Bug#165416: xfree86: no warning of meta/alt/windows key behavior change
On Sat, 19 Oct 2002 08:09:24 -0700, Marc Wilson wrote:
> On Fri, Oct 18, 2002 at 02:58:48PM -0700, David Parker wrote:
> > After reading the changelogs I figured out that I had to add the
> > following to my XF86Config-4 file:
> >
> > Option "XkbOptions" "altwin:meta_win"
> >
> > There was no warning of this change, and for a long time I was
> > unable to fix the problem.
>
> I'm sure Branden will be considerably more caustic, but the best possible
> answer would be... whyinhell hadn't you read the changelog in the first
> place, where this is explicitly described? This *is* unstable, after all.
I routinely do read changelogs, but in this case that wasn't enough. Maybe I
scanned over it, maybe I saw it and didn't recognize it as something that
affected me, but in the end I didn't get the message that something so
significant had changed. Let me quote the changelog entry:
* patch #052: deleted; dropped this Debian-specific meta/alt/windows key
thing because XFree86 4.2 has the "altwin" symbols file; Emacs users need
to check it out (/etc/X11/xkb/symbols/altwin)
First of all, I don't use emacs, so I can effectively disregard the last part
of the line. The remaining information mentions that "XFree86 4.2 has the
'altwin' symbols file", but to one who simply uses XFree86 and isn't intimate
with its inner workings, that piece of information means very little to me. It
also doesn't mention that it requires some sort of user intervention (since I
don't use emacs) so the changelog entry wouldn't be significant in my mind
until things stopped working and I was wondering why.
> You really gotta love the way people think nowadays... oh, no, I don't see
> any reason why I should look at a changelog and find out what changed in
> package "X"... I'll decide the KERNEL, which hasn't changed, must suddenly
> be at fault.
>
> Sheesh.
As I said, I had looked over the changelog. For lack of a better explanation, I
simply pointed out that the kernel keymap, as manipulated by the console-tools
package, was my first idea of where to look to fix the problem. I'm just
pointing out how a Debian user who doesn't also happen to be an X internals
guru reacted to this problem. No need to get edgy and elitist here.
> > I think that if this setting isn't added automatically to the user's
> > config file, it should at least be documented in README.Debian and
> > displayed in a dialog before the package is installed.
>
> For every person that expects that ALT is ALT and Windows is META, there's
> another person who expects that ALT is META and is going to lame about how
> you've broken their copy of emacs. Don't gratuitously add options to
> configuration files.
Agreed, modifying config files isn't the best idea. But the fact remains that
this behavior was (except for the changelog entry) silently changed from a
previously established behavior. There's no harm in displaying a debconf
informational message, which would get the message across clearly and ensure
that any whining users are really whining out of their own ignorance.
Maybe I'm missing something, but won't this change in functionality hit users
in the future when they migrate from sarge to the next stable codename release?
If you're suggesting that a relatively casual Debian user upgrading her system
from one stable version to another should be forced to read thousands of lines
of changelogs as part of the process, research and understand each one, and act
upon each one that might be a problem, that would be ridiculous. In my mind,
one of the goals of Debian is to make it accessible to users, and in this case
a simple change would prevent it from being inaccessible.
Let me end this message by thanking Branden and everyone else who contributes
to the xfree86 packaging effort. Don't let this bug report send the message
that I don't appreciate it; in fact, I'm simply trying to make it flawless.
David
-------
David Parker <david@neongoat.com>
Neon Goat Productions
http://www.neongoat.com
0xF90FFFE5 / F362 51F7 6D51 85EB AF68 75B9 D29B 1AFC F90F FFE5
-------
Reply to: