Michael A. Peters mpeters at
Thu Jun 19 22:42:59 PDT 2003

On Thu, 2003-06-19 at 11:34, Dagmar d'Surreal wrote:
> Here's some facts for you...  You don't appear to really understand
> which parts are responsible for what, which you demonstrated by
> repeatedly referring to whether or not GDM could handle missing hardware
> (among other things).

Doesn't matter.
XFree86 4.3 is what the blfs book has.
My XFree86 was configured according to the BLFS book.
There were NO issues with starting gpm without the mouse that would not
also happen with the init script method.

Perhaps previous versions of XFree86 would croak if started without the
mouse device when configured to have one, but the current doesn't.
Instead it handles it intelligently by simply not loading support for
any input devices.

This is a problem - but NOT a problem with XFree86 - it is a problem
with gdm. gdm should check for the devices and if a mouse/keyboard is
missing, it should start gdm but not switch to that tty (problem being
once you have switched to that tty there is no way to switch to another)

So I was correct that the bug is with gdm.
It's called a usability bug. It shouldn't put you into a console you can
not get out of.

>   You weren't even able to properly break your own
> system to simulate any of the suggested failure conditions, otherwise
> you would have been able to see what happens when XFree86 can't find a
> core pointer as opposed to how init reacts to GDM failing.

Bull fricken sh*t.
I unplugged the mouse device.
My XFree is configured to use a mouse.
There were no issues - XFree started just fine.
Wasn't usable - but DID NOT DIE or have any issue with lack of core
pointer. It (XFree) just simply started w/o support for core devices
(hence the dead keyboard) rather that die - as older versions of XFree
might have done. This would also happen if started as an init script.

You are flat wrong here.

>   You claimed
> that scripts being invoked through inittab were subject to accidental
> user-interruption, and originally cited this as your reason for invoking
> gdm directly from inittab, even though later you, yourself stated that
> /usr/bin/gdm is _also_ a shell script.  About the only thing you've
> managed to get right is your assertion that gdm doesn't need a valid
> login shell.  If what you said about RedHat calling /usr/bin/gdm as the
> script for runlevel 5, I am very sorry for them, but I gave up on
> expecting sensible behaviour from them long, long ago.

There have been no reports that I have heard of this being in issue -
and they started doing it in RH 7.0 which is enough time for an issue to
have surfaced.

> Repeatedly I have tried to point out these errors to you, and you're
> either impossibly dim, unbelievably clue-resistant, or deliberately
> being argumentative to waste other people's time and bandwidth.

Anyone who wants to repeat what I have tested will see who is the one
who has errors, is impossibly dim, unbelievably clue-resistant, and
deliberately arguementative.

In fact - you can feel free to repeat my tests and see for yourself.
This thread can't bear any more good fruit.

My case has been made, objections have been made, the authors of the
book can decide what they want to do - be it a total change, a comment,
or nothing at all.

But this thread should probably die.
When threads resort to name calling they have gone beyond useful

Unsubscribe: send email to listar at
and put 'unsubscribe blfs-dev' in the subject header of the message

More information about the blfs-dev mailing list