From: kennu@mits.mdata.fi (Kenneth Falck) Subject: Re: Linux vs. the world Date: Sat, 8 Aug 1992 18:34:37 GMT
In article <1992Aug8.035226.29102@colorado.edu> drew@hamlet.cs.colorado.edu (Drew Eckhardt) writes:
>>I see only one problem there; if all the hackish kind of programmers
>>that actually port the largest and most difficult programs use
>>new "hackers'" kernels, won't the ports easily become incompatible
>>with the old kernels and everyone has to update his/her kernel anyway?
>
>Probably not. Most well behaved programs (This does not include
>windowing systems, that need some hackish way of accessing the hardware
>directly) use a standard interface to the syscalls. While the
>underlying code for the sys calls may change in newer kernels,
>the interface remains the same and user level software will be portable
>across the different systems.
>
>Of course, there are exceptions to this, such as the MGR and X11 ports
>which needed the IO permissions bitmap, mmap(), and unix domain sockets,
>and the new mount(2) call that works with different types of file
>systems.
Hmm.. I wonder if releases of this "really big" software like
X-windows (and maybe also MGR) could be synchronized with the
major Linux releases; so that for some time we would have
"Linux 1.0", "X for Linux 1.0", "MGR for Linux 1.0", "GCC for
Linux 1.0", and then after some time "Linux 1.1", "X for Linux
1.1", etc...
Of course, nothing would prevent releasing something like
"GCC 1.1 for Linux 1.0", but it'd be cool if you could always
always tell just what version of Linux the software is meant
to be run on just by checking the program's version.
As far as I can tell, this far you've usually had to check
comp.os.linux all the time or try by yourself to see whether
some big piece of software works with some kernel. (Of course,
usually everything works with the latest kernel...)
I don't know then, would this work nicely if everyone recompiles
his/her kernels and makes big changes to them. I hope the kernel
becomes more externally configurable in the future, though I
suppose most of the changes people make themselves are just
little bug patches or keyboard configuration.
I wonder if there should be (or is there?) some "Linux committee"
that would coordinate everything, as I'm sure Linus can't do
everything by himself. (I don't think this newsgroup acts as
any kind of authority, but I haven't introduced myself to the
mailing lists yet though.)
-- kennu@mits.mdata.fi <here's where the dumb quote is supposed to be>-- kennu@mits.mdata.fi