[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: gpm in standard or not [was: Re: interesting times installing 2.4.17 in Woody...]



On Sun, Feb 24, 2002 at 12:04:03AM +0100, Josip Rodin wrote:
> > > I've been on the edge of putting it in optional for a while now, if the
> > > general consensus is that it should not be standard then my next upload
> > > (which should fix the current big bugs, though they are VERY much kernel
> > > bugs as well.).
> > 
> > There's still room to make a task pull it in, if we can pinpoint a good
> > task to put it in. The closest existing task is unix-server, but it's
> > not _quite_ the right place. Still, I don't think we'll get any more
> > upset users leaving it out of standard than we did not installing
> > standard at all in potato!
> 
> Any task for people who want to use Linux console should include gpm, or any
> other program that allows copy&paste. I still believe there's a lot of
> people out there that won't run X for all their needs.

Agreed, but if gpm is installed as a standard package, it reasonably must
not be configured automatically.  The other option is to push for wider
adoption of the input core drivers in 2.5.x, which likely will happen
anyway.  As it stands, gpm is an unnecessarily complex program.  It's much
a house of cards.

You have to understand that the very act of probing the PS/2 port to see
if there is a mouse connected causes all input from the AT keyboard to
stop.  This is unreasonable for any package installed as standard since
failure to boot the machine with a mouse plugged in should not involve ssh
to shut down the machine or hitting the reset button and praying nothing
gets corrupted.  Journalling filesystem?  Er, in potato?  This problem's
existed for years now.

Of course, the kernel knows if there is no mouse connected to the PS/2
port; it checks that as soon as it identifies that there is a PS/2 port.
Userspace programs, however, get to take their chances.  It's a lovely
problem that's been given a bandaid solution so far, the input core
drivers are the only real fix in sight.  The Linux Console project is also
trying to revamp the VT system quite a bit.  Hopefully when they're done
with it, gpm may not be necessary any longer.


> I've just been told about this thread -- I wouldn't think we'd make
> decisions about whether a package is to stay in standard as a subthread
> of a thread called like this. (One could even understand this as sneaky
> and evasive. >:>)

*shrug* I saw it, and I don't read a third of what gets posted to this
list anymore.  =)

-- 
Joseph Carter <knghtbrd@bluecherry.net>      You're entitled to my opinion
 
Isn't it embarrassing when you have to go to the drugstore for some
"special items", and when you're checking out, the cashier looks at you
like, "oh, I know what YOU'RE doing tonight..."
 
Yep, that cashier read all the signs... canned chicken soup, TheraFlu,
Halls, NyQuil, the bigass bottles of OJ and grapefruit juice... he knew
and I knew that I had a date with the teevee and a down comforter. Awww
yeah.
        -- Elizabeth Kirkindall

Attachment: pgpecq71Vm3Vg.pgp
Description: PGP signature


Reply to: