Re: New draft-draft of sftp...

[Pontus Skoeld <pont_secsh%soua.net@localhost>]

"Joseph Galbraith" <galb-list%vandyke.com@localhost> writes:

> Seriously, though, as your example pointed out, a VMS user and a
> unix user are expecting very different things in terms of how
> filenames are globbed.

This is an extremely important point. IMO globbing behaviour shall be
consistent throughout the client, independent of what server it is
connected to. A user that is used to file names being case-sensitive
given the following file list

alittlefile
AnotherFile

probably refers to alittlefile by a*, whereas someone used to
case-insensitive name mean both files. This is a user property (and
should probably be encoded in client behaviour), not a server
property.

Furthermore, I think this is really a non-problem. The server supplies
a list of file names that can be used for globbing. There is no
problem with doing case-sensitive (or case-insensitive) globbing with
those names (the names returned by the server are valid, independent
of how the user the user selects those file names).

Neither are there any new problems introduced by the server above
being able to refer to AnotherFile by the names anotherfile. A server
that does case-sensitive matching might have both AnotherFile and
anotherfile but not return anotherfile in the answers to readdir.

To rephase this; to check for the existence of a file, use
SSH_FXP_STAT, to avoid overwriting it, use SSH_FXF_EXCL. Using the
answers from readdir for this is *wrong*.

I have not yet had time to review the FTP thread, but I'll try to get
around to that ASAP, but I'm not certain how much of that
argumentation will be valid for sftp.

	/Pontus
-- 

Pontus Sköld, see <URL:http://soua.net/> for more information.