On 01/19/2018 01:34 AM, Burton, Ross wrote:
On 18 January 2018 at 16:47, Randy MacLeod
<[email protected] <mailto:[email protected]>> wrote:
On 2018-01-11 11:18 AM, Burton, Ross wrote:
On 9 January 2018 at 01:40, Chen Qi <[email protected]
<mailto:[email protected]> <mailto:[email protected]
<mailto:[email protected]>>> wrote:
* hostname is explicitly enabled to keep the same with
previous recipe's
behaviour.
But buildhistory-diff:
packages/corei7-64-poky-linux/coreutils/coreutils: FILELIST:
added "/usr/bin/hostname"
So the old coreutils wasn't building hostname at all (we use
/bin/hostname from busybox).
Are you asking Chen to package coreutils.hostname and
use the alternatives mechanism with busybox's hostname as the default?
(busybox hostname has annoyed me in a past life!)
I'm not sure what I'm asking, just pointing out that the log is wrong
as the old coreutils wasn't building a hostname binary.
If someone sets base_bindir and bindir to the same directory then
there will be a conflict, so this does need to be sorted somehow.
Ross
Hi Ross,
You are right about this. The old coreutils doesn't build out hostname.
I cannot recall why I thought it did. Anyway I can see from this mistake
that using buildhistory is indeed a big help, and I'll use it in my
future work.
I've sent out a fix to resolve the conflict problem about hostname
between coreutils and other packages (busybox and net-tools).
Best Regards,
Chen Qi
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core
