[OE-core] [PATCH v2] Introduce new SERIAL_CONSOLES to add multiple consoles for your MACHINE

[Chris Larson]

On Tue, Nov 8, 2011 at 1:50 PM, McClintock Matthew-B29882
<B29882 at freescale.com> wrote:
> On Tue, Nov 8, 2011 at 2:47 PM, Phil Blundell <philb at gnu.org> wrote:
>> On Tue, 2011-11-08 at 19:36 +0000, McClintock Matthew-B29882 wrote:
>>> On Mon, Nov 7, 2011 at 5:03 PM, Matthew McClintock <msm at freescale.com> wrote:
>>> > I don't get it. I think the first version was still OK. The first
>>> > version is pasted below.
>>> >
>>> >> +    idx=0
>>> >> +    tmp="${SERIAL_CONSOLES}" <- If it's not defined this will catch it.
>>> >> +    for i in $tmp
>>> >> +    do
>>> >> +     j=`echo ${i} | sed s/\;/\ /g`
>>> >> +        echo "${idx}:2345:respawn:${base_sbindir}/getty ${j}" >> ${D}${sysconfdir}/inittab
>>> >> +     idx=`expr $idx + 1`
>>> >> +    done
>>> >
>>> > See my inline comment above, but if SERIAL_CONSOLES is not defined
>>> > tmp="" and the for loop won't execute. Not sure if there is a more
>>> > standard way to do this though.
>>>
>>> ping..
>>
>> Yeah, now I look at this again I think it is probably OK.  Bitbake won't
>> expand ${SERIAL_CONSOLES} if it's unset, but the shell will and I think
>> you probably will get the right results.  Assuming you've actually
>> tested that case and it does work, I think I'm happy with what you have
>> above.
>
> Thanks Phil, I wanted to make sure I was not missing something obvious ;).
>

Note that one could always define SERIAL_CONSOLES ?= "" somewhere
appropriate. Wouldn't hurt anything to define defaults for things we
use. Maybe someday we can make referencing undefined variables raise
an error rather than this whole 'if bitbake doesn't expand it, let the
shell do it' behavior which is unintuitive. Not easily knowing when
something will be evaluated is problematic.
-- 
Christopher Larson
clarson at kergoth dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Senior Software Engineer, Mentor Graphics