I agree with you that it is not deterministic. What is really important is that 
the
checksums and the tarball come from a source you trust and are in agreement.
It is a simple security measure and I am not sure there is value in making it
deterministic. I have a feeling that if you want to do that we may have to 
change
packaging tool radically.

François

> On 13/01/2015, at 12:24, mmarco <mma...@unizar.es> wrote:
> 
> If the problem is that packaging is not deterministic... what if we run the 
> hash check on the unpacked files instead of the tarball?
> 
> 
> El lunes, 12 de enero de 2015, 11:59:53 (UTC+1), Thierry 
> (sage-googlesucks@xxx) escribió:
> Hi, 
> 
> it is advised to distribute unmodified upstream tarball at much as 
> possible, so that the end-user should be able to check that the tarball 
> shipped by Sage has the same hash that upstreams. However, when size can 
> be reduced by a huge factor, integrity arguments become pretty weak and we 
> randomly upload hand-modified tarballs on tickets without a clear checking 
> process during the review process. 
> 
> In some cases, one possibility is to discuss with upstream to ship both 
> full and trimmed sources (which will benefit to other downstream, e.g. for 
> mathjax that can be considerably reduced while keeping all features). 
> 
> Another mid-term compromise could be to strip some few upstream source, 
> but in a checkable and reproducible manner, that is, with a spkg-src 
> script that will produce deterministic tarballs, so that anyone (in 
> particular the reviewer) can re-run the script and check the hashsums. By 
> default, tarballs are quite volatile because of timestamps and ownership, 
> also the file ordering seems to depend on the computer, the posix format 
> is nondeterministic, and i may have missed some other subtleties. 
> 
> In order to try such possibility on the next matplotlib update, could some 
> people (especially someone using OSX) give me (with minimal info on their 
> OS, arch, and tar --version) the result of: 
> 
> wget 
> https://downloads.sourceforge.net/project/matplotlib/matplotlib/matplotlib-1.4.2/matplotlib-1.4.2.tar.gz
>  
> tar xf matplotlib-1.4.2.tar.gz 
> rm -rf matplotlib-1.4.2/lib/matplotlib/tests/baseline_images/* 
> find matplotlib-1.4.2 | sort | tar --no-recursion -cj --format=gnu 
> --mtime='1970-01-01 01:00' --group=0 --owner=0 -f matplotlib-1.4.2.tar.bz2 -T 
> - 
> shasum matplotlib-1.4.2.tar.bz2 
> 
> Thanks, 
> Thierry 
> 
> 
> -- 
> You received this message because you are subscribed to the Google Groups 
> "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to sage-devel+unsubscr...@googlegroups.com.
> To post to this group, send email to sage-devel@googlegroups.com.
> Visit this group at http://groups.google.com/group/sage-devel.
> For more options, visit https://groups.google.com/d/optout.

-- 
You received this message because you are subscribed to the Google Groups 
"sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to sage-devel+unsubscr...@googlegroups.com.
To post to this group, send email to sage-devel@googlegroups.com.
Visit this group at http://groups.google.com/group/sage-devel.
For more options, visit https://groups.google.com/d/optout.

Reply via email to