Hi,
On Wed, 24 Nov 2021 at 20:55, Leo Famulari <l...@famulari.name> wrote:
> I'm seeing something weird, where Guix is creating different derivations
> based on the same Git commit, between two machines of the same system
> type:
Euh?!
Many things appear really weird. Just, how is it possible,
> Derive([("out","/gnu/store/81bdcd5x4v50i28h98bfkvvkx9cky63w-hello-2.10","","")]
("/gnu/store/xmq151lsv0p5q435q8l3f9yi8ddrxs00-hello-2.10.tar.gz.drv",["out"])
> Derive([("out","/gnu/store/81bdcd5x4v50i28h98bfkvvkx9cky63w-hello-2.10","","")]
("/gnu/store/78ww8sqfpxavnkjgwafs7xr2zwzza3mc-hello-2.10.tar.gz.drv",["out"])
to have 2 different fixed-outputs which lead to the same final output hash?
On my machine, running two time-machine, I get the same as machine A
/gnu/store/260bk0ch4np4h2yz5yqhf8hjbsyhwpmr-hello-2.10.drv.
Where does Guix on machine B come from? Because, Guix inherits from
some previous config, I mean, for instance:
./bootstrap && ./configure --localstatedir=/var && make
./pre-inst-env guix pull
guix pull
guix <actions>
is not necessarily the same as:
guix pull
guix pull
guix <actions>
because ’sysconfdir’ is not the same. See for instance this
<http://issues.guix.gnu.org/issue/47401> for documenting this behaviour.
Cheers,
simon
