Next message: Tim Ellison/OTT/OTI: "RE: Workspaces as versionable resources"
Date: Sat, 27 May 2000 13:36:37 -0400 (EDT)
Message-Id: <200005271736.NAA18931@tantalum.atria.com>
From: "Geoffrey M. Clemm" <geoffrey.clemm@rational.com>
To: ietf-dav-versioning@w3.org
Subject: LABEL vs. SET-LABEL-TARGET
We currently have two alternatives for marshalling label manipulation.
They are semantically equivalent, so this is just a marshalling question.
One (LABEL) is a method that is applied to a revision, and can either
add or remove a label for that revision. The body of the request
identifies the label and whether it is to be added or removed.
Another (SET-LABEL-TARGET) is a method that is applied to a versioned
resource, and specifies with revision of that versioned resource
(possibly "none") should be selected by a given label. The body of the
request specifies the label, and specifies what revision should be
selected by that label (possibly "none").
I have argued (excessively :-) for the SET-LABEL-TARGET marshalling,
while JimA as argued (possibly also excessively :-) for the LABEL
marshalling. Chris supports the SET-LABEL-TARGET marshalling, while
Tim supports the LABEL marshalling. The rest of the design team could
go either way,
As stated in my earlier message, we are proposing that a server be
required to provide "stable" URL's to the versioned resource metadata
(i.e. history-URL's). As a result, we don't actually need the
"versioned-resource" value of a Target-Selector anymore, *except* to
make it convenient to apply the SET-LABEL-TARGET method without
querying first for the history-URL. But if we used the LABEL
marshalling, there is no need for the request URL to identify a
versioned resource.
Since the LABEL marshalling does provide us with the opportunity to
simplify the Target-Selector header, I am willing to switch over to
the LABEL camp (it is a shorter method name as well :-).
We'd like to stabilize core versioning ASAP to allow for some
interoperability prototyping, so if you care about this at all,
please let me know. Otherwise, I propose we go back to the
LABEL marshalling that appeared in 4.4.
Cheers,
Geoff