[RFC] removal of server generation mechanism

[Tiago Vignatti]

Alan,

Alan Coopersmith wrote:
> 
> Why?   Not that I won't mind having to explain less often to people why
> running "xset" with no other clients connected isn't going to be useful,
> but if the server is actually being used by multiple users/sessions without
> being restarted in between, there's other cleanup that needs to happen
> to prevent memory leaks, information leaks, etc.   For instance, xdm would
> need to change it's default to terminateServer and the other display managers
> would need to be checked to see if they need similar changes (I don't
> remember which ones already default to that).

XDM and colleagues are using so far because no one never bothered to fix 
it properly. I would like to remove regen because it's being used 
without real need and because display managers are abusing from a bad 
feature defined by the server.

If we keep letting the server defines such kind of features/policies 
[0], then we will never see a "thin" Xorg, that only its essence is what 
is need: a display server. For small devices, an 
almost-nothing-and-very-thin Xorg counts a lot (I'd guess the excess of 
policy is pretty much the motivation for the creation of Wayland, for 
instace).

Therefore, I'd like to see a strong argument to why _not_ remove such 
regeneration code.

... and I want to stop hear from people that X is a pile of dead and old 
code. I can dream :)


Thanks,

[0] and the X11 principle is also against: "Provide mechanism rather 
than policy. In particular, place user interface policy in the clients' 
hands."

              Tiago