Bug#860428: reprotest: use an existing HOME in the control build

[beuc at debian.org]

Hi,

On Sun, Apr 16, 2017 at 07:11:08PM +0200, Mattia Rizzolo wrote:
> On Sun, Apr 16, 2017 at 06:55:34PM +0200, beuc at debian.org wrote:
> > After several tests (and then more) I eventually tracked it to HOME
> > being invariably non-existant in reprotest
> > (HOME=/nonexistent/first-build and HOME=/nonexistent/second-build),
> > while my normal compilation environment has an existing home (duh!).
> 
> Consider that both sbuild and pbuilder have HOME pointing to something
> non-existent.

I'm using reprotest outside of Debian packaging.  See e.g.:
http://blog.beuc.net/posts/Practical_basics_of_reproducible_builds_2/


> > - non-existing home: ./configure attempts to run conftest.exe, wine
> >   can't create '.wine', conftest.exe fails, configure assumes:
> >   checking whether we are cross compiling... yes
> 
> This feels quite buggy behaviour.  I suggest you consider this another
> bug in your package.

Yeah I wrote that this is fixed by specifying both --build and --host
to ./configure.  Btw this is one of the first ./configure tests from
autoconf, so nothing specific to my application.


> > To detect this issue, and probably others, I'd suggest making the
> > control build's HOME point to an existing directory.
> 
> By all means, I suggest instead having one build with an existent and
> writable HOME, and one with a non-existent one (this leaves out the case
> of an existent but not writable, though).

I believe we have the same thing in mind - I suggested an existing
(and indeed writable) directory only for the control build, not the
experiment build.

Cheers!
Sylvain