Hi Sean,

thanks for caring about reproducible builds and reaching out to us!

On Fri, Apr 29, 2016 at 07:12:32PM -0700, Sean Whitton wrote:
> My package ocrmypdf builds reproducibly on my machine.  But it fails to
> build at all on the reproducible-builds.org cluster. 

that's actually not quite true, it only fails on amd64 in our setup, but
builds nicely on i386 and armhf! eg check

> The problem is
> that the test suite is getting permission denied errors while trying to
> read and write files, which causes the tests to fail.
> Do you know of any restrictions on reading and writing files on your
> cluster? 

no. Also: our amd64 and i386 setup are pretty much identical…

> I'm pretty sure that the test suite only tries to read and
> write files within its source tree, though I haven't verified that.

…except that i386 uses an ext4 filesystem for the pbuilder workspace,
while amd64 uses a tmpfs filesystem. 

(There are more variations between amd64+i386 (cpu type is varied on
amd64 but not yet on i386, OTOH there is 32+64 bit kernel variation on
i386, but not on amd64…)

> Kindly keep me in the CC.



Attachment: signature.asc
Description: Digital signature

Reproducible-builds mailing list

Reply via email to