Johnny Lam wrote:
>> When installing or deinstalling a package, it is not enough to lock the
>> WRKDIR against foreign processes. Instead, the PKG_DBDIR is locked,
>> preventing other (pkgsrc) processes from modifying the set of installed
>> packages while another one is doing that.
> 
> I think this commit is incorrect.  The stages for a single package 
> should keep locking within the package work directory.  If you want to 
> gate the package addition and deletion, then we should have a separate 
> set of lock macros just for that purpose

ok, I agree with this one.

> and they should be dependent on the package system flavor.

but I don't agree with this. I think locking LOCALBASE should not depend 
on the package system flavor, but should be done the same for every 
flavor. The reason is simply that it doesn't matter which tools you are 
using, when other processes modify files in LOCALBASE. The print-PLIST 
will get confused in each case, except if we make it depend on the 
package tools flavor.

Maybe my choice for _PREFIX_LOCKFILE was bad. I could rename it to 
_LOCALBASE_LOCKFILE and set it to ${LOCALBASE}/.lockfile.

Roland