Re: Debian specific gpm brokenness
In answer to Josip here too..
On Thu, Mar 13, 2003 at 02:24:03PM +0100, Thomas Hood wrote:
> I'm testing the new gpm. Unlike gpm 1.19.6-12, this version
> lacks the "fups2" protocol, so I switched to the "ps2" protocol.
> At first it didn't work. I ran /etc/init.d/gpm restart twenty times
> and on elevent of those occasions the mouse worked (I was able to
> move the pointer around the console); on nine occasions it didn't.
> Version 1.19.6-12 never failed on me.
Could you please grab the source package and try adding
usleep(50000) between the write and read in write_to_mouse
(and maybe read_mouse_id too) in src/mice.c to see if that
fixes it for you. If not, could you post any error output
gpm gives when it fails.
And could someone please explain exactly what it is that fu is
designed to fix, the patches have no documentation as to their
intent and much of it looks like a gratuitous (even if cleaner)
rewrite, which does nothing to ease my suspicion of it since the
changelog shows fups2 being added in -7, after this:
gpm (1.19.6-4) unstable; urgency=low
* Grr, changed the write_ps2 function to be more, tolorant, of
really really broken mice which reply with 2 ACKs instead of 1 for
setting sample rate. (closes: #120581, #120701)
When to my knowledge, what is really really broken is the write_ps2
function which doesn't wait for the first ACK (which the protocol
defines should be sent) before sending the rate data byte.
It seems write_ps2 works ok with tolerent mice and was then kluged to
'work around' otherwise conforming ones :(
(note both versions worked with all my mice, so I'm not much help
finding the real bugs here)
btw:
It still sounds like what we need here is an active maintainer
prepared to work *with* the new upstream (who is keen, but self
admittedly no code guru). I'm happy to help get it sorted out
but we need someone with time and interest in *gpm*, my interest
in it is far too oblique to stretch myself even thinner on a full
time basis. If someone has idle hands to raise I'd support any
move to get this package a dedicated maintainer again...
Ron
Reply to: