[gentoo-project] Some focus for Gentoo

[gentoo-project] Some focus for Gentoo

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 357 bytes --]

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/

-- 
Thanks,
Donnie

Donnie Berkholz
Council Member / Sr. Developer, Gentoo Linux <http://dberkholz.com>
Analyst, RedMonk <http://redmonk.com/dberkholz/>

[-- Attachment #2: Type: application/pgp-signature, Size: 966 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Jeroen Roovers]

On Tue, 13 Jan 2015 21:43:23 -0600
Donnie Berkholz <dberkholz@gentoo.org> wrote:

> http://dberkholz.com/2015/01/13/gentoo-needs-focus-to-stay-relevant/

I generally agree.


[ = People developing software = ]

It reads like we should better *advertise* Gentoo as a good distro to
choose for the purpose of building a development environment, custom
distro or educational environment. Aren't we doing that properly now?
Aren't we heading in that direction already?

If we want to attract developers, we might need to point out how they
can (easily) write ebuilds to make deployment easier of things we don't
already offer ebuilds for. This would have to chime with their ideas of
what a deployment should look like.


[ = People who need extreme flexibility (embedded, etc.) = ]

Development environments for embedded systems generally focus on a
"stable" OS where they can plant down flashy (sorry) IDEs so that users
can focus on the applications they want to write for those targets. How
do you convince them to instead build an embedded system from the
ground up using Gentoo Linux?


[ = People who want to learn how Linux works = ]

The last category is a bit problematic. People who flock to Gentoo with
little Linux experience are generally after a specific feature, like
the 3D desktop (oh, that was 10 years ago), a replacement for a
favourite desktop environment gone insane (this happens all the time)
or access to the latest version of some newfangled application they've
read about and want to try out (like cryptocoins or geolocation or
containers to name a random few).

Gentoo already naturally gravitates toward offering those, since writing
and distributing ebuilds for brand new software is easy and since Gentoo
developers equally tend to volunteer their time to such novelties. More
generally interest in a certain category of applications coincidentally
means better quality in how we deliver those in ebuilds.

This category of users generally doesn't start using Gentoo for the
learning experience - they might appreciate that only after they've
got comfortable with it. People who want that tend to go with LFS.
Attracting people who want to "learn Linux" is therefore impossible, and
having those that we do attract "learn Linux" instead of "learn yum"
is a nice side-effect.

For half a year now I have been trying to find a Linux related job, and
actual Linux sysadmins/application developers generally don't know what
Gentoo Linux is (i.e. they've never heard of it) or they assume the
anti-ricer pose and the "we use an Enterprise distro" defence. All my
(nearing) 10 years of Gentoo Linux development mean nothing to them,
and nothing seems to have changed there in 10 years. Even if it cures
your curiosity about Linux, it doesn't offer much "office cred" in
return.

People who want or need to use software that their distro doesn't yet
offer in a straightforward manner tend to find HOWTOs and thereby simply
follow the script (copying and pasting commands apparently teaches you
how stuff works?). When that succeeds, they don't need to think about
learning Linux or find another distro.

I generally agree that attracting young/inexperienced users is a bonus,
but it's extremely hard to "focus" on.


     jer

Re: [gentoo-project] Some focus for Gentoo

[Jeroen Roovers]

On Wed, 14 Jan 2015 08:52:44 +0100
Jeroen Roovers <jer@gentoo.org> wrote:

> If we want to attract developers, we might need to point out how they
> can (easily) write ebuilds to make deployment easier of things we
> don't already offer ebuilds for. This would have to chime with their
> ideas of what a deployment should look like.

On that note, I guess one way of doing this is to advertise Gentoo
Prefix better and integrate it better. Get the QA warnings in, make all
developers aware of the common pitfalls, and generally move toward
supporting Gentoo on any platform out there that can reasonably
accommodate it.

Why on Earth would anyone need to install a new platform to start
developing (on) another platform? Almost ten years ago, Gentoo on
Windows was a joke[1]. A few years later it suddenly wasn't.


     jer


[1] http://gentooexperimental.org/nt/gentoo-nt-quickinstall.html
    https://www.gentoo.org/news/en/gwn/20050404-newsletter.xml

Re: Re: [gentoo-project] Some focus for Gentoo

[Andreas K. Huettel]

Am Mittwoch 14 Januar 2015, 08:52:44 schrieb Jeroen Roovers:
> On Tue, 13 Jan 2015 21:43:23 -0600
> 
> Donnie Berkholz <dberkholz@gentoo.org> wrote:
> > http://dberkholz.com/2015/01/13/gentoo-needs-focus-to-stay-relevant/
> 
> I generally agree.
> 
> 
> [ = People developing software = ]
> 
> It reads like we should better *advertise* Gentoo as a good distro to
> choose for the purpose of building a development environment, custom
> distro or educational environment. Aren't we doing that properly now?
> Aren't we heading in that direction already?

++

Advertising, public image, etc is extremely important.

(A stylish, modern-looking website too...)

Side note, why do I have to go to the Funtoo website to see something as 
beautiful as this?
http://www.funtoo.org/Gentoo_Ecosystem


-- 
Andreas K. Huettel
Gentoo Linux developer
kde, council

Re: [gentoo-project] Some focus for Gentoo

[Markos Chandras]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On 01/14/2015 11:27 AM, Andreas K. Huettel wrote:
> 
> (A stylish, modern-looking website too...)
> 
> Side note, why do I have to go to the Funtoo website to see
> something as beautiful as this? 
> http://www.funtoo.org/Gentoo_Ecosystem
> 
> 

because we have no manpower to do anything else besides maintaining
packages. And even that, sometimes feels it's beyond our limits. See
all the abandoned areas we have (even core packages seem to suffer
from time to time eg gcc)

I've been around for a long time and I don't seem to remember getting
new developers on board for non-technical stuff (wallpapers, PR,
artwork etc). It seems we simply can't attract this kind of interest
anymore. We might get some contributions from time to time but we
really can't compete what other distros do in that area.

- -- 
Regards,
Markos Chandras
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=
=A7Kv
-----END PGP SIGNATURE-----

Re: [gentoo-project] Some focus for Gentoo

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 896 bytes --]

On Tue, Jan 13, 2015 at 09:43:23PM -0600, 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/

Donnie,

I have been a Gentoo dev for 10 years, almost as long as you have, and I
agree with your post.

Over the years, I have seen this distro go from being the distro that
was ahead of the curve, always innovating and embracing new tech, to a
distro that is stagnating.

I also keep hearing that Gentoo is a meta distribution and Gentoo is
about choice. Well, the issue is that we can't be all things to all
people, and that is what we have been attempting to be for the last few
years now.

I agree that we need to find a focus for Gentoo, and I like your use
cases.

What do others think?

William


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Eray Aslan]

On Tue, Jan 13, 2015 at 09:43:23PM -0600, 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.

Agree in general.

Gentoo does provide a good environment for development.  I am not sure
about how much awareness there is in the community though.

Not enough experience with embedded but agree that flexibility is/should
be a major point.

Meh for people wanting to learn linux.  I see your point but am not sure
if it is really relevant in this time and age.

As a side note, Gentoo for production is certainly doable and has some
advantages provided that one has the necessary knowledge/ability
in-house.

I am assuming that you want to focus distrowide efforts into points you
mentioned in your blog.  Great news as some focus and direction will
help.  Yay for bringing it up.

-- 
Eray

Re: [gentoo-project] Some focus for Gentoo

[vivo75]

Il 14/01/2015 04:43, Donnie Berkholz ha scritto:
> 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/
>
People who want to build containers & co (lxc, kvm ...), these people
would benefit the same as embedded from an easier to strip down gentoo.

Re: [gentoo-project] Some focus for Gentoo

[Ultrabug]

On 14/01/2015 04:43, 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 general feeling is that the focuses we really need must tend to
attract more users. I'll give here my 2 cents in light on how it
evolved here at work.


1. People who need extreme flexibility

I can't agree more tho I don't have embedded in mind : I can't help to
think we're too greatly understaffed to shine on all those side and
time consuming topics of application so it would be a waste of a main
distro focus.

Every sysadmin I talked to or recruited coming from another distro
always tries to argue on how easy it is on Ubuntu/Debian/Bla until
they need something specific on their package or simply don't need
something else and keep dependencies to a minimum.

Then they understand what is to me one of the true value of Gentoo and
why we chose and run it in production for so long here.


2. People who want to learn how Linux works

They are rare or can easily get tired on failures. I still think we've
a lot to ease there (see next point).


3. People developing software ARE NOT People who want to learn how
Linux works

I heard about two upstream devs running Gentoo for the exact reasons
you mentioned and I guess we all fit your description here but I'm not
sure we represent a large part of the developers community.

Let's face it, you've all seen it in any dev conference : most devs
run on Mac now. Because "it just works out of the box and I don't have
to deal with X or Y" I hear all the time [1].

While I enforce a quite strong Gentoo training to any new recruit
joining *my* teams (both ops & devs) I have to admit that we chose
Ubuntu for the rest of the devs here because the entry ticket of
Gentoo is too high and too demanding to the lambda developer who's not
Linux minded (meaning, running Windows or Mac at home).

Hearing "I prefer to spend X hours coding than battling with my
system" is common (and makes sense sometimes).



4. That being said

I think a "Make Gentoo usable quicker and faster" focus would greatly
improve the "People developing software" and "People who want to learn
how Linux works" ones.

This is what I believe makes Arch Linux attractive to devs and people
willing to understand Linux better and I'm convinced that it's one of
the few topics Gentoo should really focus on.

Thanks again for the topic mate.

Ultrabug

[1] It still amazes me that people developing free software can
endorse Apple hardware but it is the reality and my lonely point of view.

Re: [gentoo-project] Some focus for Gentoo

[Andreas K. Huettel]

Am Dienstag 13 Januar 2015, 21:43:23 schrieb Donnie Berkholz:
> 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/

Hi Donnie, 

I agree that your three main points are very good use cases for Gentoo, I'm 
just not sure yet how they exactly lead to more focussing.

I'd like to add another category, which definitely attracts people to Gentoo 
(as long as the relevant parts of Gentoo are well-maintained):

"People who want to run bleeding edge code"

^ As an example, this was for a long time and likely still is the attraction 
of the Gentoo KDE ebuilds (live, live stable branch, and alpha, beta, release 
candidates, releases all as zero day bumps). The KDE team has historically 
attracted a lot of developers to Gentoo, who now contribute in all sorts of 
different projects. 

[Now if we could only get Qt 5 unmasked...]

Cheers, 
Andreas

-- 
Andreas K. Huettel
Gentoo Linux developer
kde, council

Re: [gentoo-project] Some focus for Gentoo

[Brian Wiborg]

Hi,

I am - except for one small ebuild some years ago - a plain Gentoo user
and not a dev at all. Still I'd like to comment on your post.

I am personally very passionate about Gentoo. Especially because of
top-3 of you blog post. Whenever people ask me why the heck I would take
the struggle of running Gentoo on my laptop and desktop I answer

    "because nowhere else I have been able to learn so much about
    Linux and have so much choice about what I want to dive into."

I work as Linux Admin for a living and am planning to do so until I
retire. I work in a business hosting company, so my clients are Linux
Admins themselves and not plain regular users or customer. If I want to
do my job well, I need to have a deeper understanding than them about
everything they could consult me about.
When I first got in contact with Linux to host a game-server for friends
about 6-ish years ago I didn't even know how to run an executable
(prefix it with ./) and had to ask. I should leave my hands off of Linux
if I don't even know how to run an executable, they said. I went from
Dapper Drake over some Ubuntu versions to Squeeze. After that I went
directly to Gentoo and have been sticky for only about three years now.
As of today, my boss believes it to be a great idea to have me in
management and so I've been promoted to lead of the hosting-team.

    Thank you Gentoo, I wouldn't have come so far without you!

I conclude that your point is absolutely valid and it would break my
heart to see this use case get out of focus.

Additionally, I would like to propose another use case:

= Felxibility =

In the age of cloud computing where rock-solid and hardened setups have
evolved to fire-and-forget throw-aways, people seem to forget that there
are still some classical environments out there. Environments of more
static nature, than disposal hosts that one just provisions on-the-fly
and destrys when done. Environments such as good old shared-hosting.
Due to the nature of portage and the ability to work in slots and mask
specific ebuild version, I know of no other distribution so perfectly
fit for such environments than Gentoo.

Ever went through a PHP upgrade with a couple thousand web-roots and
databases in a shared-hosting Debian environment? It is painful!

As a matter of fact, I know of another hosting company in my area that
builds entirely upon Gentoo while specializing in individualized
high-performance e-commerce hosting. There is no better distro for
that. Of the top of my head I would guess that is around 2-3k Gentoo
setups right there (my company and the other hoster). Consider that.

baccenfutter
o7

Re: [gentoo-project] Some focus for Gentoo

[Amadeusz Żołnowski]

[-- Attachment #1: Type: text/plain, Size: 1588 bytes --]

1. People developing software

It is especially true for C/C++.  We should focus on bringing
bleeding-edge GCC, Clang and core libraries like Boost and Qt.  Next
thing is that we should keep track of new languages, IDE-s, text
editors, and other developer tools coming up and not only adding
silentely ebuilds, but also informing upstream developers that we get
their software into our distro - it's quite common practice to inform in
which distros app is available.

To me Gentoo is the only choice if it comes to development.
Unfortunately I must use Ubuntu at work and it really sucks at that.

2. One install for ever

This is really cool about Gentoo that you need to install it only once -
no reinstalls, no risky release upgrades. If you keep Gentoo up to date
you get a stable system until your machine dies (by overheating?  ;-)).

3. Web apps

I feel that's the topic which other distros avoid, because of too rapid
development of web apps. We could improve on that topic and provide
bleeding edge ebuilds. We could also ask an upstream to give a note that
on Gentoo it's just a matter of emerge thewebapp and a user doesn't have
to follow complicated set-up guide.

I find this especially problematic on Ubuntu which cannot keep up with
Ruby releases.

4. Security - Gentoo hardened

Gentoo gives choice about security models. Do other distros provide any
choice on that?

5. Choice of init system

In the light of recent flame wars we can take an advantage of giving a
choice on that topic. (-:


Cheers,

-- 
Amadeusz Żołnowski

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 472 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 484 bytes --]

On Wed, 14 Jan 2015 22:22:21 +0100
Amadeusz Żołnowski <aidecoe@gentoo.org> wrote:
> 1. People developing software
> 
> It is especially true for C/C++.  We should focus on bringing
> bleeding-edge GCC, Clang and core libraries like Boost and Qt.

But Gentoo is the only distribution that doesn't allow multiple
versions of g++ to be run side by side... It's also not possible to
release dist tarballs using libtool that are built on a Gentoo host.

-- 
Ciaran McCreesh

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Amadeusz Żołnowski]

[-- Attachment #1: Type: text/plain, Size: 497 bytes --]

Ciaran McCreesh <ciaran.mccreesh@googlemail.com> writes:

> On Wed, 14 Jan 2015 22:22:21 +0100
> Amadeusz Żołnowski <aidecoe@gentoo.org> wrote:
>> 1. People developing software
>> 
>> It is especially true for C/C++.  We should focus on bringing
>> bleeding-edge GCC, Clang and core libraries like Boost and Qt.
>
> But Gentoo is the only distribution that doesn't allow multiple
> versions of g++ to be run side by side...

GCC is slotted, isn't it?


-- 
Amadeusz Żołnowski

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 472 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 735 bytes --]

On Wed, 14 Jan 2015 22:42:56 +0100
Amadeusz Żołnowski <aidecoe@gentoo.org> wrote:
> Ciaran McCreesh <ciaran.mccreesh@googlemail.com> writes:
> > On Wed, 14 Jan 2015 22:22:21 +0100
> > Amadeusz Żołnowski <aidecoe@gentoo.org> wrote:
> >> 1. People developing software
> >> 
> >> It is especially true for C/C++.  We should focus on bringing
> >> bleeding-edge GCC, Clang and core libraries like Boost and Qt.
> >
> > But Gentoo is the only distribution that doesn't allow multiple
> > versions of g++ to be run side by side...
> 
> GCC is slotted, isn't it?

Yes, but you can only have one "active" libstdc++ at a time, thanks to
gcc-config weirdness. No other distribution has this problem.

-- 
Ciaran McCreesh

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[gentoo-project] Re: Some focus for Gentoo

[Martin Vaeth]

Ciaran McCreesh <ciaran.mccreesh@googlemail.com> wrote:
>
> It's also not possible to
> release dist tarballs using libtool that are built on a Gentoo host.

Can you be more verbose on this?
What will be different if you call "make dist" on a Gentoo host?

Re: [gentoo-project] Re: Some focus for Gentoo

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 693 bytes --]

On Thu, 15 Jan 2015 06:53:30 +0000 (UTC)
Martin Vaeth <martin@mvath.de> wrote:
> Ciaran McCreesh <ciaran.mccreesh@googlemail.com> wrote:
> > It's also not possible to
> > release dist tarballs using libtool that are built on a Gentoo host.
> 
> Can you be more verbose on this?
> What will be different if you call "make dist" on a Gentoo host?

Gentoo's libtool has all kinds of patches applied to it. These patches
are included in the libtool code that's shipped as part of a dist
tarball. These patches are not necessarily portable.

There's a good explanation of one instance here:

http://blog.bz2.nl/2005/12/12/compiling-irssi-0810-on-solaris-10/

-- 
Ciaran McCreesh

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

[gentoo-project] Re: Some focus for Gentoo

[Martin Vaeth]

Ciaran McCreesh <ciaran.mccreesh@googlemail.com> wrote:
>
> There's a good explanation of one instance here:
>
> http://blog.bz2.nl/2005/12/12/compiling-irssi-0810-on-solaris-10/

Thanks for the link. This looks worrisome, indeed.

Re: [gentoo-project] Some focus for Gentoo

[Zac Medico]

On 01/13/2015 07:43 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/
> 

I think that improved binary package support will allow Gentoo to
compete with binary distros, making Gentoo applicable to a much wider
audience.

I am currently working on an new binary package repository (PKGDIR)
layout [1] and binary package soname dependencies [2] (soname deps will
require us to establish a naming convention for multilib ABI identifiers
[3]).

[1] http://thread.gmane.org/gmane.linux.gentoo.portage.devel/5031
[2] http://thread.gmane.org/gmane.linux.gentoo.devel/94145
[3] http://thread.gmane.org/gmane.linux.gentoo.devel/94176
-- 
Thanks,
Zac

Re: [gentoo-project] Some focus for Gentoo

[Amadeusz Żołnowski]

[-- Attachment #1: Type: text/plain, Size: 681 bytes --]

Zac Medico <zmedico@gentoo.org> writes:

> On 01/13/2015 07:43 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/
>> 
>
> I think that improved binary package support will allow Gentoo to
> compete with binary distros, making Gentoo applicable to a much wider
> audience.

As far as I understand we should focus on something that other distros
fail at.  (I'm not saying we should ignore binary package support, but
definitely it's place is not in top-3.)


-- 
Amadeusz Żołnowski

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 472 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Zac Medico]

On 01/14/2015 01:46 PM, Amadeusz Żołnowski wrote:
> Zac Medico <zmedico@gentoo.org> writes:
> 
>> On 01/13/2015 07:43 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/
>>>
>>
>> I think that improved binary package support will allow Gentoo to
>> compete with binary distros, making Gentoo applicable to a much wider
>> audience.
> 
> As far as I understand we should focus on something that other distros
> fail at.  (I'm not saying we should ignore binary package support, but
> definitely it's place is not in top-3.)

Well, maybe you could say that other distros fail at providing an
extremely flexible build-system to feed their binary package
repositories? If you look at it this way, Gentoo's missing piece is the
binary package repository support.

Thanks,
Zac

Re: [gentoo-project] Some focus for Gentoo

[Brian Dolbec]

On Wed, 14 Jan 2015 14:33:27 -0800
Zac Medico <zmedico@gentoo.org> wrote:

> On 01/14/2015 01:46 PM, Amadeusz Żołnowski wrote:
> > Zac Medico <zmedico@gentoo.org> writes:
> > 
> >> On 01/13/2015 07:43 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/
> >>>
> >>
> >> I think that improved binary package support will allow Gentoo to
> >> compete with binary distros, making Gentoo applicable to a much
> >> wider audience.
> > 
> > As far as I understand we should focus on something that other
> > distros fail at.  (I'm not saying we should ignore binary package
> > support, but definitely it's place is not in top-3.)
> 
> Well, maybe you could say that other distros fail at providing an
> extremely flexible build-system to feed their binary package
> repositories? If you look at it this way, Gentoo's missing piece is
> the binary package repository support.
> 
> Thanks,
> Zac
> 

Yes, while maybe not a top 3,, at least a top 5. It is definitely a good
one to accomplish.

Better binary support will improve production use in multi-system
environments, making maintenance much easier and Gentoo a much more
attractive option.  Both increasing our user base, and most likely,
contributing members, even if they are not devs.

-- 
Brian Dolbec <dolsen>

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.

Re: [gentoo-project] Some focus for Gentoo

[Daniel Campbell]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 01/13/2015 07:43 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/
>
>
> 
I'm not a developer, but I am interested in becoming one. That said, I
don't understand why relevance or focus are important. Is Gentoo not a
distribution made by people who feel that Gentoo meets their needs and
(for those who contribute) want to improve it? One of the core values
of Gentoo is the concept of choice. Without choice, none of the three
things you listed on the blog post are possible. People who were
trying to avoid things being pushed on them from other distros would
never have come here, and wouldn't be contributing. And that's only
one reason that people come to Gentoo.

With any other "migration" of sorts, you can rightfully expect that
most of the "refugees" will just be leeching users, but I would wager
that a not-insignificant portion of them will be users that, like me,
want to contribute or join the development effort and make Gentoo a
better distro. Choice and the "meta distribution" concepts are part of
that. I don't think they're the end-all, be-all of Gentoo, but those
two things open up a plethora of options and possible "focuses".

Gentoo is the furthest thing I can think of from being "for everyone".
But to anyone with the know-how or dedication, Gentoo really can fit
any use-case. You might need to build a number of scripts or tools on
top of the already great ones that Gentoo provides in order to meet
your needs, but there's a lot already in place that creates a lot of
value for a lot of different use cases. One's grandma or
non-tech-savvy friend doesn't really fit in on Gentoo since it
requires a passing understanding of a GNU/Linux system, and that's
okay. Gentoo, from what I can tell, is meant for the people dedicated
to running specific, lightweight, and/or versatile systems. Whether
they're a dev, need to manage 100+ machines, prefer a lean and fast
desktop, or need to pare it down for an embedded product, Gentoo can
and does meet all those needs.

Maybe I'm misunderstanding you, but it seems like "focus" is referring
to actively marketing or advertising Gentoo to garner more interest.
Given the other marketing and pushing seen in the FOSS world
(specifically related to a certain audio stack and init system) I
don't see people taking kindly to that sort of behavior. Gentoo should
(and does) stand on its own as a great distribution to do many
different things with, without a marketing agenda.

But assuming I do misunderstand, are you talking about bringing those
benefits up if/when someone asks about Gentoo and/or putting it on the
front page of the wiki and the main site(s)? I see little to no harm
in adding bullet points to a list of benefits, and maybe Gentoo could
benefit from some good guides (in blogs or the wiki) for getting a
sane dev environment or setting up something for an Arduino or
Raspberry Pi, etc. I think doing that would be really good for giving
outsiders a more concrete example of things that Gentoo can do more
easily than other distributions. And that will almost certainly bring
in users. But what about the real problem?

The real problem, based on what I've read here on gentoo-dev,
gentoo-project, and so on, is a lack of manpower. A lot of Gentoo devs
supposedly don't work together too much, and prefer doing their own
thing. And to a degree that's the way it should be, since devs are all
volunteers. Nobody should be forced to do something on Gentoo unless
what they're doing is damaging to the distro as a whole. But let me
get back to my point...

More users means more bugs, more requests, more ideas, and
essentially... more work. There needs to be a smoother process for
people to become developers without sacrificing (too much of) the
integrity of Gentoo. More developers means more maintainers, more eyes
on bugs, and yes, more trouble coordinating group efforts. If the
pathway to become a developer is a bit smoother and easier, more users
- -- with considerable knowledge and/or a strong will to learn -- will
become valuable assets to the Gentoo team. A more prominent way to
tell the Gentoo community what challenges Gentoo is facing (like the
multilib migration from emul-linux-* to the abi_x86_32 USE flag), and
a clear pathway to joining the development efforts (say, with
developers who have the time to mentor adding their name to a list,
their specialties, etc) would do a lot to help visibility of Gentoo's
need for manpower. Then these new developers can lend their hand to
areas that they feel strongly about and are personally invested in.

For example, if/when I become a Gentoo dev I want to assist in ebuilds
for games, a few tools I personally use, and would want to join
efforts in OpenRC and eudev to help keep GNU/Linux free from vertical
integration efforts that most other distros are pushing for. I'm sure
I'm not the only Gentoo user out there who feels strongly about a few
things and could offer help on things that Gentoo seems to be in dire
need of. Personally, I'm not limited to what I listed above. If I
think I can learn enough to help, I'd be glad to help in any way I
can. Gentoo's a great distro with a knowledgeable and enthusiastic
community. If there's a way to get more of the community involved in
Gentoo's maintenance and progression, I think that's the way forward.
Switching the documentation to a wiki was a fantastic move that seems
to be improving Gentoo's docs by quite a bit and a great example of
what can be done to improve the distro.

The current issues I think Gentoo faces are more related to manpower
than relevance, focus, or usefulness. Maybe some new blood (to assist
the older, knowledgeable blood) is needed to help Gentoo move forward
faster.

Just my 2¢.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iQEcBAEBAgAGBQJUtxoVAAoJEJUrb08JgYgHIB4IAIFPbtVEXnd92HAxXa5t3upK
jLzm6xP9MyV4cDDyvHwUc//FBF1uoriwdtA9NOiNCBxszyoEhAfvCtXAQHlADrUx
yHEjserNPLBkkR75UCfrSzSCMCT/4AO4AjmX81f+ZSrh/O24Vc+p1Y6LzW+CHkYP
V+Yz7KIQHaExFeUmz8cwNQDweSxqDlsYo2l05FrUZjcqwPtj2nB46jXiGbXwCZ8I
nURVKBotmquOKHEO1azwnbW1WkP0tsJSdU3lFtTw0hEMNQu/9zsOVDWyvfL19Fh4
T6pgK9LLKGNRZuWWEYvV3oDPJzp2lWwDP10yDy827JhqNbCmjh6xBT4t7K9n4bQ=
=34t2
-----END PGP SIGNATURE-----

Re: [gentoo-project] Some focus for Gentoo

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 4036 bytes --]

On 17:38 Wed 14 Jan     , Daniel Campbell wrote:
> I'm not a developer, but I am interested in becoming one. That said, I
> don't understand why relevance or focus are important. Is Gentoo not a
> distribution made by people who feel that Gentoo meets their needs and
> (for those who contribute) want to improve it? One of the core values
> of Gentoo is the concept of choice. Without choice, none of the three
> things you listed on the blog post are possible.

Daniel,

Thanks for your comments. Gentoo has been about choice where it makes 
sense and is pragmatic — not about choice for the sake of choice. This 
subtlety is often lost through the mists of time. But it drives other 
key tenets of Gentoo like packaging proprietary software, and all the 
weird restrictions that enable you to package things that can't be 
mirrored, etc.

Here's one way to think of my proposal. Imagine a REST API. The provider 
of that API might also have a few SDKs for some popular languages or 
frameworks, along with documentation to make it really easy for them. 
But *the REST API is still there*, usable by anyone in any language. It 
just takes a bit more work. For Gentoo, this would work in much the same 
way — it essentially requires the existence of an underlying platform, 
but the focus would be on the use cases.

There are two major problems that drive this. One is that Gentoo's 
popularity has greatly declined in the past decade. While you could 
argue that fewer users isn't necessarily a problem in its own right, 
it's creating severe difficulties for us in recruiting new developers. 
That is a problem, and it's an existential one. We need to change, 
because doing what we're doing will just keep us on the same road.

Second, and this is one thing I've learned from my professional life, is 
that platforms are too ephemeral of a concept in isolation. Platforms 
(or metadistributions) succeed when it becomes crystal clear exactly how 
to use them, typically through creation of sample applications, early 
public customers, or something along those lines.

> Maybe I'm misunderstanding you, but it seems like "focus" is referring 
> to actively marketing or advertising Gentoo to garner more interest.

Focus includes but goes far beyond marketing. Here are some examples of 
what this kind of focus would imply:

- The council would base its decision-making on these use cases
- The QA project would help test that these use cases worked well
- PMS enhancements would be driven in part by these use cases
- Releng goals could be to provide image/package types to support them
- Individual developers would make ebuild-level decisions based on them
- We could use them as a rallying cry to unite Gentoo devs
- And yes, the website would tell these stories to draw in new users

> The real problem, based on what I've read here on gentoo-dev, 
> gentoo-project, and so on, is a lack of manpower.

This is kind of a problem that isn't a problem. We've supposedly had a 
lack of manpower ever since Gentoo's founding, because there are more 
things that we could be doing but aren't.

> The current issues I think Gentoo faces are more related to manpower 
> than relevance, focus, or usefulness. Maybe some new blood (to assist 
> the older, knowledgeable blood) is needed to help Gentoo move forward 
> faster.

And where will that new blood come from?

There's a limited number of ways to improve the conversion rate through 
our recruitment funnel. You can either lower the barriers along the way 
or pour more into the top of it. And while lowering the barriers would 
certainly get people their commit rights faster, I very rarely see 
people who give up halfway through because it's too frustrating. 
(Granted I haven't checked lately but I've been an on-and-off 
recruiter.)

-- 
Thanks,
Donnie

Donnie Berkholz
Council Member / Sr. Developer, Gentoo Linux <http://dberkholz.com>
Analyst, RedMonk <http://redmonk.com/dberkholz/>

[-- Attachment #2: Type: application/pgp-signature, Size: 966 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[hasufell]

Donnie Berkholz:
> 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/
> 

How do you want to _ensure_ focus with 263 developers having direct push
access without any strict review policies?

How do you want to ensure focus if the council and GLEP 39 say that we
may have conflicting ideas in ONE single repository and that we may
voluntarily break tree consistency (can give examples)?

You are tackling the wrong problem. The problem is not lack of ideas and
people having focus on these ideas.

Re: [gentoo-project] Some focus for Gentoo

[Donnie Berkholz]

[-- Attachment #1: Type: text/plain, Size: 789 bytes --]

On 14:41 Thu 15 Jan     , hasufell wrote:
> How do you want to _ensure_ focus with 263 developers having direct 
> push access without any strict review policies?
> 
> How do you want to ensure focus if the council and GLEP 39 say that we 
> may have conflicting ideas in ONE single repository and that we may 
> voluntarily break tree consistency (can give examples)?

I discussed some of this, in terms of what specifically "focus" would 
look like, in my response to Daniel.

> You are tackling the wrong problem. The problem is not lack of ideas and
> people having focus on these ideas.

Instead, it is...?

-- 
Thanks,
Donnie

Donnie Berkholz
Council Member / Sr. Developer, Gentoo Linux <http://dberkholz.com>
Analyst, RedMonk <http://redmonk.com/dberkholz/>

[-- Attachment #2: Type: application/pgp-signature, Size: 966 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 345 bytes --]

On Thu, 15 Jan 2015 13:10:01 -0600
Donnie Berkholz <dberkholz@gentoo.org> wrote:
> > You are tackling the wrong problem. The problem is not lack of
> > ideas and people having focus on these ideas.
> 
> Instead, it is...?

The complete lack of technical progress thanks to it being so hard to
work with Portage.

-- 
Ciaran McCreesh

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Jeroen Roovers]

On Thu, 15 Jan 2015 19:13:26 +0000
Ciaran McCreesh <ciaran.mccreesh@googlemail.com> wrote:

> On Thu, 15 Jan 2015 13:10:01 -0600
> Donnie Berkholz <dberkholz@gentoo.org> wrote:
> > > You are tackling the wrong problem. The problem is not lack of
> > > ideas and people having focus on these ideas.
> > 
> > Instead, it is...?
> 
> The complete lack of technical progress thanks to it being so hard to
> work with Portage.

Good, the usual two trolls are done now?


     jer

Re: [gentoo-project] Some focus for Gentoo

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 298 bytes --]

On Thu, 15 Jan 2015 20:33:33 +0100
Jeroen Roovers <jer@gentoo.org> wrote:
> Good, the usual two trolls are done now?

Probably not just yet. I'm still waiting to hear it extolled that the
community and environment are more important than delivering a decent
product.

-- 
Ciaran McCreesh

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 181 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Seemant Kulleen]

[-- Attachment #1: Type: text/plain, Size: 754 bytes --]

Seems like the same discussion.  What, exactly, is the product?  In a micro
view, you are right, ciaran, that the technology is lacking.  In the macro
view, portage is an engine not the product.  The product, in my view, is
the experience of using Gentoo.

In many ways,  the community is the "product".

2 pennies,

seemantk empathic design
http://seemantk.com
On Jan 15, 2015 11:45 AM, "Ciaran McCreesh" <ciaran.mccreesh@googlemail.com>
wrote:

> On Thu, 15 Jan 2015 20:33:33 +0100
> Jeroen Roovers <jer@gentoo.org> wrote:
> > Good, the usual two trolls are done now?
>
> Probably not just yet. I'm still waiting to hear it extolled that the
> community and environment are more important than delivering a decent
> product.
>
> --
> Ciaran McCreesh
>

[-- Attachment #2: Type: text/html, Size: 1197 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[hasufell]

Donnie Berkholz:
> On 14:41 Thu 15 Jan     , hasufell wrote:
>> How do you want to _ensure_ focus with 263 developers having direct 
>> push access without any strict review policies?
>>
>> How do you want to ensure focus if the council and GLEP 39 say that we 
>> may have conflicting ideas in ONE single repository and that we may 
>> voluntarily break tree consistency (can give examples)?
> 
> I discussed some of this, in terms of what specifically "focus" would 
> look like, in my response to Daniel.
> 
>> You are tackling the wrong problem. The problem is not lack of ideas and
>> people having focus on these ideas.
> 
> Instead, it is...?
> 

Sorry to answer so late. I didn't have the time yet.

The main problem in my opinion is that our organizational concept as a
whole doesn't work so well... or at least not any more. With concept I
don't just mean focus on technical stuff, but the question where does
that focus come from and how do we process ideas?

As I said... we have a lot of people with ideas and some are very
focussed. Ofc we can discuss a technical focus and you/we might even be
lucky and the majority agrees with you... now. And in 3 months?

Afais gentoo is a very loose group of devs with optional communication,
but everyone having access to one repository. Conflicting ideas and
inconsistencies are allowed, unless it's about EAPI. Maybe this has
worked for some time, but I think that was rather coincidence.
And this has lead to several problems:
1. very high bus factor in some areas (as in: lets not hope mgorny,
vapier or jer quit gentoo... commit rate will go down a lot or bugzilla
just die)
2. point 1 also resulted into some devs getting special privileges which
sometimes amplifies point 7
3. low QA
4. difficult collaboration model
5. major conflicting ideas not being properly mediated: e.g. multilib vs
portage-multilib, because portage-multilib wasn't in-tree anyway and
multilib was an eclass concept
6. allowing inconsistencies that may break user experience: e.g. we have
games.eclass, but the council says... well, you may or may not use it
(instead of saying we wipe it out completely or we follow the concept of
special permissions consistently)
7. a lot of organizational problems these days and a high burn-out rate
for people who come up with new ideas

I think there are only two ways out of it:
1. Make gentoo more centralized to ensure focus. One possibility would
be to give a lot more power to the council and make it the main hive for
new ideas (already proposed that a year or more ago during council
election, afair).
2. Make gentoo more decentralized and reduce the number of core-devs to
allow conflicting ideas which is one of the main points of GLEP 39, IMO.
But now make this idea actually possible on the technical and
methodology level.
This would imply a major restructuring of the organizational model and a
redefinition of "core development", which would also ensure focus of
that core development. (already proposed this a few weeks/months ago)

I think we are currently a hybrid of both concepts and I think that is a
problem. It's not enough to come up with ideas to focus on. You also
have to come up with a way to ensure focus... or a way that doesn't need
to ensure focus, at least for some areas.

Re: [gentoo-project] Some focus for Gentoo

[Rich Freeman]

On Tue, Jan 20, 2015 at 11:00 PM, hasufell <hasufell@gentoo.org> wrote:
> 2. Make gentoo more decentralized and reduce the number of core-devs to
> allow conflicting ideas which is one of the main points of GLEP 39, IMO.
> But now make this idea actually possible on the technical and
> methodology level.

You've talked about the first sentence in this suggestion before, but
not really about the second.  Just what does making this possible from
a technical level look like, other than how things work today?  How
can we have both games.eclass and no games.eclass but due to a
technical/methodology change there are now less problems?  Ditto for
the two multilib implementations?

It seems to me that your #2 and #1 are really the same thing - having
fewer core devs actually eliminates conflicting ideas and increases
focus, simply by virtue of the fact that there are fewer people left
to disagree.  It is nice to speak of having your cake and eating it
too, but saying it doesn't make it happen, and I don't want a
bait-and-switch where we sell everybody on a concept but then say,
"well, there was no way this was ever going to work out like we talked
about..."

My main concern with this approach is that it seems reasonably likely
to just result in having zero devs.  When I've seen big shakeups in
other organizations with the goal of revitalizing things the result is
often that the organization just disbands.  The chances of a new
person becoming committed tends to be low, while the chance of an
existing contributor remaining committed is much higher.  Anytime you
shake things up you tend to have far more to lose than to gain as a
result.

I think that a major shakeup only makes sense if we can actually
demonstrate a new model in operation, and it actually solves our
problems.

-- 
Rich

Re: [gentoo-project] Some focus for Gentoo

[Julian]

Rich Freeman:
> On Tue, Jan 20, 2015 at 11:00 PM, hasufell <hasufell@gentoo.org> wrote:
>> 2. Make gentoo more decentralized and reduce the number of core-devs to
>> allow conflicting ideas which is one of the main points of GLEP 39, IMO.
>> But now make this idea actually possible on the technical and
>> methodology level.
> 
> You've talked about the first sentence in this suggestion before, but
> not really about the second.  Just what does making this possible from
> a technical level look like, other than how things work today?

We would need much better tools and PM support for overlays. I've talked
about this too and there are examples of at least two distros that are
running a more decentralized model:
1. NixOS, also see my thread on nix-dev about how they want to ensure
focus [0]
2. exherbo: I consider this at least some sort of proof of concept for
tool-driven distributed development with automated inter-overlay
tinderbox runs. The PM has also very good overlay support. Unfortunately
they don't attract people with their arrogance (it's all over their
docs) and they haven't understood that there is a difference between
distributed and fragmented.

And there are more things that can be done, I've talked about some of
them too already.

If you have a different idea of decentralization, then please share it.

> How
> can we have both games.eclass and no games.eclass but due to a
> technical/methodology change there are now less problems?  Ditto for
> the two multilib implementations?
> 

Well, for once: forking an overlay is easier than forking the whole
gentoo tree.
But that's not necessarily the main point. Ideas would not be decided by
"who does something first in the tree", but by more dynamic processes
about approval in the community. Someone starts a repo and does stuff.
If people like it, they are going to use it and focus their contribution
there.

> It seems to me that your #2 and #1 are really the same thing - having
> fewer core devs actually eliminates conflicting ideas and increases
> focus, simply by virtue of the fact that there are fewer people left
> to disagree.

They are fundamentally different things. They just try to fix the same
problem.

I can't really say which one is better for gentoo specifically.

> My main concern with this approach is that it seems reasonably likely
> to just result in having zero devs.  When I've seen big shakeups in
> other organizations with the goal of revitalizing things the result is
> often that the organization just disbands.  The chances of a new
> person becoming committed tends to be low, while the chance of an
> existing contributor remaining committed is much higher.  Anytime you
> shake things up you tend to have far more to lose than to gain as a
> result.
> 

If you want to go business analyst, all right.
An austrian millionaire (Gerald Hörhan) sometimes gives lectures in
german economy universities and he's often talking about the 3 main
points why companies fail. Among those 3 is: conflicts between
shareholders/associates (not sure what's the precise translation). How
about having 200 of those? Sounds great.

> I think that a major shakeup only makes sense if we can actually
> demonstrate a new model in operation, and it actually solves our
> problems.
> 

I've already given two examples of a new model in operation. Also, I
haven't said "let's do this tomorrow". IMO it will never happen anyway.
I am convinced that gentoo will rather die slowly, because people are
afraid to change the course which leads us to a nice thick wall, because
we might derail.
I'm still not sure if it will die because of "lack of manpower" or
because our technology is so messed up that it becomes unusable.


--
[0] https://www.mail-archive.com/nix-dev%40lists.science.uu.nl/msg12446.html

Re: [gentoo-project] Some focus for Gentoo

[Rich Freeman]

On Wed, Jan 21, 2015 at 11:36 AM, Julian <hasufell@hasufell.de> wrote:
>
> We would need much better tools and PM support for overlays. I've talked
> about this too and there are examples of at least two distros that are
> running a more decentralized model:
> 1. NixOS, also see my thread on nix-dev about how they want to ensure
> focus [0]

I still don't get how NixOS avoids some fundamental issues.  As I
understand it they assign a unique ID to every build of everything,
and the dependencies are expressed in the same way.  So, my build 123
of appfoo v1.1 depends on build 456 of libz v2.  Your package might
depend on build 457 of libz v2, and that is fine since both versions
will be installed side-by-side.  Then we get to have two copies of
libz in RAM and maybe build 456 has a security vulnerability.  It is
better than static linking, but not by far.

Sure, we could do that with Gentoo, but I'm not sure this really
addresses the concerns you brought up, like...


>> How
>> can we have both games.eclass and no games.eclass but due to a
>> technical/methodology change there are now less problems?  Ditto for
>> the two multilib implementations?
>>
>
> Well, for once: forking an overlay is easier than forking the whole
> gentoo tree.
> But that's not necessarily the main point. Ideas would not be decided by
> "who does something first in the tree", but by more dynamic processes
> about approval in the community. Someone starts a repo and does stuff.
> If people like it, they are going to use it and focus their contribution
> there.

So, which is it?  You point out that we have inconsistencies in our
tree because we have games.eclass but we don't force everybody to use
it.

How is it better if the games are scattered across 47 overlays
instead, and some of them use games.eclass, others use something else
that sticks games in an entirely different group/path/whatever, others
follow upstream, and maybe half the overlays do security updates and
the other half don't?  Maybe half of them use the multilib eclass and
half use portage-multilib.

So, are we OK with inconsistencies or not?  If we are OK with them,
then why complain about it?

>
>> I think that a major shakeup only makes sense if we can actually
>> demonstrate a new model in operation, and it actually solves our
>> problems.
>>
>
> I've already given two examples of a new model in operation. Also, I
> haven't said "let's do this tomorrow". IMO it will never happen anyway.
> I am convinced that gentoo will rather die slowly, because people are
> afraid to change the course which leads us to a nice thick wall, because
> we might derail.
> I'm still not sure if it will die because of "lack of manpower" or
> because our technology is so messed up that it becomes unusable.

So, I'm not convinced that we won't find our way, but if Gentoo stops
existing because we've all found a better way and moved on to it, then
really there is no loss unless you're really attached to the name,
"Gentoo."  None of us own stock in the Gentoo Foundation.  If
something new already exists and is doing it better, and we all just
decide to switch, then we all reap the benefits.

However, it seems likely to me that if there were an overwhelming
sense that a new model would really work for most of us, we'd change.
The problem is that not everybody is here for the same reason, and
we're best off trying to tap into that rather than fighting it.

-- 
Rich

Re: [gentoo-project] Some focus for Gentoo

[hasufell]

Rich Freeman:
> On Wed, Jan 21, 2015 at 11:36 AM, Julian <hasufell@hasufell.de> wrote:
>>
>> We would need much better tools and PM support for overlays. I've talked
>> about this too and there are examples of at least two distros that are
>> running a more decentralized model:
>> 1. NixOS, also see my thread on nix-dev about how they want to ensure
>> focus [0]
> 
> I still don't get how NixOS avoids some fundamental issues.  As I
> understand it they assign a unique ID to every build of everything,
> and the dependencies are expressed in the same way.  So, my build 123
> of appfoo v1.1 depends on build 456 of libz v2.  Your package might
> depend on build 457 of libz v2, and that is fine since both versions
> will be installed side-by-side.  Then we get to have two copies of
> libz in RAM and maybe build 456 has a security vulnerability.  It is
> better than static linking, but not by far.
> 

All copies are known to the PM. That's nothing like bundles libs or
static linkage. It's probably a triviality to develop a tool that checks
for vulnerable libraries (not sure if there already is one).

I didn't say we should become NixOS. It was just one example how
decentralized packaging CAN work. There are other examples.

> 
>>> How
>>> can we have both games.eclass and no games.eclass but due to a
>>> technical/methodology change there are now less problems?  Ditto for
>>> the two multilib implementations?
>>>
>>
>> Well, for once: forking an overlay is easier than forking the whole
>> gentoo tree.
>> But that's not necessarily the main point. Ideas would not be decided by
>> "who does something first in the tree", but by more dynamic processes
>> about approval in the community. Someone starts a repo and does stuff.
>> If people like it, they are going to use it and focus their contribution
>> there.
> 
> So, which is it?  You point out that we have inconsistencies in our
> tree because we have games.eclass but we don't force everybody to use
> it.
> 
> How is it better if the games are scattered across 47 overlays
> instead, and some of them use games.eclass, others use something else
> that sticks games in an entirely different group/path/whatever, others
> follow upstream, and maybe half the overlays do security updates and
> the other half don't?  Maybe half of them use the multilib eclass and
> half use portage-multilib.
> 
> So, are we OK with inconsistencies or not?  If we are OK with them,
> then why complain about it?
> 

It is better, because you can actually choose what you want without
messing with complex masks and workarounds.
And because people could even set up monolithic repositories (actual
distros) that merge the work of several overlays (that's what happens in
NixOS afaiu, some companies seem to be interested in that work too),
straightening out things, checking for compatibility, removing
overlapping parts, etc.
People already use overlays so extensively that I don't think the world
would collapse from the user perspective. And keep in mind that a lot of
gentoo projects already moved their main work to github, because they
cannot keep up the massive amount of work without community contributors.

I think modularity does not only make sense for programming in general,
but also for packaging. But there must be a concept behind it as well,
otherwise you just get terrible fragmentation and ::mynick overlays (see
exherbo).

The main point about decentralized packaging is to shrink the tree (and
needed manpower) and allow more community effort to shape the "distro"
while still maintaining QA through a review workflow (to not become the
next arch linux user repository, aka AUR).
Then you can focus on what is most important: PM, EAPI, toolchain maybe
and most importantly... review.

But I've said most of this already.

>>
>>> I think that a major shakeup only makes sense if we can actually
>>> demonstrate a new model in operation, and it actually solves our
>>> problems.
>>>
>>
>> I've already given two examples of a new model in operation. Also, I
>> haven't said "let's do this tomorrow". IMO it will never happen anyway.
>> I am convinced that gentoo will rather die slowly, because people are
>> afraid to change the course which leads us to a nice thick wall, because
>> we might derail.
>> I'm still not sure if it will die because of "lack of manpower" or
>> because our technology is so messed up that it becomes unusable.
> 
> So, I'm not convinced that we won't find our way, but if Gentoo stops
> existing because we've all found a better way and moved on to it, then
> really there is no loss unless you're really attached to the name,
> "Gentoo."  None of us own stock in the Gentoo Foundation.  If
> something new already exists and is doing it better, and we all just
> decide to switch, then we all reap the benefits.
> 
> However, it seems likely to me that if there were an overwhelming
> sense that a new model would really work for most of us, we'd change.
> The problem is that not everybody is here for the same reason, and
> we're best off trying to tap into that rather than fighting it.
> 

I don't have a completely worked out idea for gentoo. I couldn't even
write a GLEP about this and I certainly won't, because it would be a
huge waste of time. The level of enthusiasm is just too low and any such
thing cannot and will not happen on GLEP level. It would have to be a
process, a community effort, not something the council decides on and
says "ok, now we can implement it" and would imply studying other
distros, checking out their workflow etc.

And btw., this was only one of two proposals.

My point was: it's not enough talking about "what should we focus on?".
How do you implement that focus?

What's your idea there, except of keeping status quo which clearly does
not work since even gentoo oldtimers seem to realize something is going
wrong?

Re: [gentoo-project] Some focus for Gentoo

[Rich Freeman]

On Wed, Jan 21, 2015 at 3:11 PM, hasufell <hasufell@gentoo.org> wrote:
> Rich Freeman:
>> On Wed, Jan 21, 2015 at 11:36 AM, Julian <hasufell@hasufell.de> wrote:
>>>
>>> We would need much better tools and PM support for overlays. I've talked
>>> about this too and there are examples of at least two distros that are
>>> running a more decentralized model:
>>> 1. NixOS, also see my thread on nix-dev about how they want to ensure
>>> focus [0]
>>
>> I still don't get how NixOS avoids some fundamental issues.  As I
>> understand it they assign a unique ID to every build of everything,
>> and the dependencies are expressed in the same way.  So, my build 123
>> of appfoo v1.1 depends on build 456 of libz v2.  Your package might
>> depend on build 457 of libz v2, and that is fine since both versions
>> will be installed side-by-side.  Then we get to have two copies of
>> libz in RAM and maybe build 456 has a security vulnerability.  It is
>> better than static linking, but not by far.
>>
>
> All copies are known to the PM. That's nothing like bundles libs or
> static linkage. It's probably a triviality to develop a tool that checks
> for vulnerable libraries (not sure if there already is one).

I'm not saying that anything bundles libs.  My understanding is that
packages are basically all content-hashed and all dependencies
including linking is against one of these.  So, if you want you can
have 14 versions of the same library with the same SONAME and
completely different ABI/APIs and it will all work, because each
package is linked against the EXACT library it was built against.  Two
packages can still share the same library (in disk and in RAM), but
they could also use two different libraries which on most other
distros would collide (identical version, SONAME, etc).

So, when you use a package you're using the exact same code the
maintainer uses, including all the libraries.

I'm sure this breaks down at some point - you can't have 47 different
versions of dbus all running at the same time if you want processes to
actually talk to each other.  You obviously can only run one kernel at
a time, and there is only one port 80/25/etc on the system unless you
run every process in its own namespace.

I don't see how you'd check for vulnerable libraries other than using
heuristics (ie actually run the exploit against the library).  If
ANYBODY can create a repository, and a repository can contain any
number of builds of any number of versions of zlib, then how could you
possibly tell whether any particular build is patched for any
particular vulnerability otherwise?  Certainly you could offer a tool
that checks for known-vulnerable versions from a selection of
supported repositories.

Such a model also only promotes choice to a degree.  You can choose to
use Fred's version of libfoo, and Tony's version of java, but you
can't use them together unless tony also chooses to support this
configuration.  Tony's java could very well link to Sam's libfoo and
you don't like that version of it, but java isn't all that popular
these days so beggars can't be choosers.

With Gentoo you can mix/match dependencies but the same thing that
allows that to work will also cause the system to break horribly when
somebody changes an ABI without changing an SONAME and so on.

>
> My point was: it's not enough talking about "what should we focus on?".
> How do you implement that focus?

I do agree with this, which is one of my concerns with the original
post.  We can certainly talk about what we should focus on, but the
only thing as a Council we really have the power to do is to forbid
people from working on other things, or to expel hot air.  The latter
can accomplish some good, but not much.

>
> What's your idea there, except of keeping status quo which clearly does
> not work since even gentoo oldtimers seem to realize something is going
> wrong?
>

If I had solutions I'd be posting them.  Apologies if it seems like
I'm nitpicking.  The thing is that while the status quo has some clear
deficiencies, in some ways it is also the best Gentoo has really ever
been.

Or put another way, where else are we going to go?  Maybe that is why
so many of us that seem to have so much conflict with each other all
stick around.

-- 
Rich

Re: [gentoo-project] Some focus for Gentoo

[Seemant Kulleen]

[-- Attachment #1: Type: text/plain, Size: 3239 bytes --]

Please understand my context: I absolutely love Gentoo Linux -- it's been
that way since some March day in early 2001.

On 21 January 2015 at 12:44, Rich Freeman <rich0@gentoo.org> wrote:

>
> I'm not saying that anything bundles libs... SONAME, etc).
>
<snipped unnecessary technical tangent.>


> I don't see how you'd check for vulnerable libraries other than using
> heuristics
>
<snipped unnecessary technical supposition>.  T



> Such a model also only promotes choice to a degree.
>
<snipped redundant argument about choice -- not different from current
status quo>


> <snipped bginning of this> but the
> only thing as a Council we really have the power to do is to forbid
> people from working on other things, or to expel hot air.  The latter
> can accomplish some good, but not much.
>

The Council's power is irrelevant this early in the discussion.
No offense intended, but I see too many posts talking about Council's
limited powers, and the fact that the Council can really only talk, but all
that is fodder for a different thread.

I would absolutely LOVE IT if the Council (and its members) would refrain
from further expellations of hot air.  Silence is not only more preferable,
but more factually correct in such situations.


> If I had solutions I'd be posting them.  Apologies if it seems like
> I'm nitpicking.  The thing is that while the status quo has some clear
> deficiencies, in some ways it is also the best Gentoo has really ever
> been.
>

It's not the nitpicking as such, but the irrelevant tangents that make for
difficult to consume wall-of-text emails.

And no, it is NOT the best that Gentoo has ever been.  Gentoo has yet to
actually be the best it can be.
That requires -- drum roll -- FOCUS!

Gentoo is bloated and slow and takes forever and is full of fighting on the
mailing lists.  That last item is the only thing that's remained consistent.


> Or put another way, where else are we going to go?  Maybe that is why
> so many of us that seem to have so much conflict with each other all
> stick around.
>

Not sure what this is about (I almost snipped it).

As someone who only ever uses Gentoo Linux, the thing I want to know from
the developer community is: what exactly are you building?

It's not a choice machine.  Gentoo hasn't ever been a choice machine. I've
been seeing variations of "Gentoo is about choice. Right?  Right???" for a
while now.

It isn't -- it's about providing sensible options as alternatives to
sensible defaults.  The danger of stopping at "but but it's about choice!!
CHOICE!!!!!!!eleven!!"  is that the shit will not stop at that point.  Any
arbitrary damned thing becomes a choice (USE flag bloat makes those flags
basically useless, for example).

Here's what hasn't changed in the Gentoo experience: the emotion of
fulfillment, achievement and satisfaction after that initial install -- the
install you really care about.

That feeling, that "je ne sais quoi" (I dunno what) -- that emotion is
unique to the Gentoo experience.

Instead of endless technical upmanships in these posts, I'd prefer to see
the developer and leader community focus on the Gentoo experience as a
product, and figure out how to maximise delivery of the Gentoogasm.

Cheers,
Seemant

[-- Attachment #2: Type: text/html, Size: 4907 bytes --]

Re: [gentoo-project] Some focus for Gentoo

[Wulf C. Krueger]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 21.01.2015 17:36, Julian wrote:
> Unfortunately they don't attract people with their arrogance

The word you're looking for is "honesty" and I can't complain about a
lack of people:

https://galileo.mailstation.de/egitstats/authors.html#list_of_authors

> (it's all over their docs)

Our docs were mostly written by contributors...

> and they haven't understood that there is a difference between 
> distributed and fragmented.

Weren't you earlier mentioning arrogance, pot?

- -- 
Best regards, Wulf

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2

iEYEARECAAYFAlTBPcEACgkQnuVXRcSi+5rJdQCg1gc1tLT31nSSJ9Lsx0zlkApt
j/EAn12hIlWy11fA1rr8DlCnkgjlUlE/
=+hT5
-----END PGP SIGNATURE-----

Re: [gentoo-project] Some focus for Gentoo

[Rich Freeman]

On Tue, Jan 13, 2015 at 10:43 PM, Donnie Berkholz <dberkholz@gentoo.org> wrote:
> Wanted to share my thoughts on where I think Gentoo should go, in terms
> of direction. Would love to hear your thoughts.

So, I think that remaining viable doesn't necessarily require focusing
on some goals to the exclusion of all others.

I certainly agree that Gentoo is a great choice for users who need
extreme flexibility.  However, isn't that just "blathering" about
Gentoo being about choice?  The very thing that makes Gentoo so
suitable for people who want to depart from the beaten path is that we
DON'T try to focus on any one thing.

I may or may not be a "typical" Gentoo user, but I think the thing
that really keeps me around here is that I'm a tinkerer, and Gentoo
seems to be full of tinkerers.  This fits into your "people who want
to learn how Linux works" and "people who need extreme flexibility"
categories, but I think it goes well beyond that.  If I wanted to run
a specific configuration in a production setting there is almost
certainly some other distro more tailored to it than Gentoo.  However,
if I want to dabble in this for a few months, and that for a few
months, and generally dabble in those things before it is popular to
do so, then there is a pretty good chance that somebody is already
doing it with Gentoo, and if not chances are that if I am the first
others will be interested in using my contributions.

Sure, having a working toolchain is very helpful when you're blazing
ground, but if all I wanted to do is build abc I could easily set up a
container/VM/whatever following the guidelines of the folks who work
on abc (whether that is a project/language/platform/whatever).  Linux
makes it pretty easy to run things in specialized environments when
you need to, and without worrying about dumping garbage in /usr when
every language wants to bundle its own package manager.

I guess I just have trouble envisioning what a "more focused Gentoo"
looks like other than what it looks like right now.  I guess we can
start saying no to stuff that is outside our area of focus, but is
that what we really want to be?  This isn't really a zero-sum game -
we don't have to exclude contributions for the sake of being focused.

I'd certainly value the perspective of those who have been
contributing longer.  For most of my early years I didn't really
follow -dev all that closely, so I don't really have a sense for how
Gentoo's focus at that time made it more successful.  I couldn't
really tell you what Gentoo's focus was back when I first started
using it.  I just was interested in trying out a source-based distro.

I would love to see more bleeding-edge work in Gentoo, but to be fair
we do actually get a fair bit of that.  We had systemd available very
early on, we are probably the only distro that ever supported X32, our
prefix capabilities are fairly unparalleled (other than on OSX, and
I'm not sure how well the alternatives there actually compare), and
even though my sense is that Gentoo hardened isn't quite as active I
think we provide a lot of unique capabilities there as well.  It is
pretty rare that somebody who wants to do something new is unable to
do it.  Sure, there can be resistance, but I think we manage it when
it happens as long as there is perseverance.

From my observation the only thing Gentoo really needs to make a
certain use case work is continued advocacy.  Typically a few devs
form a project to support some use case, and they advocate for it, do
work to get packages to support it, educate other maintainers as to
how to best help them out, and so on.  Not every Gentoo dev is going
to join up on it, but most will at the very least stay out of their
way, and most are going to do what they can to facilitate things.

I apologize for this being a bit disorganized/random.  I just want to
get my thoughts out there.  This isn't an area I feel particularly
strongly about.  However, I do want to make sure that we don't declare
some kinds of innovation as unwelcome.

-- 
Rich

Re: [gentoo-project] Some focus for Gentoo

[Dirkjan Ochtman]

On Wed, Jan 14, 2015 at 4:43 AM, Donnie Berkholz <dberkholz@gentoo.org> 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/

Interesting post, and I liked bits of the discussion in this thread.

As a software developer, I really like how easy Gentoo makes it for me
to pull in bits I need for any kind of project I might come up with.
On the other hand, I think we suffer from the competition of
language-specific packaging solutions (e.g. NPM, pip/virtualenv,
RVM/gems). Maybe containerization/Docker will swing this back, though
I don't quite like the direction Docker takes. I'm really looking for
something even more lightweight that would fit in better with Gentoo,
e.g. some solution where I can have a container in which I can use
portage to install stuff, but depend on the @system, from the host
system. Integration between our package management and
containerization feels like it could be very powerful.

Mostly, I worry about our contribution process debt (this term was
recently coined by a guy from the Mozilla community, read
http://gregoryszorc.com/blog/2015/01/09/firefox-contribution-process-debt/
and subsequent posts for background). Given our heavy-weight
recruiting process, and CVS, and the lack of manpower in e.g. infra,
it definitely feels like we have issues here. I think that could
potentially be very important to keep Gentoo alive (and keep a fresh
inflow of new developers going).

Cheers,

Dirkjan

Re: [gentoo-project] Some focus for Gentoo

[Zac Medico]

On 01/17/2015 04:56 AM, Dirkjan Ochtman wrote:
> On Wed, Jan 14, 2015 at 4:43 AM, Donnie Berkholz <dberkholz@gentoo.org> 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/
> 
> Interesting post, and I liked bits of the discussion in this thread.
> 
> As a software developer, I really like how easy Gentoo makes it for me
> to pull in bits I need for any kind of project I might come up with.
> On the other hand, I think we suffer from the competition of
> language-specific packaging solutions (e.g. NPM, pip/virtualenv,
> RVM/gems). Maybe containerization/Docker will swing this back, though
> I don't quite like the direction Docker takes. I'm really looking for
> something even more lightweight that would fit in better with Gentoo,
> e.g. some solution where I can have a container in which I can use
> portage to install stuff, but depend on the @system, from the host
> system. Integration between our package management and
> containerization feels like it could be very powerful.

I've had some passing thoughts about this too. I think this may be a
special case of what the prefix team calls prefix-chaining [1].

> Mostly, I worry about our contribution process debt (this term was
> recently coined by a guy from the Mozilla community, read
> http://gregoryszorc.com/blog/2015/01/09/firefox-contribution-process-debt/
> and subsequent posts for background). Given our heavy-weight
> recruiting process, and CVS, and the lack of manpower in e.g. infra,
> it definitely feels like we have issues here. I think that could
> potentially be very important to keep Gentoo alive (and keep a fresh
> inflow of new developers going).

Yeah, anything we can to to reduce the barrier to contributions would be
great. Maybe using a github pull-request workflow may help with some
aspects of this.

[1] http://thread.gmane.org/gmane.linux.gentoo.alt/4821
-- 
Thanks,
Zac