[Reproducible-builds] Bug#774496: ncurses: please make ncurses build reproducibly

Jérémy Bobbio lunar at debian.org
Sat Jan 3 14:47:43 UTC 2015


Sven Joachim:
> I am not strictly opposed to your patch, but I think the problem should
> be tackled elsewhere, so that fewer packages need to be modified.

In our experimental toolchain, packages built using `dh` will
automatically call `dh_strip_nondeterminism`. Packages using `dh` are
the 60% plateau you can see on:
https://jenkins.debian.net/userContent/stats_pkg_state.png

Since a few days, we are getting more packages considered reproducible
as we moved generation of the .buildinfo files to dpkg-buildpackage
instead of doing it as a dh stop.

> Have you considered filing a wishlist bug against binutils to build with
> "--enable-deterministic-archives", so that ar and ranlib operate in
> deterministic mode by default?

Yes. This is how we did the experiment in January last year:
https://wiki.debian.org/ReproducibleBuilds/Rebuild20140126

During discussions at DebConf14, it was acknowledged that having
`--enable-deterministic-archives` could break some build systems in
subtle ways and that it would be better to do without. Unfortunately,
I have never been able to get Matthias Klose's opinion on the matter.

> Also, what's the current state of #759895?

The original patch became part of strip-nondeterminism. I guess it
should be updated to ask the inclusion of the following patch:
https://anonscm.debian.org/cgit/reproducible/debhelper.git/commit/?h=pu/reproducible_builds&id=7fbd7ba

-- 
Lunar                                .''`. 
lunar at debian.org                    : :Ⓐ  :  # apt-get install anarchism
                                    `. `'` 
                                      `-   
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.alioth.debian.org/pipermail/reproducible-builds/attachments/20150103/18320831/attachment.sig>


More information about the Reproducible-builds mailing list