Mailing List Archive

I would say, yes. This is a very important for some dialup users. I personally
ditched the default way on my system because I couldn't do this.

With regard to the concerns leveled at users defining their own scripts, it
shouldn't cause too much trouble as long as the user knows a bit about what
he is doing.

I do have one question though, how will this affect programs like pppconfig?
Programs such as these may not look at these scripts and break. Pretty much
my one worry is about breaking compatibility with other programs in the tree.

--Thomas

P.S. Since this is a dialup type question, how many devs have done dev work
over 56K dialup net connections?
On Sunday 26 August 2007 02:05:08 Alin Năstac wrote:
> A gentoo user requested in bug 190143 [1] to change the way pppd deals
> with interface up/down events. He requested to break current
> ip-up/ip-down functionality into different scripts contained in
> /etc/ppp/(ip-up.d,ip-down.d}.
>
> What do you think about? Is it worth it?
>
> Personally I think it is a good idea, but I have reserves when it comes
> to user defined {ip-up,ip-down}.local scripts. IMO the best way to solve
> this is to add this code to pkg_postinst():
> if [ -f "${ROOT}"etc/ppp/ip-up.local ]; then
> mv "${ROOT}"etc/ppp/ip-up.local "${ROOT}"etc/ppp/ip-up.d/99-local
> fi
>
> [1] http://bugs.gentoo.org/show_bug.cgi?id=190143



--
2.6.22-gentoo-r2
--
gentoo-dev@gentoo.org mailing list

Thomas Anderson wrote:
> I do have one question though, how will this affect programs like pppconfig?
> Programs such as these may not look at these scripts and break. Pretty much
> my one worry is about breaking compatibility with other programs in the tree.
>
pppconfig do not care what the /etc/ppp/ip-up script is doing, nor any
other package AFAIK.

The current scripts handle /etc/resolv.conf generation in our own way.
From this point of view, it wouldn't matter if you use a dialer program,
baselayout PPP support or run pppd by yourself.

Alin Năstac wrote:
> A gentoo user requested in bug 190143 [1] to change the way pppd deals
> with interface up/down events. He requested to break current
> ip-up/ip-down functionality into different scripts contained in
> /etc/ppp/(ip-up.d,ip-down.d}.
>
> What do you think about? Is it worth it?
>
> Personally I think it is a good idea, but I have reserves when it comes
> to user defined {ip-up,ip-down}.local scripts. IMO the best way to solve
> this is to add this code to pkg_postinst():
> if [ -f "${ROOT}"etc/ppp/ip-up.local ]; then
> mv "${ROOT}"etc/ppp/ip-up.local "${ROOT}"etc/ppp/ip-up.d/99-local
> fi
>
> [1] http://bugs.gentoo.org/show_bug.cgi?id=190143

If anything does actually change, be sure to open a docs-team bug with
a patch or list of changes against what the networking handbook(s)
currently have for PPP. It'll need to be documented all over again.

>From a users point of view - dont do it.

Stay simple and standard (i.e., equal to upstream as per the gentoo
philosophy).

Its easier for users to understand (and find documentation for) and less
likely to break.

Expert users will make changes anyway and they have the knowledge to do
so - its almost impossible to second guess what everyone is going to do
- the experts will do their own thing anyway as no matter what you do,
it wont be suitable for everyone. If you make the configuration more
complex so its "experts only" to understand it, its not going to do
anyone a favour.

This proliferation of non-standard configuration files
(i.e., /etc/conf.d) that gentoo does, and the changes that keep
happening to them is one of the hardest things for a user to keep up
with - for an example of the problems this causes, see the networking
scripts! (constant changes, broken networking etc - yes they are now
better at this point in time, but I wonder when the next change will
occur that will unexpectedly kill my remote systems ...)

Billk


On Sun, 2007-08-26 at 09:05 +0300, Alin Nãstac wrote:
> A gentoo user requested in bug 190143 [1] to change the way pppd deals
> with interface up/down events. He requested to break current
> ip-up/ip-down functionality into different scripts contained in
> /etc/ppp/(ip-up.d,ip-down.d}.
>
> What do you think about? Is it worth it?
>
> Personally I think it is a good idea, but I have reserves when it comes
> to user defined {ip-up,ip-down}.local scripts. IMO the best way to solve
> this is to add this code to pkg_postinst():
> if [ -f "${ROOT}"etc/ppp/ip-up.local ]; then
> mv "${ROOT}"etc/ppp/ip-up.local "${ROOT}"etc/ppp/ip-up.d/99-local
> fi
>
> [1] http://bugs.gentoo.org/show_bug.cgi?id=190143
>
--
gentoo-dev@gentoo.org mailing list

W.Kenworthy wrote:
> >From a users point of view - dont do it.
>
> Stay simple and standard (i.e., equal to upstream as per the gentoo
> philosophy).
>
I don't see it:
a) The current ip-up & ip-down scripts are provided by us, not by upstream.
b) If this is going to be implemented, it will be done by altering our
ip-* scripts, not by patching the pppd.
c) Users were not supposed to modify these scripts, but make their own
ip-*.local scripts. If these scripts exists, they are going to be
automatically moved into ip-*.d directories.

W.Kenworthy wrote:
>>From a users point of view - dont do it.
>
Firstly, can I ask you not to top-post? It makes it harder to follow the
flow of conversations.
> Stay simple and standard (i.e., equal to upstream as per the gentoo
> philosophy).
>
It's been years since I worked with a dial-up Linux box, but iirc RH5 had
the capability to hook scripts into the ppp sequence (it's how I got my
connection going ;)

> Its easier for users to understand (and find documentation for) and less
> likely to break.
>
Sure, but here we're talking about enabling users, in the characteristically
simple and clean Gentoo fashion, to hook in functionality they need.
Documentation has always been one of Gentoo's strong points (as you'll have
noticed one of the doc team already asked that any changes be reflected in
web docs.) Quite apart from which, the config files are always
well-commented, since they are designed to be edited by humans. Some of
them are mini-tutorials (eg make.conf.example.)

> Expert users will make changes anyway and they have the knowledge to do
> so - its almost impossible to second guess what everyone is going to do
> - the experts will do their own thing anyway as no matter what you do,
> it wont be suitable for everyone. If you make the configuration more
> complex so its "experts only" to understand it, its not going to do
> anyone a favour.
>
Sure, but this isn't complex, it's just one level of metadata. As such it
enables a whole lot while still being simple to understand. As for
experts-only, ime with dial-up, any user who sets such a connection up has
to edit a file or two. I didn't get anywhere with dial-up guis in RH6, so I
admit my knowledge is out-dated, but the stuff I did by hand was simple in
comparison to other Gentoo stuff. There's no point pretending you can set
up and administer a Gentoo box without being willing to look at text files;
a generic user may be able to get on fine with just a GUI, but someone's
got to maintain the box (even if remotely.)

To my mind the changes proposed would make it simpler to script around PPP
which would make the sysadmin/developer's job a lot easier, and thus the
end-user experience much smoother.

> This proliferation of non-standard configuration files
> (i.e., /etc/conf.d) that gentoo does, and the changes that keep
> happening to them is one of the hardest things for a user to keep up
> with - for an example of the problems this causes, see the networking
> scripts! (constant changes, broken networking etc - yes they are now
> better at this point in time, but I wonder when the next change will
> occur that will unexpectedly kill my remote systems ...)
>
I've never had to mess with any networking voodoo (we're not allowed WIFI
where I work in any case) so I can't really comment. I would say that if
you see changes of the type you mention, and you have some constructive
input (such as how to upgrade, or even just warning others) forum posts
would be welcomed. Patches to docs and scripts on bugzilla are even more
welcome.


--
gentoo-dev@gentoo.org mailing list