A poor maintainer's problem

Artem Bityutskiy dedekind1 at gmail.com
Wed Nov 5 23:28:35 PST 2014


On Wed, 2014-11-05 at 22:05 +0100, Robert Jarzmik wrote:
> "Keller, Jacob E" <jacob.e.keller at intel.com> writes:
> 
> > I believe that Aiaiai tries to re-use the same trees in order to conserve space?
> >
> > Are you running jobs in parallel or serial? If we can isolate exactly why
> > objects are being kept around, maybe we can determine a way to refactor so that
> > we don't keep this data if it's no longer useful, and thus we can shrink the
> > size of required disk space.
> 
> Hi Jake,
> 
> In serial, one after another.
> 
> And as far as I understand aiaiai-test-patchset, it does:
> 	for defconfig in $defconfigs; do
> 		test_configuration "$defconfig"
> And test_configuration does :
> 	aiaiai-make-kernel $verbose $sparse $smatch $cppcheck $coccinelle -o
>         "$obj1" ...
>        	aiaiai-make-kernel $verbose $sparse $smatch $cppcheck $coccinelle -o
>         "$obj2" ...
> And there is no rm anywhere around.
> 
> I was wondering if at the end of test_configuration I could add a :
> 	rm -rf "$obj1" "$obj2"

I'd encourage you to just try and send a patch. I think it should be
just fine.

The general approach in all these script was that things get cleaned-up
upon exit in the exit handler (called on various kinds of signals
leading to termination). The handler which is always somewhere near the
top of the script.

So when writing the scripts, I would just add a big rm -rf <tmpdir> or
something to the exit handler, and forget about clean-up stuff - it is
handled already.

Artem.




More information about the aiaiai mailing list