Roger asked to submit a wishlist bug, so... here you go. :)

Hi,

what is the purpose of doing this within sbuild?

Is it for what is now called reproducible builds?

Would it be sufficient to do two individual sbuild runs and then compare the
resulting .buildinfo files to fix this bug?

Can you give a pointer to the original thread that sparked this idea?

Thanks!

cheers, josch

Hi,

if the purpose is to check whether the debian/rules clean target works, then
this is already possible using sbuild external commands. This is because in
§4.9 it says about the clean target: "This must undo any effects that the build
and binary targets may have had". So building twice is neither sufficient nor
necessary to check this. Building twice could even mean that the second build
passes but has output different from the first. Instead, using an sbuild hook,
the state of the unpacked source could be recorded before the build starts and
then again after the build ended and the clean target has been executed again.

Thanks!

cheers, josch

Hi,

thanks to Stefano in #424846 I now know where the quoted message from Roger in
the original message of this report comes from:

http://lists.debian.org/debian-devel/2007/05/msg00553.html

Thus, the intend indeed seems to have been to check for the effectiveness of
the clean target.

Thus merging with 424846.

Thanks!

cheers, josch