[mythtv-users] Protocol version mismatch

Paul Bender pebender at san.rr.com
Wed Sep 27 11:47:27 UTC 2006


Nowhere wrote:

> I was able to upgrage everything I needed to individually. Is this normal
> for a non-version changing upgrade?

If by "is this normal" you mean "is it normal to change the protocol 
version", then the answer is yes and no. The purpose of the fixes branch 
is to have a branch that is more stable than the trunk. Kind developers 
backport fixes from the trunk to the fixes branch.

The goal is to make as few changes as possible in the fixes branch. 
However, at some point, one is bound to run into a bug that requires a 
protocol version change. At this point, the decision is between fix the 
bug or keep the protocol version the same. As with any bug fix on a 
stable branch, the developer decides whether a bug is severe enough to 
justify the potential impact of the fix. Sometimes, as in this case, the 
developer decided that it was worth it.

In some applications that communicate of a network, the major version 
number of the application changes whenever there are protocol changes 
that cause protocol incompatibility. However, while MythTV communicates 
over the network, it does not follow this sometimes followed convention. 
Isaac and other MythTV developers have stated that you must be running 
the same MythTV code on all computers in your MythTV system. If you do 
otherwise, then you are on your own. Since this is the rule for MythTV, 
there is no need to tie the application version number to the protocol 
version number.

More testing before a release can reduce the chance of this happening. 
However, more testing delays the release and creates more work for the 
developers. Since many people have integrated MythTV backends and 
frontends or run the same Linux distribution on their MythTV backends 
and frontends, keeping the versions in sync is usually not a problem. 
Therefore, with respect to the protocol version stability, a longer test 
cycle is not as necessary.

Personally, I delay upgrading my software (including Myth) until is has 
been out for a little while. No matter how much testing is done, once 
software is release there will be many more testers (I mean users). Once 
  the new set of testers has used the software for awhile and the first 
set of bug fixes is released, I figure the chances of me having a 
problem are with the upgrade is greatly reduced. This applies to both 
proprietary and open source software.


More information about the mythtv-users mailing list