On Tue, 2005-10-18 at 11:29 +0200, Jules Colding wrote: > On Tue, 2005-10-18 at 14:40 +0530, Parthasarathi Susarla wrote: > > On Tue, 2005-10-18 at 10:05 +0200, Jules Colding wrote: > > > Forcing unique sibling folder names on creation will partly solve the > > > dilemma, but what should happen when identically named siblings are > > > found anyway? > > If indeed a protocol(server) allows you to have identically named > > siblings, then the matter of concern would be how we would store it > > locally. The filesystem would obviously not allow us to have identical > > siblings. > > > No, but we can do almost as good by appending a ' ' to the name of the > first identically named sibling and ' ' the next and so forth. I am > pondering how to do this the best. > > I do not think that forcibly changing the name on the remote server will > be any good, so I think that a local mapping agent must be used. > > Something like: > > Case A - N identically named folders/objects are found on the remote > server: > > A map of display names are created whereby the remote names used to > construct a one-to-one "local name" <==> "remote UUID" mapping. > The local names are constructed from the remote names by appending > 1..(N-1) ' ' characters to the remote names. > > Case B - An Evolution user wants to create an object with a name that is > identical to an existing sibling: > > We deny the request. > > Case A above can get complicated, but I don't see how we can do > otherwise if we want to cover all use cases. > > Thoughts?
We should deny, unless the protocol tells what to do. Cheers, partha _______________________________________________ Evolution-hackers mailing list [email protected] http://mail.gnome.org/mailman/listinfo/evolution-hackers
