[colug-432] February COLUG Meeting Announcement

[Scott McCarty]

I think many of us took forays into Gentoo at some point in our careers. Around 2003, I discovered it and quite honestly failed to get it working. I took another stab in 2005 and stuck with it for a few years for my laptop/desktop but never could justify Gentoo for a production server.

I think most of us understand that a "compile everything" OS is something best for a learning curriculum, which is basically the role it filled for many of my friends and I. Once I had ran Gentoo for a year or so, I completely "got it" and didn't really have the need to go further. Like you, I also went back to a compiled distribution. In fact, I tried OpenSUSE, Ubuntu, Debian, and Fedora at the time. Eventually, it just made business sense to use Fedora, as it was in-line with Red Hat, and what I found most employers wanted.

As a side note, I attended the Gentoo dev-room at FOSDEM in 2011 and their distribution is, indeed, alive and well. Though I wouldn't suggest it for production, I would suggest using it for a while to really learn Linux. The documentation strategy really is completely different than other distributions and I believe that is where it's magic lies. I will not repeat all of my reasoning here, but I did write a fairly in depth article why bootstrapping a new system through documentation works much better than installers for learning: http://crunchtools.com/bootstrapping-and-rooting-documentation-part-1/

As a final comment on code compilation, I would say I am most intrigued by the Java world, where I have the least experience. I hear about all of these build tools. I think it would be quite interesting to see a side by side comparison of:

C
Java
Python
Ruby

Along with all of the build tools that are associated and their equivalents. I am quite familiar with C build tools as I think most Unix guys are (probably many on this list), but I am much less knowledgeable when it comes to the Java/Python/Ruby worlds, yet there is an endless sea of blog posts about RVM, VirtualEnv, Maven, Jenkins, etc. This level of "compiling open source", I believe, is extremely applicable to the modern professional Linux person and I would love to see a talk expanded into this are of compilation/building free software.

Best Regards
Scott M

----- Original Message -----
From: "William Yang" <wyang at gcfn.net>
To: colug-432 at colug.net
Sent: Tuesday, February 21, 2012 7:55:28 AM
Subject: Re: [colug-432] February COLUG Meeting Announcement

On 02/20/2012 10:35 AM, Rick Troth wrote:
> Having encountered widely varying opinions about source build, what do
> YOU say about it?  Why do we use FOSS?  If we *use* FOSS, does that
> mean (to you) that we should actually *build* from source?  Why and
> when should we and should we not [re]build from sources?
> 
> I'm priming the pump for this pres.  Will be showing my own how-to,
> which to me seems like do-it-in-your-sleep easy.  But part of the talk
> will include the rationale for going this route.  Lemme know what
> questions or points come to mind.

Alas, I will not be at the meeting.  But I'd like to tell a short story
about compiled code.

I used to use Gentoo.  When I selected Gentoo (over RedHat), it was because
the platform was built from source, leading to an optimized, efficient
platform.  Or so I thought.

After 4 years, I transitioned pretty much all of my systems from Gentoo to
Ubuntu.  My move away was not caused by the breakdown at Gentoo:  I'd
discovered that the act of patching a system had become onerous, expensive,
time-consuming and inefficient.  Long compile times were the #1 reason why
I felt I needed to move to something else: I was tied down by the
requirement to compile.

I still use source to build custom appliance operating system images.  But
for day-to-day use, including reasonably standard servers and workstations,
I think it's a poor choice to base operations on source-compilation.  That
approach, which took hold in the 80's and before in *IX environments, isn't
an approach to modern computing.  It makes your environment optimized and
efficient on your existing platform, at the expensive of losing nimbleness
and the ability to move and change, as legacy code holds you down.

Source is best used for customized applications, areas where superfine
tuning is required or behavior needs to be different from in the box.
However, there's been a substantial move toward "off the shelf" software
for more than a decade in many environments... and the reason why is
because it's cheaper, faster, and more efficient.

(Free) Open source is good.  It gives you the opportunity to understand
what your system does.  It gives you the ability to control the behavior of
your system when you need to.  But it's not mandatory to build from source.
 There's going to be a balance based on interest, time, willingness to futz
with internals, cost, and other factors.  I do think people who work with
systems regularly need to understand how to compile software, and it's part
of my program when I train others in systems operations and administration.
 But as a business matter, compilation is typically a cost center.

	-Bill
-- 
William Yang
wyang at gcfn.net
_______________________________________________
colug-432 mailing list
colug-432 at colug.net
http://lists.colug.net/mailman/listinfo/colug-432