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

Re: discrepancies between uploaded and source-built .deb

On Tue, Mar 22, 2005 at 06:34:58PM -0800, Karl Chen wrote:
> Hi, 
> I'm doing something that involves building every Debian package,
> and I'm finding (usually minor) discrepancies between what I build
> from source packages, and the binary packages uploaded by
> maintainers.  I'm building each package in its own chroot which
> contains only the minimum packages (bootstrap + build-essential +
> build-dependencies).
> Are such things considered bugs?

A bit of this is ineviteable because of slight differences in host
system (builds to add hostname you compiled on to version strings, for
example), or time of build (timestamps), but most importantly, different
versions of build-dependencies. The environment changes, so does the

Usually, this doesn't give a significant difference, and certainly not a
regression. If there is a real regression, or a significant change,
yeah, that might (might!) be a bug. A build of a package needs to be
reasonably reproducieable after all. If there is no noticeable change
though, it's certainly not a bug.

Bottom line is that a difference is not a bug per se, but it may be an
indication of a bug.
> Also, what is the policy on the relationship between source
> packages and binary packages.  May two source package both produce
> the same binary package?

In transitional situations this happens for a while, this should however
not persist. The package producing the lower version cannot be updated
easily, as the archive tools would reject an upload where still the
package in question is of lower version number than the one built by the
other source package. I think it'd be good to ship sarge without such
situations, but again, this needs to be looked into on a case-by-case
basis, and I certainly dare not say that every such case must be a bug
(but I suspect so in general).

Note that for this issue, one of the Debian FTP scripts (rene) actually
checks for this situation, and I plan to file bugs based on that test
real soon now.


Jeroen van Wolffelaar
Jeroen@wolffelaar.nl (also for Jabber & MSN; ICQ: 33944357)

Reply to: