V6 test-in-container patch

[DJ Delorie]

"Carlos O'Donell" <carlos@redhat.com> writes:
>> There's also preclean and postclean triggers, which were named
>> preclean.txt and postclean.txt.  I've renamed those to preclean.script
>> and postclean.script although they're not really scripts.
>
> Any reason we don't just add a "preclean" or "postclean" command to the script
> language used in the *.script file and process it that way?

The triggers have to be acted on before we build the testroot, but the
script has to be run *after* we build the testroot.  So it could be
done, but it would be a bit tricky.

> Yes. A new file. If the file gets too big slipt logical utility code into
> other sources files and build them separately or include the C code.

I will refrain from splitting it up, until such time as the weird
internal functionality is needed elsewhere too ;-)

> Add a comment that the number is randomly chosen to avoid calling
> realloc too often. We want to guide future readers by stating the
> intent of your choice.

Done.

>>> Check for free failure. Or add xfree. Just kidding ;-)
>> 
>> Oh sure, joke about this one, don't mention xopen() ;-)
>
> ... we have xopen() and you should be using it? :-}

I prefer to be able to print a more suitable error message ;-)