[mythtv-users] protocol version mismatch, what's the best solution?

[Osma Ahvenlampi]

On ke, 2006-10-04 at 02:01 +0200, Axel Thimm wrote:
> On Tue, Oct 03, 2006 at 07:43:00PM -0300, Rick wrote:
> > None of this explains why a yum update and smart update upgrade on the 
> > appropriate machines running the same os, Fedora,  one after the other 
> > gives the protocol mismatch on the backend (for me at least).  Surely 
> > this is the exact way one is to update a system via the packages.
> 
> Indeed, that's the proper way. Assuming that you don't have any funny
> options with exclude/include/flagging/pinning/repo weiging or any
> mechanism that can be summarized as selective/partial then the only
> thing that you may be missing is a restart of the frontend and/or
> backend.

I think the problem many users were bitten by was one of "yum update
myth*", which did not include libmyth. Why would people do this? Because
many of them would rather avoid updating most parts of their (possibly
dedicated) HTPC system in search for a more stable base (for example,
updating the kernel necessitates updating or reinstalling other
components like NVidia or ATI drivers, lirc, etc).

Rightly or wrongly, since users do this, it is by definition a
reasonable thing to do. One must assume that users are not acting
unreasonably in significant numbers. They might be acting under limited,
partial or even false information, but they ARE acting reasonably given
the information they have.

Thus problems like this invariably are a matter of a) less than perfect
packaging, b) less than perfect user education, c) less than perfect
change management or d) combination of these factors. 

This is not meant as a critique, but as a statement which includes no
judgement of good vs bad. It's a fact of life, and only by addressing
one or more of these factors the problems are reduced. 

What doesn't work is saying that users are unreasonable (as some posts
in this thread have claimed, though not those of Rick's or Axel's) -
while some may be, that is simply not true in large numbers.

-- 
Osma Ahvenlampi   <oa at iki.fi>    http://www.fishpool.org