[PATCH] Include netinet/in.h in libbb.h

Bernhard Fischer rep.nop at aon.at
Mon Jun 5 03:35:53 PDT 2006


On Sun, Jun 04, 2006 at 06:22:26AM -0400, Robert P. J. Day wrote:
>On Sun, 4 Jun 2006, Bernhard Fischer wrote:
>
>> On Sat, Jun 03, 2006 at 06:46:06PM -0400, Robert P. J. Day wrote:
>> >On Sat, 3 Jun 2006, Bernhard Fischer wrote:
>> >
>> >> On Sat, Jun 03, 2006 at 12:25:36PM -0600, Shaun Jackman wrote:
>> >> >On 7/25/05, Rob Landley <rob at landley.net> wrote:
>> >> >...
>> >> >>But applied anyway...
>> >> >>
>> >> >>Rob
>> >> >
>> >> >This patch was never applied; please apply it now.
>> >> >
>> >> >libbb.h uses struct sockaddr_in which is defined in netinet/in.h but
>> >> >never included.
>> >>
>> >> This patch shouldn't be needed anymore since the order of includes
>> >> were restructered recently. Applets have to include busybox.h as the
>> >> first include, lib*/* has to include libbb.h as the first include.
>> >>
>> >> Applets and library functions MUST not include any header which is
>> >> included by platform.h, busybox.h and platform.h, libbb.h
>> >> respectively.
>> >
>> >all right, this is just plain creeping me out, when people start
>> >talking about a *required order* of header file inclusion.  this is
>> >just plain not good.
>>
>> go understand how some aspects of dmalloc work: http://www.dmalloc.com/
>
>i'm not sure i see your point.  you're referring to a specific example
>of how an enhanced version of some functionality can be used in a
>program.
>
>*i'm* talking about the apparent need to enforce an ordering scheme on
>the all-encompassing header files being used for busybox overall.  i

Not sure if i understand what you mean.

>don't think those two situations are comparable.

Our platform.h serves (umong other things) as a "config.h" so we don't
get errors when an applet does #include <macos/snafu.h> which will most
likely not exist on your cygwin installation nor on your linux
installation. See?


More information about the busybox mailing list