Re: [gentoo-dev] Making the developer community more open

[Brandon Edens]

On Mon, Mar 20, 2006 at 11:07:37PM +0000, Daniel Drake wrote:
> I'm looking for ideas - preferably big, drastic, shiny ones. Ignore any
> issues relating to migration away from our current system. What would be
> the _ideal_ way for Gentoo to handle contributions from anyone? (note
> that I'm dropping the user/developer community separation in that
> question, as the boundary between those could change in these ideas)
> How would an ideal recruitment process work, if there would be one at all?


When I was a system administrator working with Gentoo I would've appreciated a
way to interact with the other Gentoo system administrators. (ie gentoo-server).
I would've liked it even more if I could've communicated with the Gentoo
University Department System Administrators. When I say communicated I mean
interacted with.

Now that I'm an AMD64 laptop Gentoo user I would like a concise way of
communicating back to my community the AMD64 users and specifically the laptop
users. In fact I'd like to know what other people are using the Compaq Presario
V2000Z AMD Turion64. I'd also like to know what software they're running on
their laptops and if they consider it stable. What kernel configurations they're
using. What functionality is broken and what needs to be fixed.

I'd like an easy way to communicate with them, pass them notes about problems
with packages. If I trust them a-lot then I'd like to use their binary built
packages as well as allow them to use the ones built by me.  I guess we'd create
a sort of p2p mini-pocket of gentoo users with our relationship built upon
trust.

I imagined Gentoo on Win32, something like Cygwin. I maintained a computer lab
filled with Windows machines. I'd like to install Gentoo Win32 on one machine.
Install it on the next machine then tell that machine about the first. I'd like
to install it on a third machine then tell it about the first or the second but
have that third machine then know about both. I'd like to compile a piece of
software on one of the machines then know that any of the other two will
automatically get the binary version of the package from the one that compiled
it because they trust that machine. The damage to unnamed others from the
existence of a system like this would be quite excessive IMO.

So I've structured this email as a want-list. However, I'm not oblivious to how
this is implemented. I suppose the idea is to restructure Gentoo into a tiered
community (as mentioned by other posters). Make it easy for tiers to birth and
die (we might like a rhode island gentoo users group for instance, might not
last for more than a year.) Maybe one tier is just me and my friends sharing our
hacks, customizations, letting each other know about some exciting package,
etc... I need my portage/emerge to act as a meta system that pulls from the
various communities based upon how much I trust those communities.

How do we get there from here? I suppose just start adding functionality to
portage to support this. One part could be just expanding upon the portage
overlay. It might be nice if portage became better defined so that we could
implement it in a variety of programming languages (I'm into LISP programming
for fun).
Portage as a daemon?

Another concrete feature would be to allow by convention a specific directory
that could be created and used for applying patches during the build process
(without modifying ebuilds). A place where portage will automatically apply that
patch during the build of some piece of software. So lets say this feature of
portage existed, perhaps I could further put the patch in various community
portage overlays and others in that community could learn about that patch. I
suppose its those sorts of massive optimizations/conventions/intelligence that I
always appreciated about Gentoo and its packaging system.

I appreciate that Gentoo is more hacker friendly and less Debian ivory tower.  I
think some of that is source-built distro versus a binary distro.

Some of this is the gentoo-stats project that died. It'd be nice to know about
the other people in my version of the Gentoo community. If I'd had statistics
that MIT was using Gentoo on 5000 machines with each having an average up-time
of 100 days (information gleamed from portage's community functions); then I
could've marketed Gentoo better to my bosses.

I'm rambling.
Enjoy,
Brandon Edens