I recently read that Microsoft has recently improved the startup speed
of Internet Explorer to be much faster than Netscape. They accomplished
this by putting large chunks of what is done in IE startup into the
"operating system".
I was hoping that we could accomplish the same with XEmacs by dropping
as much of it as possible into device drivers. Or, if that is not
possible, into the upper half of the kernel. Disk driver code looks
like a good place to hide a lot of the lisp engine since that code is
already rather complex. Good software design mandates that complexity
be seperated out; this is why all complex code in a software system
should always live in the same file.
A major win would be that we could use all of the familiar Emacs
commands would be available at the login prompt and in single-user
mode. I have already begun writing a white paper and hope to get a
government grant for a study on rebinding "C-x C-c" to "sync; sync;
sync; halt" in single user mode.
I believe that this effort would win many converts from the FSF Emacs
crowd. We can begin with the more popular and freely available
operating systems like GNU Hurd and then move on based on that
experience. I will begin writing letters on impressive government
letterhead to Microsoft explaining our noble effort and demanding
source code to Windows 98 and NT. They have no choice but to comply
since they are already running scared because of the severe legal
action that the US government has arrayed against them.
Since Emacs already supports the concept of "the editor is the
operating system" I think that it is logical to rally around the idea
that "the operating system is the editor". This has prompted me to
file for a US Patent entitled "Software" that would give me rights to
this idea and many others dealing with programmable devices. (As a side
note I discovered that Apple never patented the idea of putting the
user interface and windowing system in the kernel and so I am also
proposing that X11 also be dropped into the kernel although into the
clock tick handling portion of it since X11, after all, also
contains event loops and timer procs).
I will soon be announcing a mailing list dealing with "XEmacs in
kernel" issues. Until then, let's not make any changes to XEmacs that
would make it incompatible with interrupts.
lonhyn
Show replies by date
Some time ago, Lonhyn T. Jasinskyj wrote...
|+
| I believe that this effort would win many converts from the FSF Emacs
| crowd. We can begin with the more popular and freely available
| operating systems like GNU Hurd and then move on based on that
| experience. I will begin writing letters on impressive government
| letterhead to Microsoft explaining our noble effort and demanding
| source code to Windows 98 and NT. They have no choice but to comply
| since they are already running scared because of the severe legal
| action that the US government has arrayed against them.
There's a simpler approach. Just sign a non-disclosure agreement, pay
$250,000, ans NT source code is all yours. Getting Windows 95 source
is rewarded $5 and a can of beer, but no one needs this crap anyways.
Kirill