[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Package Pool Proposal



Craig Sanders <cas@taz.net.au> writes:

> On Fri, Dec 03, 1999 at 12:08:38AM +0000, Philip Hands wrote:
> > The whole point was to not separate the sources & archs, because that
> > makes things easier to find for humans, and makes no odds to machines.
> 
> it makes it a whole lot more difficult to mirror only the binaries
> without source.

With rsync it make almost no difference.  It's a question of either
editing the list of paths to include, or editing the list of
extensions to exclude.

For FTP and HTTP, I agree it does make it more difficult.

If that's thought to be important, then it's not difficult to generate
partial views of the pool, containing only one architecture or source
each, populated with hard links to the real package pool.  That would
double the number of inodes used for the archive, but other than that
would have practically no impact on things.

One real problem with splitting the pool is that in the case of
binary-i386 and binary-hurd-i386 there will be a load of packages that
will be common to both.  This would mean that we'll need to have files
hard linked between directories, which will make archive maintenance
harder.

Cheers, Phil.


Reply to: