Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/boost: boost-1.46.1-r1.ebuild metadata.xml boost-1.49.0-r1.ebuild boost-1.51.0-r1.ebuild ChangeLog boost-1.47.0.ebuild boost-1.35.0-r2.ebuild boost-1.47.0-r1.ebuild boost-1.39.0.ebui

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/boost: boost-1.46.1-r1.ebuild metadata.xml boost-1.49.0-r1.ebuild boost-1.51.0-r1.ebuild ChangeLog boost-1.47.0.ebuild boost-1.35.0-r2.ebuild boost-1.47.0-r1.ebuild boost-1.39.0.ebui

[Jamie Learmonth]

[-- Attachment #1: Type: text/plain, Size: 1513 bytes --]

Firstly, why are you guys always so mad, and secondly why don't we just
start packaging more of these packages as binaries then or bundling the
needed version like the rest of the world does anyways? For many packages
and systems (in this day and age of personal computing power) the advantage
of source or not bundling is negligible but mostly I am starting to get
slightly concerned about everyones health having been reading this list for
years now. Things like slotted boost and a new glibc really wind some
people up, it's not good - it's very concerning. Your health is more
important than being forced to increase your compiled binary size by a few
KB, despite the tempting memory gain, or because your package won't compile
because someone removed one of the 8 available versions of that library you
depended on. It is easier to buy more RAM than to reduce ones blood
pressure, let me tell you.

So how about we transform this distro into one where decisions are made on
real world benefits, not purist ideals, and technical giants help each
other to make breakthroughs unimaginable to others because they are
smarter, work together, and the most kick-ass geeks on this interweb. I
want to keep using Gentoo not just because of the pretty colours on the eix
output or because when I run emerge --world in a coffee shop all the chicks
think I'm that guy from the Hackers movie, or even because my favourite
color is purple. I want to keep using Gentoo because the team effing rocks!

Who's with me?

Jamie

[-- Attachment #2: Type: text/html, Size: 1629 bytes --]

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/boost: boost-1.46.1-r1.ebuild metadata.xml boost-1.49.0-r1.ebuild boost-1.51.0-r1.ebuild ChangeLog boost-1.47.0.ebuild boost-1.35.0-r2.ebuild boost-1.47.0-r1.ebuild boost-1.39.0.ebui

[Diego Elio Pettenò]

On 01/11/2012 06:50, Jamie Learmonth wrote:
> 
> Who's with me?

Not me.

If you're just looking for another binary distribution, CentOS is there.
Or Sabayon. Or Ubuntu. Pick your poison.

But I use and _develop_ Gentoo because I _want_, and sometimes just
_need_ source based! Sometiems it's because I'm doing embedded work and
I have to shave even those 4KiB, most of the time is because I want the
flexibility.

And unbundling almost never has "negligible advantages" as it's not the
_size_ or performance that we care about in those situation but security
and reliability.

Would you rather have to track down Internal Compiler Errors in two
dozens packages, or on one, even if it takes slightly longer when it
breaks compatibility? (And almost never it takes _that_ long anyway.) I
would definitely prefer the one.

-- 
Diego Elio Pettenò — Flameeyes
flameeyes@flameeyes.eu — http://blog.flameeyes.eu/

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/boost: boost-1.46.1-r1.ebuild metadata.xml boost-1.49.0-r1.ebuild boost-1.51.0-r1.ebuild ChangeLog boost-1.47.0.ebuild boost-1.35.0-r2.ebuild boost-1.47.0-r1.ebuild boost-1.39.0.ebui

[Georg Rudoy]

2012/11/1 Jamie Learmonth <jamie-lists@boxlightmedia.com>:
> Firstly, why are you guys always so mad, and secondly why don't we just
> start packaging more of these packages as binaries then or bundling the
> needed version like the rest of the world does anyways?

So are you suggesting to package all the binaries that depend upon too
old boost, or upon too new boost, or whatever, as binaries?

I always thought those few -bin packages are -bin just because they
take quite a lot time to be compiled.

> So how about we transform this distro into one where decisions are made on
> real world benefits, not purist ideals, and technical giants help each other
> to make breakthroughs unimaginable to others because they are smarter, work
> together, and the most kick-ass geeks on this interweb. I want to keep using
> Gentoo not just because of the pretty colours on the eix output or because
> when I run emerge --world in a coffee shop all the chicks think I'm that guy
> from the Hackers movie, or even because my favourite color is purple. I want
> to keep using Gentoo because the team effing rocks!
>
> Who's with me?

I hope you'd forgive I'd oppose that suggestion, given I'm not a
Gentoo dev but just a C++ developer who uses Gentoo as his primary
(and, in fact, only) distro and loves it a lot.

Gentoo has a wonderful community which I truly love, but first thing
about a distro is how it helps you to solve your problems and achieve
your goals. I'd surely miss the slotted Boost, since it was one of the
things I needed for my C++ thingies and which I loved, but Diego's
arguments are sane and reasonable, and I also had quite some PITA with
old/new/whatever Boost versions, so OK, I can live without slotted
Boost. But please, leave Gentoo as that powerful distro that really
helps to work and code. I want it source-based. I want to be able to
seamlessly have Qt 4.8.9999, or cabal integrated with Portage (kudos
to gentoo-haskell guys, they're doing great work), and things like
that.

That makes even more sense given the new C++11 standard and it's
gradual implementation in gcc. One could just as well say "let's just
keep the newest gcc 4.7 in portage, since there is software that fails
to build with gcc 4.6 and earlier", for example.

-- 
  Georg Rudoy
  LeechCraft — http://leechcraft.org

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/boost: boost-1.46.1-r1.ebuild metadata.xml boost-1.49.0-r1.ebuild boost-1.51.0-r1.ebuild ChangeLog boost-1.47.0.ebuild boost-1.35.0-r2.ebuild boost-1.47.0-r1.ebuild boost-1.39.0.ebui

[hasufell]

On 11/01/2012 02:50 PM, Jamie Learmonth wrote:
> Firstly, why are you guys always so mad, and secondly why don't we just
> start packaging more of these packages as binaries then or bundling the
> needed version like the rest of the world does anyways?

You lost faith in our ways, maybe dberkholz can help you.

watch
http://dberkholz.com/2012/07/10/gentoo-linux-or-why-in-the-world-you-should-compile-everything/
3 times

now compile gcc by hand
then chant the portage prayer and "lolcat" at LFS people (you might need
to emerge games-misc/lolcat first)

after that install ubuntu in a vbox (close all windows and lock your
door beforehand)

If you survived, chant the portage prayer.

> 
> So how about we transform this distro into one where decisions are made on
> real world benefits, not purist ideals, and technical giants help each
> other to make breakthroughs unimaginable to others because they are
> smarter, work together, and the most kick-ass geeks on this interweb. I
> want to keep using Gentoo not just because of the pretty colours on the eix
> output or because when I run emerge --world in a coffee shop all the chicks
> think I'm that guy from the Hackers movie, or even because my favourite
> color is purple. I want to keep using Gentoo because the team effing rocks!
> 

How did it turn out with the chicks?

> Who's with me?

I will forgive you... this time.

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/boost: boost-1.46.1-r1.ebuild metadata.xml boost-1.49.0-r1.ebuild boost-1.51.0-r1.ebuild ChangeLog boost-1.47.0.ebuild boost-1.35.0-r2.ebuild boost-1.47.0-r1.ebuild boost-1.39.0.ebui

[Ian Stakenvicius]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 01/11/12 10:10 AM, Georg Rudoy wrote:
> 2012/11/1 Jamie Learmonth <jamie-lists@boxlightmedia.com>:
>> Firstly, why are you guys always so mad, and secondly why don't
>> we just start packaging more of these packages as binaries then
>> or bundling the needed version like the rest of the world does
>> anyways?
> 
> So are you suggesting to package all the binaries that depend upon
> too old boost, or upon too new boost, or whatever, as binaries?
> 
> I always thought those few -bin packages are -bin just because
> they take quite a lot time to be compiled.
> 

They are.

This idea wouldn't work tho -- providing the old boost as binaries
isn't actually going to help things, unless they are fully static, as
it's the breakage against the toolchain that invalidates them
(otherwise it wouldn't be an issue to leave 'em in the tree and for
that matter leave boost slotted and have all rdeps just depend on the
slot they were written for).  And fully static binary packages are
just plain wrong on any number of levels for something like this imo.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)

iF4EAREIAAYFAlCSixYACgkQ2ugaI38ACPB72AD9EUYVEovDTDkHBmURJ3XGWt7Z
EdPNP7F5k46lZAM6LscA/0rO3wjaVfBZDwKi88kX6NL3nWEUgpDxmNASrN42xs+O
=KC5a
-----END PGP SIGNATURE-----

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/boost: boost-1.46.1-r1.ebuild metadata.xml boost-1.49.0-r1.ebuild boost-1.51.0-r1.ebuild ChangeLog boost-1.47.0.ebuild boost-1.35.0-r2.ebuild boost-1.47.0-r1.ebuild boost-1.39.0.ebui

[Georg Rudoy]

2012/11/1 Ian Stakenvicius <axs@gentoo.org>:
> On 01/11/12 10:10 AM, Georg Rudoy wrote:
>> 2012/11/1 Jamie Learmonth <jamie-lists@boxlightmedia.com>:
> This idea wouldn't work tho -- providing the old boost as binaries
> isn't actually going to help things, unless they are fully static, as
> it's the breakage against the toolchain that invalidates them
> (otherwise it wouldn't be an issue to leave 'em in the tree and for
> that matter leave boost slotted and have all rdeps just depend on the
> slot they were written for).  And fully static binary packages are
> just plain wrong on any number of levels for something like this imo.

Moreover, one may run into serious runtime troubles if several copies
of one static library are loaded into the same address space. So, this
cannot be decided for all packages IMO, rather on per-package basis
with opt-in strategy.

-- 
  Georg Rudoy
  LeechCraft — http://leechcraft.org

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-libs/boost: boost-1.46.1-r1.ebuild metadata.xml boost-1.49.0-r1.ebuild boost-1.51.0-r1.ebuild ChangeLog boost-1.47.0.ebuild boost-1.35.0-r2.ebuild boost-1.47.0-r1.ebuild boost-1.39.0.ebui

[Ben de Groot]

On 1 November 2012 21:50, Jamie Learmonth <jamie-lists@boxlightmedia.com> wrote:
> Firstly, why are you guys always so mad,

We're not, honestly. But you do see some of the worst we can do on
this list (which is why some people tend to ignore this list). Altho
on the other hand, some of the best cooperation happens here too.

> For many packages
> and systems (in this day and age of personal computing power) the advantage
> of source or not bundling is negligible

So go use a binary distro. You're always welcome to come back when you
find out that building from source and not bundling libs does have
significant advantages.

> So how about we transform this distro into one where decisions are made on
> real world benefits, not purist ideals, and technical giants help each other
> to make breakthroughs unimaginable to others because they are smarter, work
> together, and the most kick-ass geeks on this interweb.

That is exactly what we are trying to do. But people (being human)
don't always agree on how to do that, and which benefits have more
priority. So sometimes there is a bit of rough and tumble before we
get things sorted out.

> I want to keep using Gentoo because the team effing rocks!

Just say 'fucking' like a real man, or don't say it at all...

-- 
Cheers,

Ben | yngwin
Gentoo developer
Gentoo Qt project lead, Gentoo Wiki admin