[Qemu-devel] release date of qemu 0.9.2?

[Qemu-devel] release date of qemu 0.9.2?

[C.W. Betts]

-----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?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (Darwin)
Comment: Macs Rule\!

iEYEARECAAYFAkj9WHAACgkQIXkdd5hN+dt1FACfQoHQDQbRCMSCD0gzR7ODpywl
iucAnjjSdBhMbwvzCuG536U7ikRlV6QJ
=+urm
-----END PGP SIGNATURE-----

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

[Laurent Desnogues]

On Tue, Oct 21, 2008 at 6:19 AM, C. W. Betts <computers57@hotmail.com> wrote:
>
> 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.

You probably meant gcc 3 which is a requirement for dyngen targets.

> Also, what targets still depend on dyngen?  And is anyone working on porting
> them to TCG?

Only PowerPC targets still depend on dyngen, and it's being worked on.


Laurent

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

[Aurelien Jarno]

C.W. Betts a écrit :
> -----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?

Only PowerPC depends on dyngen. I am currently working on converting it
to TCG as my free time allow it, and a few other persons are also
sending patches. I expect that the conversion will be finished in about
one month (this target has a high number of existing dyngen-ops due to
the number of supported CPU with specific extensions).

Unfortunately that also means I will spend less time on merging patches
from the mailing-list.

-- 
  .''`.  Aurelien Jarno	            | GPG: 1024D/F1BCDB73
 : :' :  Debian developer           | Electrical Engineer
 `. `'   aurel32@debian.org         | aurelien@aurel32.net
   `-    people.debian.org/~aurel32 | www.aurel32.net

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

[Anthony Liguori]

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.

Regards,

Anthony Liguori

> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.8 (Darwin)
> Comment: Macs Rule\!
>
> iEYEARECAAYFAkj9WHAACgkQIXkdd5hN+dt1FACfQoHQDQbRCMSCD0gzR7ODpywl
> iucAnjjSdBhMbwvzCuG536U7ikRlV6QJ
> =+urm
> -----END PGP SIGNATURE-----
>
>
>
>

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

[Thiemo Seufer]

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.


Thiemo

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

[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.

Regards,

Anthony Liguori

> Thiemo
>   

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

[Glauber Costa]

On Tue, Oct 21, 2008 at 12:51 PM, Anthony Liguori <anthony@codemonkey.ws> wrote:
> 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.

I believe there's a lot of users who grab the stable releases, but
won't grab the development snapshots.
So usually releasing official releases more often means more testing happening.

I believe it's worthy.

-- 
Glauber  Costa.
"Free as in Freedom"
http://glommer.net

"The less confident you are, the more serious you have to act."

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

[Jean-Christophe PLAGNIOL-VILLARD]

On 09:51 Tue 21 Oct     , Anthony Liguori wrote:
> 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.
I agree with you I think that the kernel way of release is good and can be
used for qemu

Best Regards,
J.

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

[Glauber Costa]

On Tue, Oct 21, 2008 at 12:57 PM, Jean-Christophe PLAGNIOL-VILLARD
<plagnioj@jcrosoft.com> wrote:
> On 09:51 Tue 21 Oct     , Anthony Liguori wrote:
>> 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.
> I agree with you I think that the kernel way of release is good and can be
> used for qemu

The kernel has too many -rcX , which I believe does not suit qemu well.
This is because of the size of the user and developer base. The merge window is
also something that we would have a hard time doing without a distribuited scm.

So, kernel + adaptations is fine.


-- 
Glauber  Costa.
"Free as in Freedom"
http://glommer.net

"The less confident you are, the more serious you have to act."

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

[Paul Brook]

> >> 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.

I agree with Thiemo. Removing dyngen is more important than new TCG host 
support.

> 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.

Cutting releases is the easy. Making sure the tree is in a suitable state to 
be released is the hard bit.  Releases without good quality control are IMHO 
pointless.

Given the speed of qemu development, I think 3 months is way too short for a 
release cycle.  You may as well just have everyone pull from SVN.

Paul

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

[Anthony Liguori]

Paul Brook wrote:
>> 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.
>>     
>
> Cutting releases is the easy. Making sure the tree is in a suitable state to 
> be released is the hard bit.  Releases without good quality control are IMHO 
> pointless.
>   

Yeah, what I meant by leg work was doing deep testing and nagging people 
to fix things up before a release.  Cutting a release is very simple.

> Given the speed of qemu development, I think 3 months is way too short for a 
> release cycle.  You may as well just have everyone pull from SVN.
>   

How do you feel about 6 months?

Regards,

Anthony Liguori

> Paul
>   

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

[Paul Brook]

> >> 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.
> >
> > Cutting releases is the easy. Making sure the tree is in a suitable state
> > to be released is the hard bit.  Releases without good quality control
> > are IMHO pointless.
>
> Yeah, what I meant by leg work was doing deep testing and nagging people
> to fix things up before a release.  Cutting a release is very simple.

Ok.

> > Given the speed of qemu development, I think 3 months is way too short
> > for a release cycle.  You may as well just have everyone pull from SVN.
>
> How do you feel about 6 months?

I think switching to a hard time driven release schedule would be a mistake.
In particular I think we should complete the TCG conversion before the next 
release.

That said, 6 months between releases may be a reasonable starting point.

Paul

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

[Anthony Liguori]

Paul Brook wrote:
> I think switching to a hard time driven release schedule would be a mistake.
> In particular I think we should complete the TCG conversion before the next 
> release.
>   

Yeah, I agree with you completely.

> That said, 6 months between releases may be a reasonable starting point.
>   

Yes, I was thinking that to start with aiming for 6 month release 
intervals is the right thing to do.

Regards,

Anthony Liguori

> Paul
>   

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

[Thiemo Seufer]

Anthony Liguori wrote:
> Paul Brook wrote:
>> I think switching to a hard time driven release schedule would be a mistake.
>> In particular I think we should complete the TCG conversion before the 
>> next release.
>>   
>
> Yeah, I agree with you completely.
>
>> That said, 6 months between releases may be a reasonable starting point.
>
> Yes, I was thinking that to start with aiming for 6 month release  
> intervals is the right thing to do.

Agreed, 6 months sounds like it will catch enough worthwile features
for a new release.


Thiemo

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

[Glauber Costa]

On Tue, Oct 21, 2008 at 2:49 PM, Thiemo Seufer <ths@networkno.de> wrote:
> Anthony Liguori wrote:
>> Paul Brook wrote:
>>> I think switching to a hard time driven release schedule would be a mistake.
>>> In particular I think we should complete the TCG conversion before the
>>> next release.
>>>
>>
>> Yeah, I agree with you completely.
>>
>>> That said, 6 months between releases may be a reasonable starting point.
>>
>> Yes, I was thinking that to start with aiming for 6 month release
>> intervals is the right thing to do.
>
> Agreed, 6 months sounds like it will catch enough worthwile features
> for a new release.
>
/me agrees too

> Thiemo
>
>
>



-- 
Glauber  Costa.
"Free as in Freedom"
http://glommer.net

"The less confident you are, the more serious you have to act."

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

[Armin]

Anthony Liguori wrote:
> Paul Brook wrote:
>>> 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.
>>>     
>>
>> Cutting releases is the easy. Making sure the tree is in a suitable 
>> state to be released is the hard bit.  Releases without good quality 
>> control are IMHO pointless.
>>   
> 
> Yeah, what I meant by leg work was doing deep testing and nagging people 
> to fix things up before a release.  Cutting a release is very simple.
> 
>> Given the speed of qemu development, I think 3 months is way too short 
>> for a release cycle.  You may as well just have everyone pull from SVN.
>>   
> 
> How do you feel about 6 months?

Shouldn't it be feature driven rather then time?

- Armin
> 
> Regards,
> 
> Anthony Liguori
> 
>> Paul
>>   
> 
> 
> 
> 
> 

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