On Fri, 2013-03-08 at 12:00 -0600, Dan Williams wrote:
On Fri, 2013-03-08 at 12:33 -0500, Adam Jackson wrote:
> I still don't understand why that timeout is there at all. What is it
> meant to fix?
Hanging for 10 seconds waiting to see if your GPU supports enough GL to
run the shell before falling back to the classic one. All that wait
time delays showing the GDM greeter, so faster bootup pretty much
requires a bound of some sort on this operation.
I don't think that argument makes sense.
One: if GL initialization takes meaningful time, it can really only be
because the kernel decided not to schedule you yet. Because,
presumably, your disk is busy with whatever, readahead maybe or maybe
you have a mail queue and sendmail is busy shoving stuff out. Or you're
in a VM and it's a busy machine. At any rate that test does not hang
forever, it merely takes time to complete. There's really not a sense
in which you will display things _faster_ if you decide to time out.
The pure-X window you create, modulo a few million instructions of GL
context creation, will show up at about the same time the GL-composited
window you would otherwise create, because whatever it is you're
contending with _is still going on_ and is the whole reason you might
bump into the timeout.
Two: Why is there still a classic greeter if there's no fallback mode?