State of X - Google Tech Talk

[Peter Hutterer]

Jojo,

(replying to both of your emails)

JoJo jojo wrote:
> Keith Packard was impressive, at the same time, it was quite light on X,
> perhaps they should forget that they are presenting to the audience sitting
> in front of them,
> and think about us, who watch it on the internet. Wish the next time he
> decides to speak,
> he goes in all the gory details of X innards.

Please think about what you are asking for here. I presume Keith and
Bart were invited by Google to give this talk. You suggest that they
should then ignore the audience of their host and concentrate about a
_possible_ audience that may watch the video on the web later.

As a speaker, ignoring the audience is not only rude, it is also
difficult. The audience will give you immediate feedback about your talk
(people dozing off is usually a hint). Ignoring the audience when being
_invited_ is rude and may hinder future public appearances.

> So the next time someone asks him to give an introductory talk, he can just
> provide them the youtube *.flv link.
> Which means that any talk he gives after that one HAS to be more technical
> than that, Good News !!!

I think we need more talks like this one. X still has an image problem,
created by years of stifling progress. Judging from some reviews I got,
the "X is something we used in the 80ies" is still ingrained in many
minds. Also, I heard many comments in the line of "X is a horrible pile
of old code, I don't want to touch this". (*)

A talk like that shows that X is not only up-to-date but interesting 
stuff is happening may change the public perception of X. It may not be 
the optimal talk for X hackers but to please everyone is difficult. The
audience was most likely not familiar with X internals. As Keith pointed
out in another email, "gory" X talks have been given at e.g. LCA and
were covered by LWN. See http://linux.conf.au/programme/ for videos of
these talks.

Our biggest problem at the moment is that the core X developers are
constantly overworked and there is little code review. A number of
patches have been sent to the ML lately and progress is being made but
it will take a while to get up to speed.

ATM, anything that may attract developers who are willing to post
patches and _review, test and sign-off_ patches is a Good Thing.

> The audience was completely irrelevant, they were complaining about device
> drivers not working etc.
> (and as soon as Q&A began, front seats were empty)

No. The live audience is _never_ irrelevant. If you give a talk to 20 
people and two complain about drivers, it does not mean that the other 
18 have the same view. In each group you have outspoken people and quiet 
ones. Judging the quiet ones is hard, but equally important.
Unfortunately, people have a tendency to assume that the vocal minority 
represents the general public. Which - quite frankly - is the reason why 
I wrote this email.

> Also why let off site googlers join in, just let them catch it on youtube,

Please consider again that Google was their host. If the host decides to
let other remote sites to virtually attend the talk, this is the host's
decision.

> Also the CAM job/cinematography reminded us all of tpb Cloverfield ;-) OK
> maybe not so much.

It is actually rather difficult to film a talk if you do not have the
required training and/or training. Especially with one camera you
constantly have to decide whether to focus on the speakers (important
for gestures and facial expressions) or get a shot of the slides as well.

Finally, the X.org website is a wiki. If you want to contribute to the
progress of X, create an account and add documentation. There is quite 
some information in these talks that can be summarised and put as
documentation online. So, even if you are not happy with the talk, you
can make the talk and the information therein more accessible to others
that may get more out of it.

Cheers,
    Peter

(*) well, it _is_ a horrible pile of old code, but that's no excuse :)