Re: [Qemu-devel] release date of qemu 0.9.2?

[Andreas Färber]

Am 21.10.2008 um 16:51 schrieb Anthony Liguori:

> Thiemo Seufer wrote:
> > Anthony Liguori wrote:
> >
> > > C.W. Betts wrote:
> > >
> > > > -----BEGIN PGP SIGNED MESSAGE-----
> > > > Hash: SHA1
> > > >
> > > > When will qemu be given the status of 0.9.2?  I was browsing the   
> > > > source code of the Qemu that ships with OpenSuSE and noticed  
> > > > that  there are a lot of patches.  It would probably be helpful  
> > > > if there was  a release that didn't depend on GCC 4.
> > > >
> > > > Also, what targets still depend on dyngen?  And is anyone working  
> > > > on  porting them to TCG?
> > > A lot of previously supported hosts are no longer supported with  
> > > TCG.  I  would think adding those hosts would be more important  
> > > than completing  the TCG conversion before cutting a new release.
> >
> > I disagree. The most important hosts are supported, the other hosts
> > were largely experimental anyway, and getting rid of dyngen / gcc3
> > is IMHO worth a release.
>
> Personally, I'd like to see much more frequent releases (every 3-6  
> months).  I'm okay with not having complete features (like the  
> dyngen->TCG conversion) or host regressions because I think there's  
> value in having releases regularly compared to feature based releases.
>
> What do other people think?  I'd be willing to do the leg work of  
> releases.

Personally I'd appreciate a decision process involving the community,  
e.g. a last call for patches period before doing the release.

Something to consider is that distros would probably prefer maintained  
releases over frequent releases. Now that we've switched to SVN, it  
would be possible to branch for a release (not just tag it) and to  
apply security/build fixes to the branch so that distros/users can  
pull from here rather than from an external location like Launchpad.  
In that case, branching more often could mean more legwork maintaining  
branches.

Regards,

Andreas