[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Request for policy advice on X server manpages

On Fri, Mar 27, 1998, Branden Robinson wrote:
>My solution to XF86Config and X server manpage issues:
>Make a new package, xserver-base, on which all xserver packages depend,
>which contains:
>/usr/doc/xserver-base/README.Debian (to be written)

Generally this sounds like a good idea. One thing though, the package
should also contain


since the X-servers need it for colour lookup. (Some clients, such as
xcolorsel, also need it, but most don't.)

>1) Strictly speaking, while users of those unaccelerated servers will
>need the xserver-base package, the XF86_Accel manpage will be useless to
>them.  But I'll be damned if I'm going to make an "xserver-accel-base"
>just to hold that one manpage.  I'm willing to waste 10334 bytes of
>space on the drives of xserver-{mono,vga16,svga} users.  Good judgement or

I for one think that xserver-accel-base would be to overdo it a bit...
10k is not going to kill anyone who intends to run X.

>2) I am not an xdm guru, and I don't know what (if anything) would be
>necessary to move into xserver-base to handle XDMCP sessions.  Or would
>such people need xbase anyway?  I've never personally configured a Linux
>machine that ran it's own xserver without also running some clients locally
>as well, so I need to know what I need to do make it possible to eliminate
>xbase entirely for some scenarios where that would be the case (no local X
>clients).  I must assume such scenarios exist, else why haven't the X
>servers depended on xbase all along?

To run an XDMCP session you need an X server (i.e. xserver-* as
appropriate). An /usr/X11R6/lib/X11/rgb.txt if you expect to use
symbolic colours (many clients do) and fonts (actually, the 'fixed'
font is enough to get going). They can come either from disk or from
a fontserver. Therefore I think that the xserver-* packages should
Suggest or Recommend (probably Suggest) xfntbase.

>3) xbase ships with a symlink /usr/X11R6/lib/X11/XF86Config ->
>/etc/X11/XF86Config . However, /etc/X11/XF86Config is not guaranteed to
>exist when the package is installed (if the user doesn't accept the
>postinst's offer to run xf86config, for instance).  I think it's nasty to
>ship a broken symlink, and I was going to have the postinst make the
>symlink if and only if xf86config created one, but this has two problems:
>a) at least one developer feels that anything in /usr should be shipped,
>broken symlink or not; I didn't hear his justification for this, but I
>generally trust his judgement.
>b) Just because xf86config doesn't buld an XF86Config file doesn't mean
>the user isn't going to have one -- he/she may have a known good one ready
>to move into place, and they shouldn't have to make the symlink manually.

I would vote for leaving the symlink in (I suspect there are users out
there who are not aware of it, and wouldn't know to make it after
copying their friends XF86Config into /etc/X11). I don't feel that the
dangling symlink is that bad, as it needs to be 'undangled' before X
will run, but perhaps you can put a placeholder of some sort in.

 -- Of course I'm crazy, but that doesn't mean I'm wrong.
Anders Hammarquist   |       Mud at Kingdoms        | iko@netg.se
NetGuide Scandinavia |   telnet kingdoms.se 1812    | Fax: +46 31 50 79 39
http://www.netg.se   |                              | Tel: +46 31 50 79 40

To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Reply to: