On Wed, 29 Mar 2006 23:37:08 -0500 Michael Jennings <[EMAIL PROTECTED]> wrote:
> Has anyone thought about auto-detecting things at runtime rather than > build time? Yes, I thought about it. It was even discussed in the mailing list. That's actually what triggers the keyboard locking problem in the first place. If you don't let X know which VT to run on, it tries to autodetect the VT, but it gets into a race condition with init or a getty which is also allocating VTs at the same time. The end result is that X gets confused, and sometimes ends up with the display on one VT, but the keyboard redirected to some other VT, and with VT switching disabled for some reason. Trying to autodetect by entrance at run time will have to do so before it starts X, and that is just as likely to fail as much as X is for exactly the same reasons. At that point in time VT allocations are being made by different processes that have no method of co operating to make sure they don't step on one anothers toes. Autodetecting at install time at least has the advantage for the user that the VTs are already in use, making it real easy to find the current X VT if X is running, and only slightly more difficult to find an unused VT if X is not running. Distro maintainers usually hardcode a VT into the other display managers, they can do the same for entrance. Package maintainers are in a middle ground. My solution is sub optimal, but it's the best one we have.
signature.asc
Description: PGP signature
