I've been thinking about Myth's stability.
Generally the releases are pretty good (!) :)
However, occasionally things happen (0.15.1, the nfs file close problem in 0.16, the ww.weather.com issue)
ie right now I'd like to see 0.17.1 include a fix for the weather problem rather than wait until 0.18 (or hacking hosts/xml)
(Well, truth be told, I'm still on 0.16 so I'd have have liked an nfs-close patch release...)
The problem with this is that the fix for weather is mixed in with a load of other CVS changes so it's hard to create a 0.17.1
I was wondering whether a leaf could be taken out of the linux kernel's recent 'stable' approach?
Essentially gather a set of patches together that all the maintainers can use to release 'stable' upgrades?
All distro packagers could submit/veto patches... that kind of thing.
These patches would offer no new features but would fix simple bugs. In general they'd have been applied to CVS.
Maybe not all bugs - some would be too intrusive, the fix may be CVS only etc etc.
The objective being that an 'upgrade' to one of these point/patch releases would be a very safe, 5 minute job with (almost!) zero risk of screwing up. Any patch that doesn't meet these criteria could be vetoed.
Users could then safely be told to upgrade to a point/patch release whenever a new one was released (hey, by version 0.20, maybe that could be (semi-)automatic a la "MS automatic updates")
I was also wondering whether this should be agreed amongst the distro maintainers (it'd be nice if all of them using the same subversion were based on at least the same source code)
Also, I'd stress that (unless they want it to be) this would not be an official developer supported thing - more a package maintainer applying a bug for his customers - essentially the same as Debian does in applying patches to the package and then forwarding them upstream.
Maybe it would be worth considering a myth-packagers mail list to promote some consistency amongst the packages?
David
PS Although I used 0.17.1 as an example, I think that Isaac 'owns' that numbering format - I'd suggest a suffix -p1, -p2 etc
But I'd like all the distros to mean the same thing.
_______________________________________________
mythtv-users mailing list
[email protected]
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
