Re: [PATCH 3/3] busybox: picking IPv6 per default is not up to the package

[Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>]

On Wed, Jun 09, 2010 at 08:18:13PM +0100, Phil Blundell wrote:
>On Wed, 2010-06-09 at 20:52 +0200, Bernhard Reutner-Fischer wrote:
>> On Wed, Jun 09, 2010 at 07:44:58PM +0100, Phil Blundell wrote:
>> >On Wed, 2010-06-09 at 19:10 +0200, Bernhard Reutner-Fischer wrote:
>> >> diff --git a/recipes/busybox/busybox-1.15.3/micro/defconfig b/recipes/busybox/busybox-1.15.3/micro/defconfig
>> >> index 8431d19..f6eaa6d 100644
>> >> --- a/recipes/busybox/busybox-1.15.3/micro/defconfig
>> >> +++ b/recipes/busybox/busybox-1.15.3/micro/defconfig
>> >> @@ -608,7 +608,7 @@ CONFIG_TIME=y
>> >>  #
>> >>  # Networking Utilities
>> >>  #
>> >> -CONFIG_FEATURE_IPV6=y
>> >> +# CONFIG_FEATURE_IPV6 is not set
>> >>  CONFIG_FEATURE_PREFER_IPV4_ADDRESS=y
>> >>  # CONFIG_VERBOSE_RESOLUTION_ERRORS is not set
>> >>  # CONFIG_ARP is not set
>> >
>> >This change doesn't look very desirable to me.  What's the reason for
>> >it?
>> 
>> As long as you don't have ipv6 in distro features somehow, ipv6 will
>> not be available. See introductory message of this series.
>
>Ah, so you're saying that ipv6 is missing from micro's DISTRO_FEATURES?

yes, like from any other distro, and like ipv4 or nls etc..

>If that's the case then it's just a bug: presumably it's been harmless
>in the past because very few packages care about it.  Irrespective of
>that, the intended behaviour is that micro should be IPv6-enabled and
>turning it off in busybox does not seem like a good idea.
>
>Actually, a quick "grep -ri ipv6" in conf/distro didn't turn up any
>matches at all

exactly. Thus my RFC for feature/dep handling heuristics.

>                so enabling IPv6 by DISTRO_FEATURES doesn't seem to be
>terribly well entrenched at the moment.  If it were to be introduced
>then I suspect it would need to be a negative property (i.e. the feature
>would be "no ipv6") in order to not break all the existing distros.

I disagree. Distros should be fixed to ask for or require a feature if
they require it.
For a nice way to flag packages with HANDLED_FEATURES, see conversation
with kergoth from today, around 18:00 UTC.