[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: toppred and bowtie now reproducible

Hi Sascha,

On Thu, May 26, 2016 at 09:35:35AM +0100, Sascha Steinbiss wrote:
> > I've uploaded toppred.
> Great, thanks!

Hope your DD application will proceed so we can stop thanking each
other. ;-)
> > I've looked at bowtie and I have a question:  In the reproducible.patch
> > you set some varying variables to fixed (basically empty) values in the
> > Makefile.  In addition you are also commenting the variables inside the
> > code which might also hide other information like build options etc.  I
> > wonder whether the latter is needed.  There might be users who are
> > interested in those settings like Build Options and other things.
> Point taken. This is what I meant the other day when mentioning in an
> email about removing information from the output.

Ahhh, I should have put more attention onto your question then.
> > I assume that the Build Options are fixed for a reproducible build,
> > aren't they?
> They should be, according to
> https://tests.reproducible-builds.org/index_variations.html. C*FLAGS,
> LDFLAGS and friends should stay the same.


> > So what about
> > 
> >     -DBUILD_HOST="\"Debian-reproducible\""
> > 
> > and inject
> > 
> >     dpkg-parsechangelog --show-field Date
> > 
> > into -DBUILD_TIME.  I also wonder whether the string
> > 
> >     gcc -v 2>&1 | tail -1 | sed 's/ *(.*//'
> > 
> > should be generic enough to solve the reproducible issue by beeing
> > informative enough for the user.
> > 
> > What do you think?
> You are right, maybe upstream has included them for a reason. I have
> changed the corresponding patch in SVN to just set these to these
> suggested fixed values. Many thanks for your hint.

Its uploaded as in SVN.  BTW, according to my criterion "use Git once
the source tarball was repackaged" which is the case here since we are
removing the seqan code copy, we should move this package to Git sooner
or later - latest with the next new upstream version.

> After another run in my local r-b pipeline, I can confirm that the
> package is still reproducible. So I guess we can go with the new version.

Very cool that somebody of the team has a local r-b pipeline.  While I
agree that this effort is really important I did not minded to install
it here.  So its very good to know that somebody else is caring.
> P.S. Just for consistency, I will also adjust the output of bowtie2
> accordingly in a later upload.

Very good.

Kind regards



Reply to: