- From: Greg Stein <gstein@lyra.org>
- Date: Tue, 17 Apr 2001 16:23:27 -0700
- To: Jim Amsden <jamsden@us.ibm.com>
- Cc: ietf-dav-versioning@w3.org
Jim, Please stop using HTML for your emails. Heck, use a better email package (your emails have looked wonky for a long time). See below for the problem with using HTML. There is *NO* separator between your message and the message you quoted. There isn't anything that says the quoted email was from Geoff. Quite impossible to read/parse/whatever. I believe netiquette generally says to use plain text for mailing lists. Not everybody out there has HTML-capable readers. Cheers, -g On Tue, Apr 17, 2001 at 07:48:13AM -0400, Jim Amsden wrote: > > OK with me. > I believe it is fair to say that based on activity in the mailing > list, there appears to be strong support for the Label header within > the working group. > As for the issues raised concerning the Label header: > - Internationalizable display of labels: > This actually has little to do with the label header, since you don't > use the label header to create labels, but rather use the body of the > LABEL method. In the body of LABEL method, you can annotate the label > with as many internationalization attributes as it wishes, and you > query for labels with PROPFIND, which will give you all those > attributes back. > - Label matching > Here is the area where the Label header does come into play, because > it is the way several methods will pass in the label to be "matched". > My impression is that most of the members of the working group that > care about labels find a simple byte string matching of the label to > be sufficient for their needs. It is OK for an Englishman to label a > version "gross", and then have this version be selected by a German > that asks for the "large" version (i.e. "gross" in German, put please > don't get too concerned about the details of this example :-). > This is no more likely to happen than for some other client to use the > word "gross" in English to mean something other than what the second > client had in mind. In fact, it is probably much *more* likely for > the two different English uses of a string to collide, than it is for > an English use to collide with a German use. > Clients that really care about his issue will just use baselines > instead of labels, since a baseline selection of a version can never > "collide" with another baseline (which is why I think we should just > get rid of the label feature altogether, but that's just me :-). > So if we accept all that, the only thing left to do is to decide which > packages (if any) will contain the label header. After staring at > our current packages (core, basic/advanced client workspaces, and > basic/advanced servers workspaces), I believe it makes sense to bundle > labels into the "client workspace" packages. > My reasoning is as follows: I believe most client-workspace servers do > in fact support labels, and Greg (who is known to be producing an > advanced-client-workspace server) was one of the vocal supporters of > labels. This makes sense, because although a server-workspace server > can just allocate a "workspace" when it wants to expose a set of > related versions, a basic client workspace server cannot, and labels > give it a simple way to do so. > So, I propose that we define the label feature as being part of the > basic-client-workspace package. Any objections? > This means that the proposed resolution of the issues raised since > draft 14 are: > The Variant-Control Feature: > Defer it to another spec. focused on variants. > The Fork-Control Feature: > Merge it into the checkout feature. > The Update Feature: > Keep it, and make it part of the basic-client-workspace package > (needed so that a new version created by checking in a working > resource can be made the "checked-in" version of its version > controlled resource). > The Label Feature: > Keep it as is (i.e. with the Label header), and make it part of the > basic-client-workspace package (so that a client can define a > lightweight configuration of versions on the server). > The "checked-out version-controlled configuration" issue: > Add a "checkout-on-update-and-keep-checked-out" variant to the > auto-versioning mechanism. Also, simplify DAV:auto-version > by replacing it with DAV:auto-checkout and DAV:auto-checkin. > Packages: > Define 5 packages: core-versioning, basic/advanced-client-workspace, > and basic/advanced-server-workspace. > In other words, the mailing list accepted 3 of the proposals from > the Minneapolis meeting, rejected 2 of the proposals from the > Minneapolis meeting, and accept the latest variant of the one > issue that originated on the mailing list (auto-checkout of a VCC). > Does anyone disagree/object to the resolutions described here? > If not, I'll write this up as draft-15, and submit it to the IETF > so that JimA can officially submit it to Ned for the next step > of the process. > Cheers, > Geoff -- Greg Stein, http://www.lyra.org/
Received on Tuesday, 17 April 2001 19:21:26 UTC