Call for feedback on a Ports-collection change
cpressey at catseye.mine.nu
Thu Jan 8 20:17:13 PST 2004
On Fri, 09 Jan 2004 06:54:42 +0300
Sergey Matveychuk <sem at ciam.ru> wrote:
> Garance A Drosihn wrote:
> > At 2:33 AM +0100 1/9/04, Max Laier wrote:
> >> 2) Changes are much harder to track:
> > On the contrary, changes should be *easier* to track. All the
> > information for any given port will be in two files. This will
> > not be true for all ports (particularly for ports which have a
> > lot of patch files).
> Let's image a situation: port has changed. What is chaneged? Let's see
> in WebCVS. Does distfile has changed? If yes, I know tarball has
> changed. pkg-plist has changed? I know a files structure has changed.
> I got this information even without opening this files. I'll open only
> Makefile to see a changes in it.
> It may be much harder to look at a big diff instead.
FWIW I agree with this point. IMO a much better idea would be:
Hack cvsup so that it can automatically create/update sharfiles of
specified directories on the client.
This approach would:
- achieve the stated goal (save inodes)
- be virtually seamless (nothing in CVS would have to change)
- have greater applicability (i.e. it could be useful to other projects,
not just the ports tree.)
Plus you'd get to code in Modula-3 :)
More information about the freebsd-ports