Karl Fogel <kfogel@galois.collab.net> writes:
> information needed to retrieve it). Thus, any directory entry could
> be a reference to a remote object. ^^^^^^^^^^^^^^^
Should be "any ancestry record or directory entry", perhaps.
A small matter of programming, heh...
-K
> The motivation for this is to be able to branch off repository X,
> storing the branch in your own repository Y. You don't even need
> write access to X.
>
> Union directories don't add any difficulties that weren't already
> there, as far as I can tell. You just store references to remote
> nodes as in any other kind of directory.
>
> Having commits be atomic across multiple repositories is probably
> difficult to the point of impossibility, granted; but the feature is
> still quite useful even if commits must be limited to a single
> repository at a time. Anyway, this limitation is not unique to
> distributed union directories -- it applies to any working copy whose
> data comes from two different repositories, by whatever means.
>
> -K
>
> > I wonder what kind of meaning one can assign to transactions if one
> > has a working copy assembled from multiple servers. Are transactions
> > meaningful across multiple independent servers?
Received on Sat Oct 21 14:36:16 2006