[CentOS] EL 6 rollout strategies? (Scientific Linux)

[Les Mikesell]

On 5/16/2011 2:52 PM, John R Pierce wrote:
> On 05/16/11 12:38 PM, Les Mikesell wrote:
>> I believe that by making
>> the process and its problems public, someone will help solve those
>> problems as they do in many, many other projects where the work is open.
>
> a very wise man[1] once said adding more bodies to a late project just
> makes it later.   9 women can't make a baby in 1 month.

If we were talking about a month, well probably no one would be talking...

> this is NOT a software development project, where pieces of it can be
> torn off and worked on independently, then later integrated.

And yet, there were 3 completely unrelated paths where only one had 
progress at a time.  Or so we've been told.  And that's not exactly a 
surprising situation even though they may not coincide frequently.

> this is a
> software build-and-integrate project where all the pieces have to be fit
> together and the bulk of the hard work is reverse engineering and
> recreating the correct build environment.

Yes, but whatever can't be automated here should benefit from doing the 
trial-and-error in parallel.   And the potential improvements might come 
in the automation process as much as the grunge work - you can't really 
predict how an open project will develop.

> [1]  Fredrick P. Brooks, The Mythical Man-Month, 1975.    36 years
> later, this book's fundamental premises are just as valid.

I wonder what he would have said about google's approach to, say, 
translation or voice recognition where the results depend on 
availability of massive amounts of input as much as anything else.

-- 
   Les Mikesell
    lesmikesell at gmail.com