Software Blog

Some my quick but important thoughts about future operating systems and how efforts of the software development should be concentrated:

Now we could weep: Oh, if Linus Torvalds would write not in C but in Ada... Oh, if then that time there would be also many other Ada programmers which would write a decent operating system like Linux and many software packages but not in C (and C++) but in Ada...

[More:]

But to weep would be irreasonable simply because what is done cannot be changed.

We have many packages, autoconf, automake, and much other messy silly stuff. And right now I deal with problems like incompatible Debian package versions which would require to much downloading for an upgrade. Oh, if Linus and friends would write in Ada instead of C, then packages would be more compatible and I even cannot imagine how much less other problems we would have. Now we can only weep about this.

But lo! what may be important in the present time, when userspace filesystems are already available and saying GNU in GNU/Linux name makes sense:

There are very long time but yet unfinished software project (commonly called GNU in the proper sense of this word) to develop an operating system operating in userspace with minimal kernel (and to develop software for this system).

Now when the time has come... We should abandon this project and instead of developing GNU concenrate as much programmer's effort as possible on rewriting as much of Linux software libraries in Ada95 instead of C/C++.

This is more important than to develop Richard Stallman's dream GNU. Filesystems in userspace anyway work even with the current Linux. I deem that we do not much need a completely new kernel now. (Really, why we would need it? Linux works.)

We (historically!) should not repeat again Linus Torvald's wrong choice of C (rather than Ada). If he would know that time... But we know now!

I want even to call strongly: Abandon system programming for GNU project protesting that it has not enough Ada support. (I do not call to abandon Debian GNU/Linux which is our so to say today bread, but that Richard's dream imaginary GNU system.) Or better yet, do Ada support for GNU/Linux. I mean: switch to other software projects, such as common inter-OS protocols, compilers, libraries, Gnome, maybe Java, etc., not GNU core. Write current device drivers for Linux better.

Maybe now it is due time when all this can be corrected. It may probably require 20 years or so or more to correct if we will err now... Take the right choice in the root. The right choice is Ada based OS backward compatible with the present state of GNU/Linux.

We need not Richard's dream GNU. We instead much more need Ada operating system (which would either run Ada software and libraries on Linux kernel) or whose would be itself written in Ada and emulate Linux. (Maybe it is not reasonable to rewrite Linux kernel in Ada, but system software and libraries need to be rewritten).

If anyone disagrees and/or deems that I insist on this too strongly and radically, you may comment below. Maybe my opinion is not right and you have a better reason...