Re: [long] Re: I-D ACTION:draft-ietf-webdav-versioning-01.txt

I think I may see one possible source of disconnect between the two
protocol proposals.  I assume there are two key use cases:

- All persistent state is maintained on the web server.
The client can GET some data via some WebDAV client and manipulate it
in memory.  Whenever there is any new state to be preserved it is PUT
back to the web server.  This would be the case if you start some activity
at work, and then need to continue it from home, and all you've got at
home is an internet connection and a computer with a WebDAV client.

- Persistent state is shared between the client and the web server.
The client can GET some data via some WebDAV client, manipulate it in
memory, *and* store it locally.  This client doesn't need to
communicate as frequently with the web server, since it can cache
server state, and can make things persistent in its private persistent
store, but users need to effectively carry this client state around
with them.  This would be the case if you carried a laptop around with
you, or always dialed up directly to the same client machine to do remote
work.

I believe our protocol MUST handle both these use cases.  I suspect
that Chris has focused only on the second case.

In particular, my comments about the scalability of the WebDAV lock
mechanism is specifically directed at the first use case.  If there is
some lock token that must be remembered, then either users have to
check-in everything before they can go home (an unacceptable
requirement, in my opinion), remember all the lock tokens for what
they have checked out (now that will be a really popular option :-),
or store them all on the server somewhere, and have some mechanism for
downloading all the lock tokens all as a clump, once they have passed
some user authentication check.  I believe the last option, involving
user authentication, is the only viable long-term answer, and I expect
many clients will be willing to use the existing web-based user
authentication mechanisms, until WebDAV choses one as a "standard".
But WebDAV must define how information is stored into and retrieved
from the "clump" or we've blown any chance of client interoperability.

My rejection of server-defined temporaries as the location for checked
out resources is based on the same reasoning.

To support the first use case, my proposal asserts that a server is
responsible for maintaining working resources as human understandable
URL's, and parallel work is supported by having a single additional
identifier (the Workspace-ID) that the user must remember, to
distinguish his work from other users' work.

This proposal is based both on a logical analysis of this use case,
as well as the pragmatic observation that this is how virtually all
CM systems work: they provide a workspace with human understandable
"pathnames" (URL's) for the checked-out working resources.  It would
be hard to make current CM systems *not* work this way, and I think
producing a protocol that reflects how current systems do versioning
is an important criterion for evaluating a protocol proposal.

Soooo, with this lengthy preface in mind, let's get back to the
current thread of discussion ... 

   From: Chris Kaler <ckaler@microsoft.com>

   > From:	Geoffrey M. Clemm [mailto:gclemm@tantalum.atria.com]
   > ...  I believe the argument to CHECKOUT, CHECKIN, and UNCHECKOUT
   > should always be a consistent human-comprehensible URL.

   [CK] This requires then that a checkout token be passed as part of
   the request.  We can do this, but how then is a CHECKOUT different
   from a LOCK?

A CHECKOUT makes a readonly URL writeable.  A LOCK restricts who can
write to a writeable URL.

   >    There is a certain utility to having the magic files in a magic
   >    directory for ease of implementation-- you know you only have to
   >    treat a file in a special way if it's in a special location.  In
   >    my opinion, if you've already done the necessary work to make MKREF function,
   >    the additional amount needed to support checked-out files and
   >    configurations going anyplace should be small.

   > I agree.

   [CK] How then does one quickly discover the defined configurations?

Ooops.  I was agreeing with the comment wrt checked-out files.  I
agree with Chris that versioning meta-data (such as workspaces,
configurations and activities) should be stored in standard places, so
that they can be easily discovered/browsed.

   > Also, I would want the "lock" to be optional (I don't think the "lock"
   > mechanism scales well, and want it to be an optional part of the
   > versioning protocol).  Again, I believe this is not an essential part
   > of your point.

   [CK] The lock portion is option in -01.  I'm curious as to why you
   don't think the lock mechanism will scale?

See comments above.

   >    (i.e.  
   >    CHECKOUT /foo/bar.html HTTP/1.1
   >    Host: www.foobar.com
   >    ...

   >    returns

   >    HTTP/1.1 201 Created
   >    Location: /foo/bar.html
   >    Revision-Id: <opaquelocktocken:rejrei-43343-rereffre>
   >    Lock-Token: <opaquelocktocken:rejrei-43343-rereffre>

   > I'd omit the Location header (it's always created at the place
   > specified by the CHECKOUT).
   > The Revision-Id is no longer necessary, since you either are
   > checking out into the default workspace, or you are checking
   > out into a specific workspace.  In any case, your workspace
   > determines which version you see, so no Revision-Id's need
   > to be passed in requests or reponses.

   [CK] Very confusing.  How did you identify the checkout?

Every workspace has at most one checkout at a given URL.  If you
tried to check it out again, it would say "already checked out".
So if you have a URL (and optionally, a workspace ID), you have all you need.

   You don't have a lock-token and you used the same URL?

Sure.  Before the CHECKOUT, the URL referred to an immutable revision.
After the CHECKOUT, it referred to a working resource.  Where is a
lock-token required to identify what has been checked-out?

   If you return a
   reivision-id then it is part of the revision graph by definition. This
   doesn't make sense.

I don't follow what you are saying here ... (but I do agree that
this doesn't make sense, for an appropriate value of "this" :-).

   > And I'd just use the LOCK command following the CHECKOUT if I want a
   > LOCK, rather than bundling LOCK with CHECKOUT (I don't believe the
   > frequency of CHECKOUT warrants worrying about an extra LOCK method call in
   > this case).

   [CK] In order for a lock to make sense, it must be atomic with the
   checkout.

Why?  You check something out to make it writeable.  If somebody jumps
in ahead of you to get a LOCK, so be it.  That just means that they
also wanted to work on it, and they beat you to it.  No different from
them beating you to a combined CHECKOUT/LOCK command.

   As well, there must be interaction between lock and
   checkout.  That is, Non-versioning-aware clients that use LOCK must be
   able to block Versioning-aware clients otherwise they will clash and
   non-versioning Clients will not be able to react.

I'd have to see a description of the particular clash you have in mind.
Some clashes in locking are inavoidable, as we have discovered in
work out the details of locking and advanced collections interact.

   [CK] I also disagree with your premise about locks.  I find that
   people frequently want exclusive locks when they checkout a resource
   even if they have merge facilities.  It is a process issue.

And some people don't want exclusive locks.
And some people don't want to use WebDAV locks to specify access control.
Which is why I resist any proposal to inextricably link WebDAV locks
with the orthogonal notion of CHECKOUT.

   [CK] I think the Workspace-Id concept is confusing and adds an extra
   layer of burden on clients.  Maybe I just don't get it.

In VSS, it's a "sandbox".  In ClearCase it's a "view".  In many/most other
CM systems, it's called a "workspace".  They've all got the concept, and
any client that doesn't understand it won't be able to interoperate in a
simple with any of the existing CM systems.

This may just be confusion over terminology ... if we get a common set
of definitions for "configuration" and "workspace" this week in Utah,
this confusion may just disappear (well, it's at least a possibility :-).

Cheers,
Geoff

Received on Tuesday, 9 February 1999 10:26:04 UTC