and the peripherals do your bidding, at the same time as developing these bits and pieces, is one approach. After all, when we put our bits and pieces together to make a script, we are essentially making a mini system designed to do a (probably) unique set of things with a unique set of data in a unique context.
If only Emacs were just a little bit more extensible….
Finally, I want to make a concrete suggestion regarding Gnome: I want a version of Gnome that comes as source code has many/most features listed in a simple way that even I can understand, so that I can check what I don’t want vs. do want, and compile a native version. I would not have tool tips or the ability to produce tool tips. I would not have desktop wigets (or whatever they re called) or the ability to install them. I would have two versions of nautilus … one fully functional the other very bare bones. And so on. This would surely be the most efficient way to deploy a desktop, yes?