[gentoo-dev] rfc: oldnet scripts splitting out from OpenRC

[William Hubbs]

All,

it has been suggested that gentoo's oldnet scripts be split out into
their own package separate from OpenRC so that they can be developed
independently. I am looking at doing this for OpenRC 0.12, which I hope
to release soon.

This means when you emerge or upgrade to openrc-0.12, the net.* scripts
will no longer be included. I am going to call the separate package that
includes these scripts gentoo-oldnet.

My plan is to write a newsitem when OpenRC 0.12 is released
telling users this and that they will have to emerge gentoo-oldnet to
get the gentoo networking scripts or turn on the newnet (maybe I'll
change this to net) use flag to get OpenRC's network scripts installed
and put ewarns in the ebuild if this use flag is turned off.

I feel that a newsitem and ewarns in the OpenRC ebuild cover live
systems well. In a nutshell, users should pay attention to their news
items and ewarns.

On the other hand, some are suggesting that I should add a runtime
dependency to OpenRC so that it pulls in gentoo-oldnet. Since OpenRC
doesn't need gentoo-oldnet in order to run, I feel like this would be
abusing dependencies.

Does anyone have any thoughts?

William

[Diego Elio Pettenò]

Please don't do this. It sounds to me like a stupid move, what are you
trying to accomplish? You already use Git/GitHub...
Diego Elio Pettenò — Flameeyes
flam...@flameeyes.eu — http://blog.flameeyes.eu/

[Ian Stakenvicius]

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

Current users should be able to do a -uDN and still have their
existing systems work as-is. Whether you do this via a static
dependency or one controlled by a use flag (and be sure use flag
defaults would have oldnet installed by default) is up to you.

It's completely understandable that openrc -the project- doesn't
contain oldnet, but the end result on gentoo still should, by default,
imo.


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

iF4EAREIAAYFAlF4Dy0ACgkQ2ugaI38ACPAQGwD+OEg6BV5rBEGylDr5R1cfMR80
oyE0Mwo4HWF/IEoTWk4A/jSvwkvz6mtx8Zuja/gzZogXVYHCVbVhARUXR195BGx2
=2754
-----END PGP SIGNATURE-----

[William Hubbs]

On Wed, Apr 24, 2013 at 05:45:04PM +0100, Diego Elio Pettenò wrote:
> Please don't do this. It sounds to me like a stupid move, what are you
> trying to accomplish? You already use Git/GitHub...
> Diego Elio Pettenò — Flameeyes
> flam...@flameeyes.eu — http://blog.flameeyes.eu/
>

robbat2 is the one who made the request. He wants to do a couple of
things:

1) he wants to be able to have independent oldnet releases so he can get
more features into the oldnet scripts and have his own development
cycle.

2) He is also interested in working on making the oldnet scripts able to
run under systemd.

William

[William Hubbs]

The issue is that OpenRC does not have any kind of dependency on
gentoo-oldnet at all. There will be a separate loopback script in OpenRC
so it is possible to run OpenRC on a system without the oldnet or
newnet scripts. In fact, this is a completely valid configuration.

OpenRC doesn't "link" to gentoo-oldnet in any way, so there is no
dependency.

The way I read the dev manual [1], a newsitem and postinst messages are
the way to go for somethinglike this.

William

[1] http://devmanual.gentoo.org/general-concepts/use-flags/index.html

[Mike Frysinger]

On Wednesday 24 April 2013 13:01:39 William Hubbs wrote:
> On Wed, Apr 24, 2013 at 05:45:04PM +0100, Diego Elio Pettenò wrote:
> > Please don't do this. It sounds to me like a stupid move, what are you
> > trying to accomplish? You already use Git/GitHub...
> > Diego Elio Pettenò — Flameeyes
> > flam...@flameeyes.eu — http://blog.flameeyes.eu/
>
> robbat2 is the one who made the request. He wants to do a couple of
> things:
>
> 1) he wants to be able to have independent oldnet releases so he can get
> more features into the oldnet scripts and have his own development
> cycle.

so release openrc more often

> 2) He is also interested in working on making the oldnet scripts able to
> run under systemd.

i don't think it's unreasonable to wait and see if this actually happens
(assuming it's even worthwhile). create a branch in the openrc git repo and
do all the work in there.
-mike

[Mike Frysinger]

On Wednesday 24 April 2013 13:23:23 William Hubbs wrote:
> On Wed, Apr 24, 2013 at 12:58:21PM -0400, Ian Stakenvicius wrote:
> > On 24/04/13 12:16 PM, William Hubbs wrote:
> > > it has been suggested that gentoo's oldnet scripts be split out
> > > into their own package separate from OpenRC so that they can be
> > > developed independently. I am looking at doing this for OpenRC
> > > 0.12, which I hope to release soon.
> > >
> > > This means when you emerge or upgrade to openrc-0.12, the net.*
> > > scripts will no longer be included. I am going to call the separate
> > > package that includes these scripts gentoo-oldnet.
> > >
> > > My plan is to write a newsitem when OpenRC 0.12 is released
> > > telling users this and that they will have to emerge gentoo-oldnet
> > > to get the gentoo networking scripts or turn on the newnet (maybe
> > > I'll change this to net) use flag to get OpenRC's network scripts
> > > installed and put ewarns in the ebuild if this use flag is turned
> > > off.
> > >
> > > I feel that a newsitem and ewarns in the OpenRC ebuild cover live
> > > systems well. In a nutshell, users should pay attention to their
> > > news items and ewarns.
> > >
> > > On the other hand, some are suggesting that I should add a runtime
> > > dependency to OpenRC so that it pulls in gentoo-oldnet. Since
> > > OpenRC doesn't need gentoo-oldnet in order to run, I feel like this
> > > would be abusing dependencies.
> > >
> > > Does anyone have any thoughts?
> >

> > Current users should be able to do a -uDN and still have their
> > existing systems work as-is. Whether you do this via a static
> > dependency or one controlled by a use flag (and be sure use flag
> > defaults would have oldnet installed by default) is up to you.
>
> The issue is that OpenRC does not have any kind of dependency on
> gentoo-oldnet at all. There will be a separate loopback script in OpenRC
> so it is possible to run OpenRC on a system without the oldnet or
> newnet scripts. In fact, this is a completely valid configuration.
>
> OpenRC doesn't "link" to gentoo-oldnet in any way, so there is no
> dependency.
>
> The way I read the dev manual [1], a newsitem and postinst messages are
> the way to go for somethinglike this.

it is reasonable to expect openrc updates to *not* break a system. that means
people shouldn't be required to read a news/postinst message to keep from
killing things.

even then, a default Gentoo system should have networking support available by
default. our manuals assume this, and people shouldn't have to install a
stage3 and then do `emerge gentoo-oldnet` just to have that happen. so
keeping a dependency in openrc (perhaps initially hard, or behind
IUSE=+oldnet) makes sense.
-mike

[William Hubbs]

I was planning on opening a bug before all of this hit stable to have
releng add gentoo-oldnet to the stages when it does hit stable.

if we keep a dependency for a while, even behind something like
IUSE="+oldnet", when we drop it, people will still be hit if they do
emerge --depclean before they emerge gentoo-oldnet.

Also, (although I don't really care about this much because we tell people
not to do it), folks who set USE="-* foo bar bas" in their make.conf would
get hit immediately with this solution.

Am I missing something? It just seems like this is putting off the
emerge command people will need to run for a while.

William

[Rich Freeman]

On Wed, Apr 24, 2013 at 1:54 PM, William Hubbs <will...@gentoo.org> wrote:
> if we keep a dependency for a while, even behind something like
> IUSE="+oldnet", when we drop it, people will still be hit if they do
> emerge --depclean before they emerge gentoo-oldnet.
>
> Also, (although I don't really care about this much because we tell people
> not to do it), folks who set USE="-* foo bar bas" in their make.conf would
> get hit immediately with this solution.
>
> Am I missing something? It just seems like this is putting off the
> emerge command people will need to run for a while.

It seems to me like network support needs to be some kind of default
for Gentoo. A USE default is probably one of the least intrusive ways
to do that - and we can always have an ewarn in the event both are
suppressed (those who override all USE defaults should be on the
lookout for trouble).

Considering our default configuration ships sshd (an argument we don't
need to rehash here), it seems a bit silly to not ship networking
support by default. I'd rather not do it as part of the system set,
though that would be consistent with what we're doing with ssh, and it
is still override-able.

Rich

[William Hubbs]

If we did this as part of @system, it would have to be a virtual imo,
since there are several things in the tree that can manage networks
(openrc[newnet], dhcpcd, wicd, nm, etc), but this is a topic for another
thread.

For OpenRC-0.12, I will put in the +oldnet use flag as suggested
previously if we go ahead with the split.

William

[Mike Frysinger]

On Wednesday 24 April 2013 13:54:07 William Hubbs wrote:
> On Wed, Apr 24, 2013 at 01:34:36PM -0400, Mike Frysinger wrote:
> > On Wednesday 24 April 2013 13:23:23 William Hubbs wrote:
> > > The issue is that OpenRC does not have any kind of dependency on
> > > gentoo-oldnet at all. There will be a separate loopback script in
> > > OpenRC so it is possible to run OpenRC on a system without the oldnet
> > > or newnet scripts. In fact, this is a completely valid configuration.
> > >
> > > OpenRC doesn't "link" to gentoo-oldnet in any way, so there is no
> > > dependency.
> > >
> > > The way I read the dev manual [1], a newsitem and postinst messages are
> > > the way to go for somethinglike this.
> >
> > it is reasonable to expect openrc updates to *not* break a system. that
> > means people shouldn't be required to read a news/postinst message to
> > keep from killing things.
> >
> > even then, a default Gentoo system should have networking support
> > available by default. our manuals assume this, and people shouldn't
> > have to install a stage3 and then do `emerge gentoo-oldnet` just to have
> > that happen. so keeping a dependency in openrc (perhaps initially hard,
> > or behind IUSE=+oldnet) makes sense.
>
> I was planning on opening a bug before all of this hit stable to have
> releng add gentoo-oldnet to the stages when it does hit stable.

"adding to the stages" isn't a magic incantation :). it's either part of
@system, or it's a dependency in a package that is part of @system. i don't
think adding it straight to @system makes sense, and i don't think there's
really an existing dependency other than openrc where this would be
appropriate.

> if we keep a dependency for a while, even behind something like
> IUSE="+oldnet", when we drop it, people will still be hit if they do
> emerge --depclean before they emerge gentoo-oldnet.

i don't think we should drop it. openrc is logically the best place imo.

i understand your position that openrc works just fine w/out these scripts.
but i don't think that pure logical distinction is really necessary here. if
people really truly don't want it, they have a USE flag to turn it off.
-mike

[Michał Górny]

On Wed, 24 Apr 2013 11:16:06 -0500
William Hubbs <will...@gentoo.org> wrote:

> This means when you emerge or upgrade to openrc-0.12, the net.* scripts
> will no longer be included. I am going to call the separate package that
> includes these scripts gentoo-oldnet.

Aside all the other, please don't name it like this :). It's just feels
wrong to start new and supposedly beneficial project and name it 'old'
like something you just thrown away off the main tree.

--
Best regards,
Michał Górny

[William Hubbs]

The name is also per robbat2's request. I asked him about other names,
but he specifically wanted oldnet in the name.

A little bit of historical background may be in order here.

In a nutshell, it is called old because Roy wanted to deprecate the
whole thing eventually and switch us over to the newnet scripts that
OpenRC has.
name.

We thought about killing off newnet entirely for a while in OpenRC, but
I have since found that people do use it. It is more similar to what
happens on the *bsd side, and it works well for simple setups.

Also, I think it is more like what some other distros do for their
network interfaces.

The primary disadvantages of newnet are that services can't depend on a
single network interface, and it is not possible to stop/start a single
interface.

William

[Diego Elio Pettenò]

On 24/04/2013 18:01, William Hubbs wrote:
> robbat2 is the one who made the request. He wants to do a couple of
> things:
>
> 1) he wants to be able to have independent oldnet releases so he can get
> more features into the oldnet scripts and have his own development
> cycle.
>
> 2) He is also interested in working on making the oldnet scripts able to
> run under systemd.

This gives it a bit more context than you originally did...

But still I don't think this is a very good move, I'll leave Mike to
follow up on it as I basically agree with his line of thought.

--

[Patrick McLean]

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

On 24/04/13 11:46 AM, Mike Frysinger wrote:
> On Wednesday 24 April 2013 13:54:07 William Hubbs wrote:
>> On Wed, Apr 24, 2013 at 01:34:36PM -0400, Mike Frysinger wrote:
>>> On Wednesday 24 April 2013 13:23:23 William Hubbs wrote:
>>>> The issue is that OpenRC does not have any kind of dependency on gentoo-oldnet at all. There will be a separate loopback script in OpenRC
>>>> so it is possible to run OpenRC on a system without the oldnet or newnet scripts. In fact, this is a completely valid configuration.
>>>

>>> even then, a default Gentoo system should have networking support available by default. our manuals assume this, and people shouldn't have
>>> to install a stage3 and then do `emerge gentoo-oldnet` just to have that happen. so keeping a dependency in openrc (perhaps initially hard,
>>> or behind IUSE=+oldnet) makes sense.
>>

+1
We should definitely install networking by default, and not make people
install it as a separate emerge. If oldnet is being split into a different
package, perhaps newnet should be split out as well.

Maybe we could have a virtual/networking that defaults to oldnet and exists
as a hard dep of openrc.

>> if we keep a dependency for a while, even behind something like IUSE="+oldnet", when we drop it, people will still be hit if they do emerge
>> --depclean before they emerge gentoo-oldnet.
>
> i don't think we should drop it. openrc is logically the best place imo.

+1
oldnet is and should remain the default networking implementation on Gentoo.
Newnet simply drops way too much functionality to become the default.

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

Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
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=iEze
-----END PGP SIGNATURE-----

[Walter Dnes]

On Wed, Apr 24, 2013 at 02:16:51PM -0400, Rich Freeman wrote

> Considering our default configuration ships sshd (an argument we don't
> need to rehash here), it seems a bit silly to not ship networking
> support by default. I'd rather not do it as part of the system set,
> though that would be consistent with what we're doing with ssh, and it
> is still override-able.

We're talking two different items here...
1) Necessary functionality for the minimal-install ISO to work
2) The installed software on a running system

Item 1) is constrained by the real world... do what you have to do to
create a functional install ISO

Item 2) will be due to customizations done by admins for their
specific needs. There's no way to predict how Joe Admin is going to
integrate the various available software services.

To handle the various possible cases, maybe we need a "virtual/net" as
part of the system set, which can be satisfied by either oldnet or
newnet or whatever. The install ISO will have a basic working network
stack (IPV4+IPV6). After the initial install, the admin can do
whatever. Maybe even invoke package.provided.

--
Walter Dnes <walt...@waltdnes.org>
I don't run "desktop environments"; I run useful applications

[William Hubbs]

On Wed, Apr 24, 2013 at 06:34:46PM -0400, Walter Dnes wrote:
> On Wed, Apr 24, 2013 at 02:16:51PM -0400, Rich Freeman wrote
>
> > Considering our default configuration ships sshd (an argument we don't
> > need to rehash here), it seems a bit silly to not ship networking
> > support by default. I'd rather not do it as part of the system set,
> > though that would be consistent with what we're doing with ssh, and it
> > is still override-able.

> To handle the various possible cases, maybe we need a "virtual/net" as
> part of the system set, which can be satisfied by either oldnet or
> newnet or whatever. The install ISO will have a basic working network
> stack (IPV4+IPV6). After the initial install, the admin can do
> whatever. Maybe even invoke package.provided.

This would actually be cleaner than a bogus dependency in OpenRC.
I would probably call it virtual/network-manager though.

Are there any issues with putting together a virtual like this and
adding it to @system?

Thanks,

William

[Rick "Zero_Chaos" Farina]

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

On 04/24/2013 07:17 PM, William Hubbs wrote:
> On Wed, Apr 24, 2013 at 06:34:46PM -0400, Walter Dnes wrote:
>> On Wed, Apr 24, 2013 at 02:16:51PM -0400, Rich Freeman wrote
>>
>>> Considering our default configuration ships sshd (an argument we don't
>>> need to rehash here), it seems a bit silly to not ship networking
>>> support by default. I'd rather not do it as part of the system set,
>>> though that would be consistent with what we're doing with ssh, and it
>>> is still override-able.
>> To handle the various possible cases, maybe we need a "virtual/net" as
>> part of the system set, which can be satisfied by either oldnet or
>> newnet or whatever. The install ISO will have a basic working network
>> stack (IPV4+IPV6). After the initial install, the admin can do
>> whatever. Maybe even invoke package.provided.
>
> This would actually be cleaner than a bogus dependency in OpenRC.
> I would probably call it virtual/network-manager though.
>

You can't call it virtual/network-manager, that calls to mind, you know,
net-misc/networkmanager. That's just too confusing imho. I wouldn't
object to virtual/net, or pretty much anything else that isn't
confusing. The net scripts are not a network manager, networkmanager,
wicd, even wpa_supplicant would be things I would consider to be network
managers.

- -Zero

> Are there any issues with putting together a virtual like this and
> adding it to @system?
>
> Thanks,
>
> William
>

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

Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
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=5kV9
-----END PGP SIGNATURE-----

[Carlos Silva]

How about someone decide which is the best version and keep it integrated in OpenRC? There's no best version? diff the two of them and merge what matters. I mean, how many distros do you know that have two sets of init scripts *just* to configure networking? I know Gentoo is about choice, but this feels a little too much choice. What will I choose next? Someone can  decide that keymap needs a refactor and just fork it, of maybe hostname, or <insert_some_stupid_but_*needed*_init_script_here>.

Just my 2 cents...

[William Hubbs]

On Wed, Apr 24, 2013 at 07:32:44PM -0400, Rick "Zero_Chaos" Farina wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 04/24/2013 07:17 PM, William Hubbs wrote:
> > On Wed, Apr 24, 2013 at 06:34:46PM -0400, Walter Dnes wrote:
> >> On Wed, Apr 24, 2013 at 02:16:51PM -0400, Rich Freeman wrote
> >>
> >>> Considering our default configuration ships sshd (an argument we don't
> >>> need to rehash here), it seems a bit silly to not ship networking
> >>> support by default. I'd rather not do it as part of the system set,
> >>> though that would be consistent with what we're doing with ssh, and it
> >>> is still override-able.
> >> To handle the various possible cases, maybe we need a "virtual/net" as
> >> part of the system set, which can be satisfied by either oldnet or
> >> newnet or whatever. The install ISO will have a basic working network
> >> stack (IPV4+IPV6). After the initial install, the admin can do
> >> whatever. Maybe even invoke package.provided.
> >
> > This would actually be cleaner than a bogus dependency in OpenRC.
> > I would probably call it virtual/network-manager though.
> >
> You can't call it virtual/network-manager, that calls to mind, you know,
> net-misc/networkmanager. That's just too confusing imho. I wouldn't
> object to virtual/net, or pretty much anything else that isn't
> confusing. The net scripts are not a network manager, networkmanager,
> wicd, even wpa_supplicant would be things I would consider to be network
> managers.

But the oldnet scripts do run wpa_supplicant, dhcp clients, etc, for
each interface they manage.

Newnet doesn't even try that, it just manages static interfaces and
assumes that you will use a dhcp client or something like wpa_supplicant
in standalone mode to control your interfaces.

William

[Ian Stakenvicius]

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

virtual/network-init ?

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

iF4EAREIAAYFAlF4iUoACgkQ2ugaI38ACPCvSAEAgSXxzswzYmOfJGl27vtb4C4w
1RjxSoLzAIAl1htEmr4A/3q3QIghpzNlG0Fmawu3eR8ML6BRf+HWwJplIWhvM6F8
=BHp8
-----END PGP SIGNATURE-----

[Zac Medico]

On 04/24/2013 06:39 PM, Ian Stakenvicius wrote:
> On 24/04/13 09:10 PM, William Hubbs wrote:
>> On Wed, Apr 24, 2013 at 07:32:44PM -0400, Rick "Zero_Chaos" Farina
>> wrote:
>>> On 04/24/2013 07:17 PM, William Hubbs wrote:
>>>>
>>>> This would actually be cleaner than a bogus dependency in
>>>> OpenRC. I would probably call it virtual/network-manager
>>>> though.
>>>>
>>> You can't call it virtual/network-manager, that calls to mind,
>>> you know, net-misc/networkmanager. That's just too confusing
>>> imho. I wouldn't object to virtual/net, or pretty much anything
>>> else that isn't confusing. The net scripts are not a network
>>> manager, networkmanager, wicd, even wpa_supplicant would be
>>> things I would consider to be network managers.
>
>> But the oldnet scripts do run wpa_supplicant, dhcp clients, etc,
>> for each interface they manage.
>
>> Newnet doesn't even try that, it just manages static interfaces
>> and assumes that you will use a dhcp client or something like
>> wpa_supplicant in standalone mode to control your interfaces.
>
>
> virtual/network-init ?

Sounds about right. Also, rather than "gentoo-oldnet", maybe a name like
"gentoo-network-init" makes sense, since I suspect that the vast
majority of users are still using it. If newnet is split out, maybe call
it "openrc-newnet-init" or something.
--
Thanks,
Zac

[Alec Warner]

So my understanding is that WIlliam does not want to break peoples shit. He doesn't want to maintain oldnet anymore. Other people do. What he wants to do is move oldnet out of openrc. I think that means that:

openrc will no longer contain oldnet.

openrc ebuilds will need to depend on <something> to make networking work. A brief discussion with him in chat seems to imply that oldnet is still fine. So in Gentoo we could just set IUSE="+oldnet" or similar, and it would get pulled in.

Someone may need to fix up the stages to work (I thought they relied on USE="-*" which would entail missing out on oldnet here.)

 

even then, a default Gentoo system should have networking support available by
default.  our manuals assume this, and people shouldn't have to install a
stage3 and then do `emerge gentoo-oldnet` just to have that happen.  so
keeping a dependency in openrc (perhaps initially hard, or behind
IUSE=+oldnet) makes sense.
-mike

Yeah after discussing in chat, I don't think the intention was to 'get everyone off of oldnet' but simply to move it out of openrc and into dedicated packages with maintainers that care about it.

-A

[William Hubbs]

I'll be involved in maintaining oldnet, but there are others (mainly
robbat2) who want to be able to release new versions of oldnet
separately from OpenRC, and to look into porting it to systemd.

I don't have a problem with the IUSE="+oldnet" issue if no one else
does, it just seemed sort of backward to me, using a use flag to pull in
something that really isn't a dependency.

But this would mean we can forget about any virtuals.

William

> Someone may need to fix up the stages to work (I thought they relied on
> USE="-*" which would entail missing out on oldnet here.)

Yes, I was planning to get with Jorge before this hits stable and figure
out what needs to happen.

[Mike Frysinger]

you've only talked about moving out "oldnet" which means "newnet" remains in
openrc. that is technically a provider of virtual/network-init and we're back
where we started: the standard Gentoo network init scripts aren't pulled in.

what providers exactly would you see live in such a virtual ?

if we do choose to go the virtual route (i don't see value here), i don't
think the transitional phase can start there. if anything other than the
standard Gentoo network scripts are provided, then it means people will end up
with a broken system as portage won't bother installing it. network-
manager/wpa_supplicant/etc... are pretty common.
-mike

[G.Wolfe Woodbury]

I have just one question....

When "gentoo-oldscripts" is pulled from openrc

WHAT will be the default network configuration method?


Without the "standard" net scripts, many systems will break.
What is being given for network configuration?
NetworkManager? (yuck!)

--
G.Wolfe Woodbury
redw...@gmail.com

[Rich Freeman]

On Thu, Apr 25, 2013 at 2:27 PM, G.Wolfe Woodbury <redw...@gmail.com> wrote:
> I have just one question....
>
> When "gentoo-oldscripts" is pulled from openrc
>
> WHAT will be the default network configuration method?

"gentoo-oldscripts"

The intent isn't to remove these scripts (unless for some reason you
don't want them), but to just move them to a different package, and
get that package pulled in by default. We're just debating the
various mechanisms for doing this. The initial proposal was to
basically tell everybody to install it manually, which isn't that hard
to do and works just fine. The downside to it is that anybody who
misses the news loses their network on the next reboot (not that this
is anything new for anybody running udev). The leading proposal is to
just pull it in with a defaulted USE flag, and maybe issue a warning
if a user has that flag overrided.

It will take little to no work to keep networking exactly as it is on
a Gentoo system, and the main debate here is how to get to "no work"
for as many users as possible while not taking away choice.

Rich

[Mike Frysinger]

On Thursday 25 April 2013 14:27:39 G.Wolfe Woodbury wrote:
> When "gentoo-oldscripts" is pulled from openrc
>
> WHAT will be the default network configuration method?

the existing defaults will remain the same. we're just debating how to
guarantee that.
-mike

[viv...@gmail.com]

On 04/24/13 21:17, William Hubbs wrote:
> On Wed, Apr 24, 2013 at 09:00:17PM +0200, Michał Górny wrote:
>> On Wed, 24 Apr 2013 11:16:06 -0500
>> William Hubbs <will...@gentoo.org> wrote:
>>
>>> This means when you emerge or upgrade to openrc-0.12, the net.* scripts
>>> will no longer be included. I am going to call the separate package that
>>> includes these scripts gentoo-oldnet.
>> Aside all the other, please don't name it like this :). It's just feels
>> wrong to start new and supposedly beneficial project and name it 'old'
>> like something you just thrown away off the main tree.
> The name is also per robbat2's request. I asked him about other names,
> but he specifically wanted oldnet in the name.
>
> A little bit of historical background may be in order here.
>
> In a nutshell, it is called old because Roy wanted to deprecate the
> whole thing eventually and switch us over to the newnet scripts that
> OpenRC has.
> name.
>
> We thought about killing off newnet entirely for a while in OpenRC, but
> I have since found that people do use it. It is more similar to what
> happens on the *bsd side, and it works well for simple setups.

s/it works well for simple setups/it work wonderfully for very complex
setups/

It can mimic very closely the 'ip' command, making it easy to test on
commandline and just copy and paste in conf.d/net.
At the same time it benefit from a lot of howtos and tutorials written
for sys-apps/iproute2

"old"net is the best network manager out there including all major distro.

[Steven J. Long]

On Wed, Apr 24, 2013 at 01:30:25PM -0500, William Hubbs wrote:
> On Wed, Apr 24, 2013 at 02:16:51PM -0400, Rich Freeman wrote:

> > On Wed, Apr 24, 2013 at 1:54 PM, William Hubbs <will...@gentoo.org> wrote:
> > > if we keep a dependency for a while, even behind something like
> > > IUSE="+oldnet", when we drop it, people will still be hit if they do
> > > emerge --depclean before they emerge gentoo-oldnet.
> > >

> > > Also, (although I don't really care about this much because we tell people
> > > not to do it), folks who set USE="-* foo bar bas" in their make.conf would
> > > get hit immediately with this solution.
> > >
> > > Am I missing something? It just seems like this is putting off the
> > > emerge command people will need to run for a while.

I'm with the others: I don't think you ever want to 'drop' that default flag,
at least not without equivalent functionality, which seems not to be provided
by other methods at this point. An end-user can always turn it off, and people
who use -* are supposed to know what they're doing.

The name is troublesome given that we're discussing basic networking.

> > It seems to me like network support needs to be some kind of default
> > for Gentoo. A USE default is probably one of the least intrusive ways
> > to do that - and we can always have an ewarn in the event both are
> > suppressed (those who override all USE defaults should be on the
> > lookout for trouble).

> >
> > Considering our default configuration ships sshd (an argument we don't
> > need to rehash here), it seems a bit silly to not ship networking
> > support by default. I'd rather not do it as part of the system set,
> > though that would be consistent with what we're doing with ssh, and it
> > is still override-able.
>

> If we did this as part of @system, it would have to be a virtual imo,
> since there are several things in the tree that can manage networks
> (openrc[newnet], dhcpcd, wicd, nm, etc), but this is a topic for another
> thread.
>
> For OpenRC-0.12, I will put in the +oldnet use flag as suggested
> previously if we go ahead with the split.

Thanks, that sounds reasonable: one minor nitpick, though. Could you not
call it 'stdnet'? Since from all the other discussion it appears like this
is not going away soon for the vast majority of users, but simply being
maintained as another package, which makes sense. And it is the standard Gentoo
networking setup.

That way, 'newnet' is clearly a more modern variant, but no-one's disparaging
the traditional setup, which is after all, still the default.

Regards,
steveL.
--
#friendly-coders -- We're friendly, but we're not /that/ friendly ;-)

[Mike Frysinger]

On Thursday 25 April 2013 15:09:28 viv...@gmail.com wrote:
> On 04/24/13 21:17, William Hubbs wrote:
> > On Wed, Apr 24, 2013 at 09:00:17PM +0200, Michał Górny wrote:

> >> On Wed, 24 Apr 2013 11:16:06 -0500 William Hubbs wrote:
> >>> This means when you emerge or upgrade to openrc-0.12, the net.* scripts
> >>> will no longer be included. I am going to call the separate package
> >>> that includes these scripts gentoo-oldnet.
> >>
> >> Aside all the other, please don't name it like this :). It's just feels
> >> wrong to start new and supposedly beneficial project and name it 'old'
> >> like something you just thrown away off the main tree.
> >
> > The name is also per robbat2's request. I asked him about other names,
> > but he specifically wanted oldnet in the name.
> >
> > A little bit of historical background may be in order here.
> >
> > In a nutshell, it is called old because Roy wanted to deprecate the
> > whole thing eventually and switch us over to the newnet scripts that
> > OpenRC has.
> > name.
> >
> > We thought about killing off newnet entirely for a while in OpenRC, but
> > I have since found that people do use it. It is more similar to what
> > happens on the *bsd side, and it works well for simple setups.
>
> s/it works well for simple setups/it work wonderfully for very complex
> setups/
>
> It can mimic very closely the 'ip' command, making it easy to test on
> commandline and just copy and paste in conf.d/net.
> At the same time it benefit from a lot of howtos and tutorials written
> for sys-apps/iproute2
>
> "old"net is the best network manager out there including all major distro.

William is talking about newnet when he says "works well for simple setups"
-mike

[Robin H. Johnson]

Hi all,

I'm the one that had asked WilliamH about splitting out OldNet from
OpenRC, so I figured rather than respond to every single part of this
thread, I wanted to give a general response.

Misc preamble:
--------------
If you're a student considering GSoC, reading this email, and this
interests you, please submit a proposal for the future portions!

Introduction:
-------------
This is NOT intended to take oldnet away from OpenRC systems at all, but
rather to encourage growth of both parts independently.

Historically, baselayout-1 had a very tight integration between oldnet and the
rc handling, because net* was a special case for dependency handling.

However, since baselayout-2, this was no longer the case. OpenRC doesn't
actually depend on oldnet in any way. Oldnet however does depend on the rc
script handling of OpenRC.

Naming & options:
-----------------
I'm not entirely set on the oldnet name, but nothing that WilliamH and I
could come up seemed entirely right. Oldnet is certainly NOT the only
option for networking in Gentoo. You've also got newnet, netctl,
network-manager, wicd, and more that I'm probably forgetting.

Complexity:
-----------
As a part of OpenRC, oldnet is roughly ~6200 lines of code, and ~1700
lines of documentation. OpenRC as whole has ~28000 lines of code.
So about 20-25% of the codebase, and most of the documentation ;-).

Of presently open OpenRC bugs, >30% are presently oldnet.

Oldnet supports both really easy simple configurations, as well as many
very complicated setups.

I have some examples in my homedir:
http://dev.gentoo.org/~robbat2/conf.d-net/
- Multi-homing, dual Internet connection, home setup
- Multi-homing, enterprise setup
- Snippet: HE.net (Hurriance Electric) IPv6 tunnel
(this gives you drop-in TunnelBroker connectivity)

Other than the above, how much of the OpenRC codebase uses or even
references oldnet?

Here's a near complete list:
init.d/devd.in: before net.lo0
init.d/hostid.in: before devd net
init.d/ipfw.in: before net
init.d/nscd.in: use net dns ldap ypbind
init.d/rarpd.in: need net
init.d/rc-enabled.in: need localmount net
init.d/rpcbind.in: use net logger dns
init.d/staticroute.in: provide net
init.d/syslogd.in: use net newsyslog
conf.d/netmount:# example, do not set rc_need to something like "net.eth0 dhcpcd".
conf.d/netmount:# If you are using oldnet, you must list the specific net.* services you
conf.d/netmount:#rc_need="net.eth0"
conf.d/netmount:#rc_need="net.eth1 net.eth2"
conf.d/netmount:rc_need="net"
(plus other examples).

And all of these simply deal with the 'net' virtual service, lots of them are
*BSD only too.

So what of this complexity? We've already had two occurrences where there was a
bug in oldnet, but we couldn't release a new stable OpenRC version easily,
because we'd started the tree on an update for other major changes in OpenRC.
It would have been extremely useful to release a stable oldnet minor version
bump with the fix, but it wasn't possible.

Testing
-------
Testing oldnet extensively is very challenging due to the complexity.

It's impossible for a single person to have a test environment for ALL
of the variants of networking - some of the modules also have a number
of variations in usage. So what I'd like to do, is establish that ALL of
the modules still work completely, and who to contact for testing of
each portion, AND/OR a known test configuration.

On a regular basis, I dogfood:
- bonding (ifenslave/sysfs)
- bridging (brctl)
- vlan
- PPPoE
- IPv6
- APIPA
- dhcpcd
- ethtool
- ifconfig
- iproute2 (esp RPDB)
- macchanger

Other parts I can't/don't test:
- adsl
- autoipd
- br2684
- ccwgroup
- clip
- dhclient
- ifplugd
- netplugd
- ip6rd
- ip6to4
- ipppd
- iwconfig
- macnet
- macvlan
- pump
- ssidnet
- tuntap
- udhcpc
- wpa_supplicant
(i probably forgot some)

More stuff pending to be added:
(most of which have bugs, some of them I just have direct email
submissions)
- plip
- l2tp
- iw
- xfsm/ipsec
- iproute2-bridge
- teaming
- vlan gvrp/mvrp
- vrrp, bgpd, ospf (run screaming for the hills)

systemd & the future:
---------------------
<warn>
Before you read this section, I suggest you don a flameproof suit,
extinguish all potential sources of ignition, and don't take it too
hard.
</warn>

I'm not sure about the future of the core of OpenRC:
Upstart & systemd have some clear architectural benefits, despite their
implementation shortcomings (either upstream or per-distro).
The /usr merge is inevitable, as is the integration of other components
into the init system (udev, dbus, ...). What has become dis-integrated instead
is the configuration: lots of hardware ships specific udev rules with few
problems.

netctl showed that systemd control of single interfaces AND/OR default
single-interfaces is feasible. However the netctl development has never
had the flexibility of oldnet.

I would like the flexibility of oldnet to continue to exist, regardless
of whichever init system we end up on. I think the best future for that
is making oldnet independent of openrc.

I look forward to a future where I can fire up my laptop, using systemd, having
network-manager on wifi, and connect my ethernet jack into a vlan-tagged port,
and connect to a little Gentoo virtual machine in the cloud that booted up with
OpenRC and a tiny udhcpc configuration.

Thank you for your time.

--
Robin Hugh Johnson
Gentoo Linux: Developer, Trustee & Infrastructure Lead
E-Mail : rob...@gentoo.org
GnuPG FP : 11ACBA4F 4778E3F6 E4EDF38E B27B944E 34884E85

[Tobias Klausmann]

Hi!

On Wed, 24 Apr 2013, William Hubbs wrote:
> The primary disadvantages of newnet are that services can't depend on a
> single network interface, and it is not possible to stop/start a single
> interface.

Which is why it doesn;t work for my not-exactly-complex,
not-exactly-simple setup (NFS and AICCU depending on assorted
interfaces being up, which they aren't always). I need to be able
to restart individual interfaces and the services need to cycle
with them.

Regards,
Tobias

[Tobias Klausmann]

Hi!

On Thu, 25 Apr 2013, Steven J. Long wrote:
> Thanks, that sounds reasonable: one minor nitpick, though. Could you not
> call it 'stdnet'? Since from all the other discussion it appears like this
> is not going away soon for the vast majority of users, but simply being
> maintained as another package, which makes sense. And it is the standard Gentoo
> networking setup.
>
> That way, 'newnet' is clearly a more modern variant, but no-one's disparaging
> the traditional setup, which is after all, still the default.

+1 It is something that had me puzzled for quite a while. Was I
supposed to migrate? Was the current somehow broken?

I'm still not quite sure what newnet does that oldnet doesn't, or
why somebody felt it was necessary to make a new package (and no,
let's not discuss that here). Whatever it is, ideally, it would
reflected in the name(s). And package descriptions.

Regards,
Tobias

[Tobias Klausmann]

Hi!

On Fri, 26 Apr 2013, Tobias Klausmann wrote:
> I'm still not quite sure what newnet does that oldnet doesn't, or
> why somebody felt it was necessary to make a new package (and no,
> let's not discuss that here). Whatever it is, ideally, it would
> reflected in the name(s). And package descriptions.

Scratch that. After reading Rob's post, I know. TYVM.

Regards,
Tobias

[Alexander Berntsen]

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

On 26/04/13 01:49, Robin H. Johnson wrote:
> This is NOT intended to take oldnet away from OpenRC systems at
> all, but rather to encourage growth of both parts independently.

For what it's worth, I think this sounds reasonable. Especially
considering, that OpenRC doesn't actually depend on oldnet. Basically,
modularity is good...

> systemd [has] some clear architectural benefits, despite [its]
> implementation shortcomings
Dæmon reincarnation would be nice...

> The /usr merge is inevitable
And *good*[0][1].

[0]
<http://www.freedesktop.org/wiki/Software/systemd/TheCaseForTheUsrMerge>
[1] <http://freedesktop.org/wiki/Software/systemd/separate-usr-is-broken>
- --
Alexander
alex...@plaimi.net
http://plaimi.net/~alexander
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iF4EAREIAAYFAlF6Uy8ACgkQRtClrXBQc7Ub0AD8CEBj2AEhwVIskqILCkCuBn9B
iR8UrZEkYSkdJvR1UscA/2JkcF7QdxuE9juTSyfdiT9A2oweWmGb91G2MiGtWBPR
=SNvb
-----END PGP SIGNATURE-----

[Michael Mol]

I must have missed that. What post?

[Luca Barbato]

On 04/24/2013 06:16 PM, William Hubbs wrote:
> All,
>
> it has been suggested that gentoo's oldnet scripts be split out into
> their own package separate from OpenRC so that they can be developed
> independently. I am looking at doing this for OpenRC 0.12, which I hope
> to release soon.

>
> This means when you emerge or upgrade to openrc-0.12, the net.* scripts
> will no longer be included. I am going to call the separate package that
> includes these scripts gentoo-oldnet.

Sounds a BAD idea.

I'm all for fostering more usage, but if you like to split it in a
separate packet then just make the split distribution but keep
everything in the same git.

lu

[»Q«]

<http://article.gmane.org/gmane.linux.gentoo.devel/85223>

[Chí-Thanh Christopher Nguyễn]

William Hubbs schrieb:

> if we keep a dependency for a while, even behind something like
> IUSE="+oldnet", when we drop it, people will still be hit if they
> do emerge --depclean before they emerge gentoo-oldnet.

In my opinion USE flags should toggle dependencies only in
metapackages or rare exceptions.
I suggest to include gentoo-oldnet (or whatever name you choose for
it) in the @system set once it is stable, and until then be a hard
dependency of OpenRC.


Best regards,
Chí-Thanh Christopher Nguyễn

[Robin H. Johnson]

Please note that openrc itself is NOT in the @system set.

sys-apps/baselayout is, and has PDEPEND on openrc.

[Ian Stakenvicius]

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

On 26/04/13 02:12 PM, Robin H. Johnson wrote:
> On Fri, Apr 26, 2013 at 07:27:06PM +0200, Chí-Thanh Christopher
> Nguyễn wrote:
>> William Hubbs schrieb:
>>> if we keep a dependency for a while, even behind something
>>> like IUSE="+oldnet", when we drop it, people will still be hit
>>> if they do emerge --depclean before they emerge gentoo-oldnet.
>>
>> In my opinion USE flags should toggle dependencies only in
>> metapackages or rare exceptions. I suggest to include
>> gentoo-oldnet (or whatever name you choose for it) in the @system
>> set once it is stable, and until then be a hard dependency of
>> OpenRC.
> Please note that openrc itself is NOT in the @system set.
>
> sys-apps/baselayout is, and has PDEPEND on openrc.
>

Are we all OK with putting the new package in PDEPEND in baselayout
alongside the openrc dep, then? Seems that would suffice, plus would
allow users that really don't want it to package.provides it away ....

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

iF4EAREIAAYFAlF6xCAACgkQ2ugaI38ACPAX6AD/WBhfhw/jV0UjyyjCgsDStHw1
WPPRJfMpICU+wT3j8J0BAKOEU+3pWXDif9jBS8X4IA2Ccyq2q9vUAxwQS4WXdgdA
=8KK6
-----END PGP SIGNATURE-----

[William Hubbs]

On Fri, Apr 26, 2013 at 02:14:57PM -0400, Ian Stakenvicius wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
> On 26/04/13 02:12 PM, Robin H. Johnson wrote:
> > On Fri, Apr 26, 2013 at 07:27:06PM +0200, Chí-Thanh Christopher
> > Nguyễn wrote:
> >> William Hubbs schrieb:
> >>> if we keep a dependency for a while, even behind something
> >>> like IUSE="+oldnet", when we drop it, people will still be hit
> >>> if they do emerge --depclean before they emerge gentoo-oldnet.
> >>
> >> In my opinion USE flags should toggle dependencies only in
> >> metapackages or rare exceptions. I suggest to include
> >> gentoo-oldnet (or whatever name you choose for it) in the @system
> >> set once it is stable, and until then be a hard dependency of
> >> OpenRC.
> > Please note that openrc itself is NOT in the @system set.
> >
> > sys-apps/baselayout is, and has PDEPEND on openrc.
> >
>
> Are we all OK with putting the new package in PDEPEND in baselayout
> alongside the openrc dep, then? Seems that would suffice, plus would
> allow users that really don't want it to package.provides it away ....

No, because the goal is o remove the pdepend from baselayout eventually
[1].

William

[1] https://bugs.gentoo.org/show_bug.cgi?id=409385