Re: xlib6g now depends on xfree86-common (?)
On Tue, 9 Feb 1999, Marcelo E. Magallon wrote:
> On Tue, Feb 09, 1999 at 04:44:44PM +0100, Santiago Vila wrote:
> > It has come to my attention that the new xlib6g base now depends on
> > xfree86-common.
> > I think this dependency is artificial, because of two reasons:
> Santiago, I really hope you are not proposing to remove that dependency.
> Did you take a look at xfree86-common's filelist?
> * A bunch of directories. Some programs don't work if those directories
> doesn't exist. Which programs? The programs that depend on xlib6g.
This is the point.
Do you guarantee that *every* program will break if those directories do
If there is a single program that does not break, the dependency is
artificial and we should allow every package to depend on xfree86-common,
on xlib6g, or both, in an independent fashion.
If every program breaks, then why two different packages?
Just merge xfree86-common and xlib6g into xfree86-common.
> (The other solution is to merge xfree86-common back in xlib6g, where it
> lived for a long time, but why do that? You have a very nice binary-all
> pacakge there)
Fine. *This* is a valid point (glad that you mentioned it).
Splitting an existing package into a binary-all and a binary-<arch>
is something useful when the binary-all package is big enough.
[ xfree86-common is 114612 bytes long. xlib6g is 943382 bytes long,
if the X maintainer says it is worth to split, I accept it ].
If this is the reason for the split, could we have it please
written in xfree86-common's README?
"c0a8c0e79e4d0fb09b776b8c861fd8d3" (a truly random sig)