Re: [Qemu-devel] CFP: 1st International QEMU Users Forum

[Alexander Graf]

Wolfgang,

On 25.11.2010, at 11:30, wolfgang mueller wrote:

Dear Alexander Graf,
thank you for contacting me.
I hope that I am right when I understand your below email as a complaint that we have not contacted you before organizing
this event.

The complaint is mostly that you have not approached the community, yes :).

First of all, we understand our event not as an official QEMU event nor have we stated this anywhere or that that we are the
owners of QEMU.

Let me explain about the background of the Workshop.
We as QEMU users, occasionally meet other QEMU users at conferences and we observed that the number was significantly growing
during the last years (thanks to the excellent work of the developers). We googled for any contacts to other QEMU users and
did neither found any workshop nor any mailing list or anything similar we could contact. The only thing we found as a main contact
was Fabrice  Bellard and a huge set of developers (~3000). We saw Fabrice Bellard as the "owner" and contacted him. Unfortunately
he rejected as he has no longer interest in QEMU.

I guess we should really remove most traces of Fabrice if he isn't even helpful enough to direct you to the mailing list. The maintainer situation should be explained in the MAINTAINERS file in the qemu source tree. Apparently the last attempt to redo that file and make it contain actual current information failed. Sigh.

There is a group of people with commit rights to the source tree. Each of them have special and in general different fields of interest. There are also several submaintainers who keep track of subsystems, but don't have commit rights and route everything through the few people with commit rights.

In general, it's similar to the Linux kernel, but with a schizophrenic Linus.

So the "right thing" to do would have been to send a mail to the qemu-devel mailing list, asking for feedback. The right people definitely do read the list.

As we did not find any current "leader" or dedicated leading group beyond the web site, we decided to move on since we were interested
to get the QEMU users together to join efforts and exchange ideas and developments (The alternative would have been to contact
all over 3000 QEMU developers and to discuss with them the organization of such a event. However, from the first idea to the
possibility to get it accepted in the framework of the DATE conference were very few weeks so that we did not see it as a viable
alternative).

I now understand that we obviously missed your role in that community as the leader or one of the leaders and address@hidden
as the community or part of the community. Therefore, we apologize that we did not contact you or address@hidden.

Don't worry about me. I feel myself in general rather representing minorities in qemu. The closest to a "leader" position in qemu these days is probably Anthony Liguori. CC'ing him.

However, currently though already very late, we could offer you that you can join and and we can list you as a third organizer.
But you have to make an immediate decision as the DATE program will go to print in about 10 days. Otherwise you will be only
listed on the web.

Additionally, we are still looking for a person who is giving a general 45-minute QEMU tutorial/overview. As we were enforce by the
hosting DATE conference to put a name there, please just take it as a placeholder for the worst case that we do not find anybody else.
Are you interested or anybody from address@hidden?

This is exactly the reason I approached you sceptic on the whole thing. Qemu is _very_ diverse. It ranges from prototyping non-existing hardware over user-space only emulation to virtualization. Each of the different aspects have their very own issues and traction. The group that's the most active currently is the virtualization crowd. Which target audience are you going for?

Finally, I actually do not share your concerns that people register to get more information "about command line switches to virtualize
on x86 hardware". I really would expect that people read the programme before they register. However, maybe you know the community
better than I do.

After checking on the pricing for previous years I agree with you :).

For the future, we are currently not planning a second Users' Forum and we are open to forward the organization of a second workshop
to anybody else.

Why don't you simply take this event as a first trigger to organize an QEMU conference for Developers and users.
We can help you by forwarding the email addresses of some of the people. I can also help you if you want.
This can be the start of a real great effort with huge benefit to the community.

As stated above, things are more difficult than that, but I agree. We do have annual KVM Forum meetings where we gather all the virtualization people from Qemu as well and talk about things there. The slides and recordings are publicly available, if you're interested: http://www.linux-kvm.org/page/KVM_Forum_2010. In general, people surrounding KVM meet quite a bit.

For the emulation side, things look different. I'm not aware of any traction on the emulation side of Qemu. Getting people together for that one is certainly lacking. I'm not sure I'm the right person to talk to there. If nobody else steps up, I could barely play the role of someone who knows what he's talking about, but I'm myself more of a virtualization person too.

So it all boils down to the audience you're expecting. From the announcement and colocation with DATE, I assume you're looking mostly for SystemC integration? That probably plays well into the emulation piece. I don't think that doing yet another 'normal' virtualization focused event would be beneficial - we get plenty of gatherings around that already.

Alex