[Pardus-devel] Conclusions to "Rethinking the Linux Distribution"

George Belotsky gb at openlight.com
Mon Jul 16 10:46:55 EEST 2007 

Thanks a lot for your interesting suggestions, Amine; these would be
great for a HOWTO site.

I was thinking, however, about reporting errors -- just like the ASRS
users do.

For example, you may have several terminal windows on your desktop,
with each window connected to a different remote machine via ssh.  In
this scenario, it is easy to accidentally give a command to the wrong
machine.  After all, the terminal windows look the same.

Here is a *fictional* narrative, showing how someone may report such
an error.

  "Working late at night.  Before running some tests, shut down the
   Apache on what I thought was the development machine.  After
   several minutes, realized that it was one of the production
   servers instead.  Restarted Apache, and the machine rejoined
   the server farm.  Fatigue was a factor, since I normally open 
   terminal windows to production Web servers only on a specially
   reserved virtual desktop.  This time, however, I neglected to do  
   so."

Note that the above situation is not due to any bug.  The terminal
emulator, window manger, ssh, etc. are all working perfectly.  Yet,
the error may occur anyway -- because the human factor is also part
of the system.

A similar situation occurred on an aircraft type during World War II.
Mysteriously, pilots would occasionally crash these airplanes during
landing.  It turned out that, because the gear and flap levers were
close to each other, it was quite possible to activate the wrong one.
So, improper interaction of a sound airplane and a trained pilot could
still lead to accidents.

The solution was to place a small wheel shape on the gear lever, and a
wedge shape on the flaps lever.  After this, the crashes stopped.
Much later, the ASRS came into being as general tool, to catch these
types of problems (as well as many others) on an ongoing basis.

With such a facility for Linux distributions, we should find areas
where great improvements can be made -- improvements that may not be
obvious from any other source.

Wish you all the best,

George.


On 10 Tue 07, Amine Chadly wrote:
> I think combining the idea of an ASRS with flexibility of tags seems
> like a big winner to me...
> If you want to have a general type of site, I guess that the content
> could be about solving particular problems in different flavors of
> linux distros.
> If it would focus on a particular distribution, then it could be about
> general problems... Were you asking for examples ?
>  - Accelerate web browsing by setting up a proxy server.
>  - Reducing the dns traffic by hosting you own inside your network.
>  - Relaying the mail for your own domain.
>  - Setting up a svn server for your local needs (or plain text backups !)
>  - Setting a vpn for accessing your local network from the outside.
> 
> and other little tips and tricks explaining caveats faced and the ways
> around them...
> 
> What did you have in mind ? ^^
> 
> On 7/8/07, George Belotsky <gb at openlight.com> wrote:
> > On 08 Sun 07, Amine Chadly wrote:
> > > I was a little confused about the aviation safter reporting system
> > > part of your email, but after having a look at it
> > > [http://asrs.arc.nasa.gov/callback_nf.htm] and at different issues, I
> > > found it very interesting !
> >
> > Great!  You can also take a look at some ASRS report sets.
> >
> >     http://asrs.arc.nasa.gov/report_sets_nf.htm
> >
> > Note that the data is formatted as name-value pairs.  The names look a
> > lot like structured tags.  Here is a fragment.
> >
> >        Person : 1
> >          Function.Flight Crew : Captain
> >          Function.Oversight : PIC
> >
> >        Person : 2
> >          Function.Flight Crew : First Officer
> >
> > As I mentioned previously, an interesting approach would be to combine
> > the idea of an ASRS report with social networking tags.  Anyone
> > submitting the report would tag it, so the classification system would
> > evolve organically.  Names used inside reports could also work like
> > tags, as stated above.
> >
> > > Having short stories telling problems faced and the solutions is an
> > > excellent concept. I would see as well a way to access to stories
> > > according to topics (security and system admin comes in mind).
> >
> > The ASRS sample reports include the narrative sections, so you can
> > read the stories and get an idea of how this works.  In the proposed
> > system, tags would allow you to find reports -- similar to the way
> > you use Flickr or del.icio.us.
> >
> > If you have any suggestions, I would appreciate them.  What would you
> > like to see in such a system?
> >
> > Wish you all the best,
> >
> > George.
> >
> > >
> > >
> > > On 7/6/07, George Belotsky <gb at openlight.com> wrote:
> > > > Hello, Everyone:
> > > >
> > > > Almost two months have passed since the publication of "Rethinking the
> > > > Linux Distribution".
> > > >
> > > > I have summarized the debate thus far, and posted the analysis to
> > > > O'Reilly's forum.  I included a thank-you note to the Pardus team, for
> > > > your great work.
> > > >
> > > > http://www.oreillynet.com/cs/user/view/cs_msg/91621
> > > >
> > > > The article mentioned the Aviation Safety Reporting System (ASRS) as a
> > > > source of ideas for governance of online collaboration.  In the
> > > > analysis, I expand on this concept.  A reporting system modelled on
> > > > the ASRS would help us make better use of the "Wisdom of Crowds", to
> > > > improve the Linux distribution.  See the link above for more details.
> > > >
> > > > Feel free to comment or ask questions.  Do you think such a system
> > > > would be useful in general?  Would you use it?
> > > >
> > > > Wish you all the best,
> > > >
> > > > George.
> > > > _______________________________________________
> > > > Pardus-devel mailing list
> > > > Pardus-devel at pardus.org.tr
> > > > http://liste.uludag.org.tr/mailman/listinfo/pardus-devel
> > > >
> > >
> > >
> > > --
> > >   Amine Chadly
> > > _______________________________________________
> > > Pardus-devel mailing list
> > > Pardus-devel at pardus.org.tr
> > > http://liste.uludag.org.tr/mailman/listinfo/pardus-devel
> > _______________________________________________
> > Pardus-devel mailing list
> > Pardus-devel at pardus.org.tr
> > http://liste.uludag.org.tr/mailman/listinfo/pardus-devel
> >
> 
> 
> -- 
>   Amine Chadly
> _______________________________________________
> Pardus-devel mailing list
> Pardus-devel at pardus.org.tr
> http://liste.uludag.org.tr/mailman/listinfo/pardus-devel

More information about the Pardus-devel mailing list