[lug] The deal with applets / notifications / panels / indicators / systray.....

[Rob Nagler]

On Sun, Feb 5, 2012 at 6:17 PM, Davide Del Vento wrote:
> everywhere worldwide. So there is a inherent latency (by design, I
> believe they call it "eventual coherence") in the way those servers
> synchronize.

Actually, it's a bug caused by the fact we still don't know how to
write application frameworks for distributed systems.  :)

Google does not have a coherent framework for cache consistency.  Try
this experiment:  Bring up two different browsers (on different
devices and maybe different continents, I haven't tried that :) with
gmail showing your inbox, email yourself from another account, e.g.
yahoo.  Hit the gmail refresh button or just click on Inbox or just
wait a few seconds if you sent from yahoo, hotmail, etc.

Try this other experiment (even in a single browser): Open two tabs
with gmail, star one of the messages in one tab, in the other either
wait a really long time or click on the gmail refresh button.  Nothing
will happen.  Hit the browsers refresh button, and the star will show
up.

> If you think this is a problem, you might even design a system that
> would solve this problem, at the expense of something else (for

I would not want to go head-to-head with the people at google when it
comes to distributed systems.  They are rockstars when it comes to
understanding stuff like distributed cache consistency. I'm
continually astounded at the low latency on virtually any search or
mapping problem I give google.  I would, however, humbly submit that
they could use a little more expertise on their application frameworks
and software methodologies. :)

> Now, why this should matter for proteus/slipstream, I don't know :-)

Because frameworks need to evolve, not be designed top-down.  Google
no doubt has some people they call architects who say how things are
to be done from 5 miles up, and they fail to notice the little
details, e.g., cache consistency needs to be in the collective
consciousness of all developers, even the guy who implements the
"star" function in gmail.

How this relates to proteus is that until you have a real problem to
solve for a real customer, you won't know how to create the right
framework, and even with the R&D budget of google, you will have a
hard time coming up with a framework which solves all problems
associated with conceptualizing all of software, which I believe,
proteus is trying to do.

The difference between wikipedia and proteus is that wikipedia's
mechanism was assimilated over time from a very simple problem
(c2.com/wiki, technically) and was used by real people over the
subsequent 18 years to where we are today.  The other solutions
available were comprehensive, consistent, and too complicated.  It was
much easier to write something like s{(\b[A-Z][a-z]+[A-Z][a-z]+\b)}{<a
href=$1>$1</a>}sg to link pages than understanding DTDs and SGML. :)
>From there it grew into the world's largest, coherent, digitized
repository, and of course, MediaWiki is the Perl of content creation
languages. :)

That being said, what do I know?  Maybe Proteus will take over the
world, and I'll be just one of the people who said, "You can't do it."
:)

Rob