Re: Source requirements and debian/missing-sources/ (was: "Browserified" stuff)
On Wed, Oct 12, 2016 at 09:00:50PM +0200, W. Martin Borgert wrote:
> Quoting Martín Ferrari <tincho@tincho.org>:
> >I had always understood that rebuilding from source was a hard
> >requirement. Is this not the case any more?
> >
> >I don't think that shipping a binary compiled upstream should be
> >allowed, so where's the line drawn?
>
> This is an interesting question indeed.
>
> If it is allowed for a package in main to ship a pre-compiled JS file
> with only some source files under debian/missing-sources/, it could be
> tempting to use all kind of pre-compiled *.c files in a package and do
> not care about a proper build process. Who cares about yaccs and
> bisons?
You're thinking small. Why not ship a pre-compiled ELF, built with some
paid version of ICC (screw silly sods on AMD chips like me[1]). Just think
of all that saved compilation time and avoided FTBFSes on new gcc versions!
Then you put a non-working dump of .c files, not necessarily the same as you
compiled, without a build system, into debian/missing-sources/ then never
update them.
> > I have never heard of debian/missing-sources. What is the
> > policy/documentation regarding this? I have repackaged tarballs many
> > times to add missing sources, I did not think there was another way to
> > do it!
>
> If I understand correctly, and please correct me if I'm wrong, this
> directory is for sources of files that are in your source package, but
> not in the binary.
>
> Example: A source package had an embedded source copy of B.min.js
> (minified, non-source), but the binary package does not have it,
> because A depends on libjs-B and has the right debian/links. Then you
> can put the B source into As d/m-s. If B.min.js is actually used in A,
> d/m-s is not sufficient. But maybe I'm on the wrong track...
That's my understanding as well.
[1]. ICC produces code that has different paths for different CPUs, using
optimized code when some instructions are available at runtime; when the
cpuid is not "GenuineIntel", it chooses a working but carefully deoptimized
variant despite the CPU having instructions that it wants.
--
A MAP07 (Dead Simple) raspberry tincture recipe: 0.5l 95% alcohol, 1kg
raspberries, 0.4kg sugar; put into a big jar for 1 month. Filter out and
throw away the fruits (can dump them into a cake, etc), let the drink age
at least 3-6 months.
Reply to: