[Wylug-discuss] best sig ever

[Dave Fisher]

On Tue, Aug 24, 2004 at 10:18:23PM +0100, Roger Leigh wrote:
> Dave Fisher <davef at gbdirect.co.uk> writes:
> >   OSS Printing Systems -- so many sub-optimal packages to choose from!
>
> This takes a lot of stick, but is it /really/ so bad?
> What are currently the worst aspects of printing?

My own experience has been mixed, but I don't think you can dismiss the
massive 'me too' reaction to Eric Raymond's cups usability rant as mere
ignorance.  Just because things _should_ work from a technical
perspective, doesn't mean they _actually_ work from a practical point of
view.

People's ignorance of may be a major source of their printing problems,
but advocates and developers need to understand why people are ignorant.

Could it be:

  * That the whole chain of tools and configs involved is simply too
    complex?

  * That there are too many factors involved and choices to be made,
    i.e.  most people don't have the time to master all of the
    specialisms that are required of them?

  * That the multiplicity of printing systems and subsystems makes
    documentation difficult and confusing?

  * That the documentation itself is so disparate and patchy that even
    knowledgable and determined people need days to collect, collate,
    understand and reassemble the stuff in a coherent manner.

As I said, my own experience has been mixed. Printing at work has always
caused the same grief (under lpr, lpr-ng and cups) i.e. jobs just
magically disappear from the queue or stick there block everything
else.

Printing with cups at home on a more uptodate version of Debian and a
newer printer is OKish ... but it was a bugger to setup in the first place
and it still doesn't seem to consistently obey the printing area
definitions in the foomatic ppd {sigh}.

I suspect that the issue at work may be ghostscript-related, beause
there does seem to be some association between dodgy postscript/PDF file
contents and queueing blocking/job-disappearance.

If so, it is still an OSS _printing_ problem, since gs is an essential
part of the printing chain.



>
> >   X Configuration -- Gurus with infinite time on their hands struggle to
> >                      wrestle its complexity to the ground!
>
> X can be nasty, but I've found most recent distributions get most
> hardware set up decently.  Hardware configuration in general is vastly
> improved now we have stuff like hotplug, which can set up all your PCI
> devices automatically.  I'll never need to touch modules.conf again!!

... hmmm ... I've just completed two installs, one Xandros the other
Debian Sarge.  Both failed miserably with PS/2 and USB mice, the latter
also barfed on a USB keyboard.

But I was actually thinking of software configs, especially X
authentication.  With the exception of the very insecure xhost + , just
about any combination of permissions blocks display from some host/app
or other.

BTW, have you ever tried to remap modifier keys in X?

For most people the Caps Lock is a complete waste of space.  For a
one-handed, right-handed person like me this is doubly cruel, since the
much-needed ESC key is usually in the far top-left.

After lots of grief and reading I finally managed to hand-code an
xmodmaprc file (xkeycaps barfs on multiple modifier alternatives) to map
ESC to the RightShift key and Super to the CapsLock.

Some time in the last year, the gnome folks decided that they knew
better than me.  Starting recent versions of gnome with X with my
original .xmodmaprc file in place disables the function keys (?), sends
the pointer mad (??!!) and ignores the desired remapping entirely.

Since I couldn't find any way of achieving the remapping through gnome's
own controls, I've gone back to a bog standard keyboard layout, no
Super/Windows key, and lots of RSI ... gnrrr!

>
> >   Character encoding -- at least Windows gets it wrong consistently!
>
> What problems have you had in this respect?
>
> This is one area which is improving dramatically, and should be fixed
> for good in a year or so (I hope!) when we are all using UCS/UTF-8 as
> the standard encoding.  Are you using a UTF-8 locale?

I try to use a UTF-8 locale, but I haven't yet mastered the art of
working a year or so in the future ;-)

Things are definitely getting better, but projected completion of the
transition seems to have been about a year off for quite some time now.

In the meantime, different bits of my home systems demand C, Latin1, or
UTF-8 depending on their mood {sigh}.

>
> >   Documentation -- Does it even exist? Which one of a dozen formats is it in?
> >                    What were they smoking when they wrote it? Have they even
> >                    heard of the concept of an illustrative example?
>
> I have to agree here.  It varies much between projects, especially
> with examples.  Some is superb (e.g. PostgreSQL), while much is
> shocking (I had to set up a DHCP and BIND setup with dynamic DNS
> updates today.  The BIND folks don't like to give too much away: two
> trivial example named.conf snippets in the whole manual, and NO
> example zone files at all--how on earth is a newcomer to understand
> it!!).

Oh yes, I had almost exactly the same experiece with BIND and DHCP
recently.  I think this now puts BIND in first place for my unnofficial
"Worst Manpage on Linux" award.  The tar manpage is a model of
transparency by comparison.

Dave