[rb-general] Debian buster, 54% reproducible in practice (Re: Core Debian reproducibility: 57% and rising!)

[Vagrant Cascadian]

On 2019-03-01, Daniel Shahaf wrote:
> David A. Wheeler wrote on Fri, 01 Mar 2019 23:01 +00:00:
>> The *majority* (54%) of packages in real-world Debian Buster are now 
>> reproducible!!
>
> (Just to be clear, that's the figure for binary packages.  I don't know
> if the figure for source packages is higher or lower.)

Certainly lower; there are a number of source packages which produce
several binary packages, of which only some of the binary packages have
reproducibilitiy issues. For example, there are issues with some of the
toolchains used to produce PDF documentation that only effect packages
that ship PDFs.

So if you're measuring from source packages, any of the produced binary
packages could make the whole source package "unreproducible".

The ~93% reproducible in theory is from the lens of a source package,
and the ~54% binary packages, so in a sense, it is sort of comparing
apples to oranges.

There are also caveats in the method used thus far, trusting that
uploaders of .buildinfo files to be acting in good faith and
buildinfo.debian.net is returning valid data. Pretty sure I saw some
TODO items to address those aspects somewhere.

All that aside, it's still really great work, and a useful comparison!
Thanks for taking an off-the-cuff idea and turning it into something
more comprehensive, Holger! :)


live well,
  vagrant
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 227 bytes
Desc: not available
URL: <http://lists.reproducible-builds.org/pipermail/rb-general/attachments/20190301/d1e1ba1a/attachment.sig>