> I´m afraid there are a lot of them. In fact installing them broke my system > completely since I need X. Here's what I still remember: > 1) xdm is started regardless whether it is confugured to do so. I have > no-start-xdm in my config file but still it starts up. This is a feature, not a bug. If xdm is installed, the daemon will attempt to start in the postinst, just like any other package that ships a daemon. Before starting, however, xdm uses the parse-xf86config script (in the xserver-common package) to attempt to determine if the XF86Config file is legal. If it isn't, xdm will not start. (no-)start-xdm was always, IMO, a crude hack to keep a daemon from running that some people might not want to use. I think a better solution is to separate out the daemons so that only people interested in them have to worry about them -- so that's what I did. xdm and xfs are now distinct packages. See /usr/doc/xdm/README.Debian, xdm.options(5), and /etc/init.d/xdm for more information. > 2) some preinsts call getreal() but the function is named readlink() Thinko. Fix will be in the next release. > 3) The server no longer is setud root. Thus it refuses to come up at all. > 4) startx only comes up under root (some authority problem). I'm now using suidregister to make /usr/bin/X11/X (which is Debian's X server wrapper) setuid root -- it ships with plain old mode 755. xserver-common (which contains the wrapper) was one of the files with the "getreal" problem, so the script bombed out before suidregister was called. With the fixing of 2) above this should no longer be a problem. > 5) xkbcomp refuses to work (some incorrect paths). I think I know what's up with this but it's going to take a little research. This should be a non-fatal problem. If the X server refuses to start, something else is wrong. > 6) Some more problems in pre/post-insts like incorrect test values or > incorrect parameters to start-stop-daemon. Fixed in the next release; Joseph Carter sent me his dpkg output and pinpointed these problems. > 7) The font packages have been renamed but don't provide the old names thus > breaking some dependecies, most notably the Gimp. Packages that depend explicitly on some X font packages should switch to the new names. Also, it's often a bad idea to have a dependency on an X font package, as the same fonts a package requires may be retrieved through an X font server. So "Recommends:" should probably be as strong a dependency as there is on the xfonts-* packages. When I next see Che_Fox in IRC I'll talk to him about the Gimp, and see if my reasoning applies in that case. > Since I take it this isn't meant to go into frozen it's not that much a > problem though. I did try to warn people. Now that the freeze has apparently been pushed back yet again, I regret the premature release, but I didn't know about it at the time. Sorry for the problems. I'm fixing them as soon as I become aware of them. -- G. Branden Robinson | It was a typical net.exercise -- a Debian GNU/Linux | screaming mob pounding on a greasy spot branden@ecn.purdue.edu | on the pavement, where used to lie the cartoon.ecn.purdue.edu/~branden/ | carcass of a dead horse.
Attachment:
pgpQLFMpo_kin.pgp
Description: PGP signature