On Thu, Jun 09, 2005 at 03:03:34PM -0500, X Strike Force SVN Repository Admin wrote:
> + Restore original documentation of how to restore automatic updates of the
> configuration file, as Ubuntu's replacement assumed 1) unprivileged users
> would be able to create files in /etc/X11 and 2) the optional "sudo"
> package would be installed.
Thanks for the catch, but your instructions are still entirely
unnecessary.
How to restore your configuration and have it rewritten:
sudo dpkg-reconfigure -phigh xserver-xorg
> + Add extensive comments documenting Ubuntu kludges and my thoughts on how
> to better solve the same problems.
> +# XXX: Kludge warning: This is used to bolt on an InputDevice section for a
> +# Synaptics touchpad. This should be 1) moved to a debconf template; 2) made
> +# configurable; and 3) more specifically detected (so that the answer to the
> +# debconf question can be predetermined). It's not dexconf's business to
> +# perform any hardware detection whatsoever.
> +#
> +# laptop-detect doesn't appear to be packaged for Debian, so this is a no-op for
> +# now. See below for further caveats if/when this code comes alive.
laptop-detect is just a small shim around dmidecode, but detecting
Synaptics devices (they just leech off PS/2) is an incredibly hard
problem that the kernel's psmouse only just barely gets right. Our
solution, which I still think is the best one, is to just throw the
synaptics driver at all laptops. I haven't seen it get any false
positives and cause breakages.
> +if which laptop-detect >/dev/null 2>&1; then
> + if laptop-detect >/dev/null 2>&1; then
> + LAPTOP=true
> + fi
> +fi
I now count one removal of this section, for two additions.
> +# XXX: This is some new template Ubuntu added. Probably better solved by just
> +# making empty entries for sync ranges syntatically valid, as Sven Luther
> +# suggested.
No, not at all. We avoid writing sync ranges whereever possible, so
u_s_r is a way for the postinst to tell Dexconf that it needs to write
out the sync ranges it has.
> +# XXX: Sync ranges kludge continued. See "use_sync_ranges" above.
It's not a 'kludge', it's working around the braindead behaviour of
writing out sync ranges whereever possible. Done, I dare say, in a
reasonably clean fashion.
> +# XXX: Hoo-ha. Another awful kludge. If this mode isn't defined upstream (or
> +# is defined wrongly) in either of
> +# xc/programs/Xserver/hw/xfree86/etc/{vesa,extra}modes, that should be patched.
> +# Furthermore, because everyone hates the limited mode selection list in the
> +# debconf template, we probably ought to generate it at package build time by
> +# grepping those files.
I agree. However, I most certainly didn't author that change, and ISTR
its lineage coming back to XFree86. (I've patched the {extra,vesa}modes
harder to include all known widescreen laptop resolutions.)
> (Of course, for laptop/flat panel users, what we really
> +# want to do is query the monitor from the X server's config script, and
> +# pre-answer the question with the display's native resolution so that people
> +# with a high question priority won't be blitzed with the truckload of modes
> +# that the aforementioned grep method would present them with.)
This is, um, what we've been doing since last October.
> -dexconf \- generate XFree86 X server configuration file from debconf data
> +dexconf \- generate XOrg or XFree86 X server configuration file from debconf data
Xorg (or X.Org), not XOrg.
> +The former is used by the
> +.BR XOrg (1x)
Ditto, but this is even worse because that manpage *does* *not* *exist*.
> +intended to replace a full\-featured X server configuration tool; that is the
> +province of
> +.BR xorgcfg (1x)
> +(for
> +.BR XOrg )
^^^^
Again. Although xorgcfg is often less useful than the Debconfage, as
much as it is a crippled series of hacks which need to be taken out the
back and put down as soon as possible.
Attachment:
signature.asc
Description: Digital signature