W3C home > Mailing lists > Public > w3c-dist-auth@w3.org > April to June 2008

Re: Thoughts on relation to WebDAV

From: Jack Cleaver <jack@jackpot.uk.net>
Date: Sat, 24 May 2008 15:01:10 +0100
Message-ID: <48381FA6.30406@jackpot.uk.net>
To: w3c-dist-auth@w3.org

Petr Tomasek wrote:
> But every webdav client I've seen REQUIRES the DAV-header to operate
> on an URL as on a WebDAV collection. So it means You would
> practically disalow any partial (e.g. a read-only) WebDAV
> implementation, even if it works correctly and the full-fledged
> functionality is not needed at all!

I don't see it that way.

A partial implementation is one that doesn't implement the spec.
A read-only implementation is one that implements the spec, but doesn't
permit write operations.

There is a variety of reasons why write operations might not be
permitted by a compliant server. The client might not have sufficient
privileges, or the operation might be permitted at some other location,
or the repository might forbid all write operations for whatever reason.
Or the server software might *by design* not allow write operations.

It seems perfectly reasonable to assert that one supports MKCOL, but to
then refuse to perform this operation in every case, giving an
appropriate status-code. After all, if one *really* didn't support the
operation, then one would return a 501. Supporting an operation doesn't
imply performing it when it "should not" be performed. What "should not"
means is determined by the implementor.
> So the spec is wrong and insane if it doesn't allow for a partial
> implementation. Sorry.

This is silly. A partial implementation of a spec is not an
implementation at all (unless the spec explicitly provides for optional
elements - in which case the implementation is arguably not a partial one).

L1 (2518) says you MUST support MKCOL. That implies that a 501 is always
an incorrect response from an L1(2518)-compliant server.

Incidentally, there's nothing wrong with borrowing those parts of a spec
that you need, and ignoring the rest; this will only cause problems if
your software is expected to interact with other software that relies on
proper compliance with that spec, and furthermore your software claims
to be compliant when it isn't.

Received on Saturday, 24 May 2008 14:01:47 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 20:01:42 UTC