[Buildroot] target busybox skeleton.

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

On Thu, Mar 08, 2007 at 02:23:09PM +0200, Heikki Lindholm wrote:
>Bernhard Fischer kirjoitti:
>>On Thu, Mar 08, 2007 at 01:14:40PM +0200, Heikki Lindholm wrote:
>>
>>>Bernhard Fischer kirjoitti:
>>>
>>>>On Sun, Feb 18, 2007 at 06:32:11PM +0100, Thomas Lundquist wrote:
>>>>
>>>>
>>>>>Maybe it's a bit too tiny?
>>>>>
>>>>>I'd suggest that /usr/lib is added.
>>>>
>>>>
>>>>hmz. And what if i don't have anything in /usr/lib? Wasted dentry.
>>>>
>>>>What about switching to "install -D" instead, where appropriate?
>>>
>>>If doing major changes, please consider taking portability issues into 
>>>account. In this case, "-D" isn't supported on OS X.
>>
>>
>>It shouldn't be too hard to ship our own working install source an
>>compile and use that for the host, but thanks for the hint anyway.
>>
>>MacOS-BSD already has trouble due to the use of "cp -a", IIRC and i was
>>gradually switching those over to cp -dpf like suggested in one
>>bug-report.
>
>Getting a bit off-topic here maybe, but I have to comment that your 
>effort has not been that useful, because such bug report has been wrong: 
>Mac OS X doesn't support the -d option for cp and thus every cp -dfp is 
>as dysfunctional as cp -a. Proper way would be cp -RPfp, which, as an 
>unavoidable side-effect, is also recursive. I also filed a bug report on 
>this a while ago.

great. It would have been helpful if you would have added a followup
note on that initial incorrect report (i assume that you filed a
different bug).

I'll go for the solution where we provide our own install.c and use that
as $(INSTALL) that can be overridden by the user. No idea if that will
handle $(INSTALL) some*.txt /there/ style multiply files, we'll see..