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

Re: Bug#521810: debian-policy: Document user defined fields starting with X-



Hi!

On Tue, 2009-03-31 at 01:10:20 -0700, Don Armstrong wrote:
> On Tue, 31 Mar 2009, Raphael Hertzog wrote:
> > On Mon, 30 Mar 2009, Russ Allbery wrote:
> > > Nils Rennebarth <nils.rennebarth@funkwerk-ec.com> writes:
> > > > Usually, unknown fields are iggnored by the debian packaging system. To
> > > > avoid conflicts of user defined fields with field that may be used by
> > > > debian in the future, we suggest to use field names starting with X- (so
> > > > you need to put X[BCS]-X-foo into the control file) which are guaranteed
> > > > to never conflict with future official fields.
> > > 
> > > Is this because the X in front of [BCS] is stripped off when the field is
> > > copied into the resulting binary or source package?
> > 
> > Yes.
> 
> Is there any reason why we can't transition official X-* headers to
> real * headers as they become widely used (and when they're inshrined
> in policy)?
> 
> Some transition period would be necessary, and dpkg-gencontrol could
> be patched to automatically rename the X-* headers to * for the
> transition period (and tools that use the header should look at both
> X-* and * headers[0]).

> 0: I submit that tools that examine the headers should let the *
> header override the X-* header when they're being written, unless
> there's some extraordinary reason not to do so.

Hmm, I'm not sure to which X-* fields you are referring here, if those
are the ones in debian/control, then that's exactly what we do when we
officialize a field in dpkg-dev. The tools get changed to understand
the X-Field as an alias for the new recognized Field.

If you meant X-Field in the binary, .dsc or .changes file, then what's
the point in having such field namespaced, if we want to consider it
a direct alias when officializing it? It also implies making all tools
dealing with binary, .dsc, and .changes fields understand both. Of
course it gives us the advantage that we might be able to mechanically
transform it in case the samantics changed, but I'm not sure that's
worth the trouble.

regards,
guillem


Reply to: