[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: destdir: 'make package' inconsistency
On Mon, Jan 05, 2009 at 08:30:35AM -0500, Greg Troxel wrote:
> Joerg Sonnenberger <joerg%britannica.bec.de@localhost> writes:
> > The question is what do you base your observation on. If you assume
> > that everyone by default is using make package, you can just use
> > package-install. Part of the point here is that the different semantic
> > is intentional. You are asking the system to create a binary package,
> > whether it is installing files to PREFIX or not is an implementation
> > detail.
> I suppose that's technically true, but tons of people have been running
> 'make package' to get a package installed and packaged both for a long
> time. The variable DEPENDS_TARGET encourages it. So I think we should
> retain compatible behavior for this (reasonable) usage. Right now
> having the bahvior change is a reason not to turn on destdir support.
There is no reason why programs shouldn't use make package-install as
default. I don't even see why install-destdir is a better name than
package-install for the target...
> > I didn't change the existing install target for a very good reason -- it
> > would make do-install and install as targets asymmetric, which is IMO
> > even worse.
> I don't follow this - I would think that whatever is done to install in
> terms of name changes could be done to do-install.
So far we have a tight semantic relation between the phases of the
packaging and the invocing targets. "make install" essentially is a
wrapper around "make install" in the package. If you think of it that
way, the changed behavior for destdir builds is not the least
Main Index |
Thread Index |