Re: ETags, next call, was: Notes on call from today ...

On Nov 29, 2005, at 11:49 AM, Lisa Dusseault wrote:

> I hadn't thought that a client could use weak etags at all for  
> authoring.  I had started from considering the cases of a JPG or a  
> Photoshop file or a Word document.  What does a weak ETag mean for  
> those documents?  If I edit a Word document on a WebDAV server and  
> all I do is fix a couple inconsistent fonts and a space or two, one  
> might think that the weak ETag need not change -- but I would not  
> find the system very usable if these changes were lost on a  
> subsequent update.

   As Julian said, the meaning of semantically equivalent is  
important.  The XML rewriting debate with respect to properties we've  
been having is an example of why it's important to define, and  
presumably a similar debate could be had as to whether such rewriting  
of a XML resource body would lend itself to a weak etag match.

   In the case of CalDAV, the iCalendar spec would be the authority  
on semantic equivalence.  If iCalendar says that the order of  
iCalendar properties and subcomponents is semantically meaningful,  
then they need to be preserved.  (Presumably the iCalendar parsers,  
storage, and generators you use would know this and preserve the  
order.)  If not, then for the purposes of iCalendar, changes in order  
are semantically equivalent, and we could probably agree that for  
CalDAV such changed in a <calendar-data> element are equivalent by  
extension.

   It should be noted that we've already agreed that normalizing out  
newlines in iCalendar data within XML is OK, and that variation would  
be semantically equivalent as well as any variations considered  
equivalent in iCalendar.

   CalDAV should probably be explicit about both of these equivalence  
measures as they related to etags.

	-wsv

Received on Tuesday, 29 November 2005 21:13:22 UTC