> A big point is that the unsafe_depends flag should be integrated into
> pkgsrc, so the manual make replace causes the cleanup rebuilds to
> happen later when you run the tool.  And, partial tool runs just do
> partial work, and one can just run again.  With the metadata in
> a private tool area, things are harder than they need to be.

Some user POV thoughts here:

I think it is important to have an option to make the tool rebuild regardle=
ss=20
of that flag, or to be able to specify the behavior of that flag.

Though trying to be conservative it is a useful method of preventing=20
unecessary re-compilation, in practice I have found with both pkgsrc and=20
=46reeBSD ports that things work much better when you just synch up=20
*everything*, regardless of specified depdency versions and whatnot, simply=
=20
because allowing mixed-tree installations adds another variable that=20
effectively means your particular combination of software has never been=20
tested by anyone. With an up-to-date tree your environment is likely to mat=
ch=20
relatively well that of the person updating and testing the packages.

=46rom a user perspective, I would often have no trouble taking the extra C=
PU=20
usage hit if it meant decreasing the chances of trouble.

=2D-=20
/ Peter Schuller, InfiDyne Technologies HB

PGP userID: 0xE9758B7D or 'Peter Schuller <peter.schuller@infidyne.com>'
Key retrieval: Send an E-Mail to getpgpkey@scode.org
E-Mail: peter.schuller@infidyne.com Web: http://www.scode.org