FTBFS in mock but not a real EL5 system

[T.C. Hollingsworth]

On Tue, Jan 10, 2012 at 8:49 PM, Kevin Fenzi <kevin at scrye.com> wrote:
> On Mon, 9 Jan 2012 15:40:43 -0700
> "T.C. Hollingsworth" <tchollingsworth at gmail.com> wrote:
>
>> Hi all!
>>
>> Node.js upstream recently switched from hardcoding LINUX_VERSION_CODE
>> to checking that __NR_accept4 is defined when checking that the
>> "accept4" syscall is available. (see patch here [1])
>>
>> This breaks building for EL5 in mock on F16 (see log tail here [2]),
>> but not on a real CentOS 5 system.  A quick grep of
>> $MOCK_ROOT/usr/include indicates that __NR_accept4 isn't defined, as
>> it should be, so I'm not sure what's going wrong here, or if it's a
>> problem in mock or upstream.
>
> Well, if something fails in mock, but works on a real system it sounds
> like missing BuildRequires. Perhaps kernel-headers?

Thanks, I'll try that.

> In any case, not sure what this has to do with EPEL... we don't ship
> Node.js. ;)

Not yet, anyway.  ;-)

Sorry, I couldn't think of a better place to ask about EL5 mock issues.

-T.C.