My first DontZap use case while testing F11 beta

[Anders Rayner-Karlsson]

* Matthew Garrett <mjg at redhat.com> [20090417 18:52]:
> On Fri, Apr 17, 2009 at 08:43:28AM -0700, Christopher Stone wrote:
> >  2009/4/16 Jesse Keating <jkeating at redhat.com>:
> > > On Fri, 2009-04-17 at 01:37 +0500, Suren Karapetyan wrote:
> > >> You will not find anyone @RH who will agree that
> > >> previews behavior was better
> > >>
> > > On the contrary, you'll likely find many people who work for Red Hat
> > > that would prefer the old behavior, they're just not wasting their time
> > > by arguing about it on a distribution mailing list.
> > 
> > So what are they doing instead? Absolutely nothing?  How ridiculous.
> > Let's just make a bunch of extremely bad decisions and keep bad
> > defaults because actually trying to convince someone that the defaults
> > or decisions people make are bad is pointless.
> 
> The people to convince are the maintainers of the code in question, and 
> in the general case sprawling threads on distribution mailing lists 
> aren't a good way to do that. Some worthwhile ideas came out of this 
> discussion and were implemented - however, at this point it's just 
> repeating the same arguments over and over again, and argument ad 
> nauseum is not the most convincing approach.

While I am not a representative figure of the user population, I had a
case today where:

 * Thunar made a "grab" and I could not click to change focus away. To
   all intents and purposes, I was in the situation that the
   contingency for preserving C-A-Bs as enabled by default is holding
   up as the primary reason.

 * Switching to vc/2 gave a blank screen

 * My sshd setup is to only allow pub-key auth coming in to my laptop
   and I did not have the right keys on another system yet set up
   (that's now rectified), so ssh in to kill Thunar didn't work

I was doing some pretty intricate stuff for several customers in
parallel, and shooting the session through C-A-Bs was out of the
question *even though it was available as an option to me*.

I ended up logging in on the console "black screen", because I knew
there was a login there, and typing blind, I killed thunar off to
recover my session and the hours worth of work I had sitting in my
session.

So - from my perspective, being one of those hate-object "@RH" people,
while C-A-Bs was available, my choice was not to use it because the
work was more valuable than the 3-4 minutes to assess the situation
and to try a few ways to recover it.


C-A-Bs is akin to Alt-SysRq-C if you have SysRq's enabled. The default
setting should be "off" so that people by accident can't trigger
it. If you need it - you get to enable it yourself. That is a sensible
default for a debug function.


In all else you wrote Matthew, I agree with you.

(And while I have not yet filed a BZ against Thunar, that is
forthcoming shortly.)

--
/Anders