Re: [gentoo-project] Some focus for Gentoo

[Michael Orlitzky]

On 01/13/2015 10:43 PM, Donnie Berkholz wrote:
> Wanted to share my thoughts on where I think Gentoo should go, in terms
> of direction. Would love to hear your thoughts.
>
> http://dberkholz.com/2015/01/13/gentoo-needs-focus-to-stay-relevant/
>

My perspective as a system administrator:

The package manager solves a lot of problems that I didn't even know I
had. When we deploy anything (a new website for example), I create a
package for it. Usually for a website the package is empty except for an
RDEPEND on php[foo] and smarty or something like that, but doing this
gives me an audit trail for every piece of software installed on every
system. It also prevents someone from uninstalling smarty on a web
server that needs it.

In theory you could do this with any distribution, but the fact that
Gentoo is source-based makes it super easy to do. I just add a few lines
of text to an overlay, and boom, new package.

It also works for more complicated software. Each programming language
has its own build system these days, and you'll hear developers
complaining about e.g. "rubygems hell" or "cabal hell" when their build
system can't resolve dependency conflicts. That's a stupid problem to
have, because it's already been solved by real package managers that
don't respect some imaginary programming language boundary.

But the reason most people avoid using a real package manager is because
all of the packages are outdated and a pain to update. This isn't true
with Gentoo. If I need a Haskell package, I run `hackport merge foo` and
99% of the time it creates a perfect ebuild. Portage makes sure the
dependencies are sane across my entire system (even if a non-Haskell
dependency is involved), and hell is avoided for one more day.

Every script, utility, and report has an ebuild that could easily be
re-emerged on another machine. By making packages easy, Gentoo lets me
keep track of everything that's happening on the systems I'm responsible
for.