Ubuntu & Linspire

[Derek Broughton]

Daniel Pittman wrote:

> Derek Broughton <news at pointerstop.ca> writes:

>> I have my doubts about "statistically significant" and "measurable" (a
>> little redundant, anyway - you couldn't have a statistically
>> significant _unmeasurable_ value:-) ), but I have to agree that it's
>> pretty pointless.
> 
> Rhetoric needs no recourse to proper grammar!

LOL.
> 
> Anyway, if you have are (probably rightly) not convinced by my assertion
> about the effect of custom kernels I suggest that you (or the silent
> audience) consult the archives of the Ubuntu developers list.

No need.  It's just a little doubt, and I'm agreeing that even if there is a
measurable difference, and if it could be statistically significant for
even some users, it can't make enough difference to the average (even
hardcore) user to go to the trouble of compiling a kernel - even if kernel
compiles are getting easier.
> 
>> I used to have to compile my own kernel for a driver that wasn't in
>> Debian. I haven't needed to do that since before Warty was a piglet,
>> and I've never compiled a kernel for Ubuntu.
> 
> These days you can typically compile a driver without needing the kernel
> source installed; Ubuntu and Debian have worked hard to ensure that
> "just works."

Well, not having compiled either a kernel or a driver in a long time, I'm
not sure, but I swear last time I loaded kernel sources it was barely
larger than the kernel headers, so I'm not sure that's so significant.

>>>> Wacom devices in xorg.conf?
>>> 
>>> I guess "hardcore" users don't own Wacom tablets, but they do own USB
>>> mice, right?
>>
>> Pretty much.  That would also match most of the "softcore" users. The
>> wacom devices in xorg.conf isn't nearly as annoying for softcore
>> users, though, because they never run KDE apps from the command line.
> 
> *nod* Personally, I just edited the X.org configuration to remove the
> input device configuration when that bothered me. ;)

If it bothered me, I would, but last time I modified xorg.conf I ran into a
confusing set of instructions about how to do it and still get updates from
the package system, so I'd rather just stick with the currently working
automagic version of xorg.conf.

>> If the errors were just suppressed you wouldn't see this complaint.
> 
> This, however, I couldn't disagree with more strongly.  Suppressing
> errors is *never* the right response, no matter what.

hmmm.  My bad, since I'm a firm believer in that.  Nothing is likely to get
me on a rant faster than trying to debug a program where the
programmer "helpfully" caught all errors, then at best printed a message
like "error".  However, these shouldn't be errors.  Not being able to
configure a device that isn't connected is not an error.  It _should_ be
normal behavior.  In syslog parlance, this would be 'debug', or even 'info'
level, but certainly not as much as a warning, and should not be displayed
in normal output (though, running X apps in a konsole is hardly "normal"
output - it's not like there's a popup dialog!).
-- 
derek