Re: Client comments?

Elias Sinderson wrote:
> 
> bugzilla@soe.ucsc.edu wrote:
> 
>> ------- Additional Comments From elias@cse.ucsc.edu  2005-12-02 09:34 
>> -------
>> Rough consensu reached on the 12/2 telecon wrt Julians proposed text, 
>> modulo
>> some minor tweaks to the language ('clients will...') and remaining 
>> nit about
>> handling of comments. Elias to raise issue regarding the handling of 
>> comments on
>> the list.
>>
> Right, so regarding the handling of XML comments in property values...
> 
> My 10,000 ft. understanding of the issue is as follows:
> * It would generally be preferred if comments were preserved
> * Currently (most? all?) servers do not preserve comments

I think all I tested recently.

> * The relevant XML specs allow for comments to be dropped by parsers and 
> they do

...allows them to be dropped by parsers, yes, but I don't think many 
parsers do so. It's the servers who drop them.

> Based on the above, it seems clear that 2518bis cannot require comments 
> MUST be preserved, although the WG does want to say something that will 
> encourage servers to attempt to do so. It would be nice to hear some 
> input from client (and server) implementers on this issue. Do note that 
> the obvious workaround for clients that want to force servers to 
> preserve comments should put them in as CDATA, which will be preserved.

...which will make those property values behave extremely strange in 
general purpose clients (will display escaped XML including the comment) 
or with things like SEARCH.

> Open questions for further discussion:
> * Do clients want or expect comments to be preserved?

We do not.

> * Would it be acceptable to server implementers for 2518bis to define 
> the preservation of comments as a SHOULD and will they make the effort? 

If there's support for that requirement, we can implement it.

> Furhter, is this even possible given the libraries they are using?

I think: in general yes, but there's no guarantee that every XML parser 
will do it. This is more likely to be a problem in size-constrained 
devices though (where there may be a "minimal" XML parser installed), 
not on servers.

> * Will client implementers accept wording to the effect that servers MAY 
> NOT preserve comments and, if they are truely required, mention the 
> above workaround?

"MAY NOT" reflects the current state of the art.

> * Should 2518bis support different requirements for live vs. dead 
> properties? This seems to make sense...

I'm not sure how the distinction between live and dead properties is of 
any relevance here (and thus confusing). In particular, the whole issue 
only applies to things you can PROPPATCH, right?

Best regards, Julian

Received on Thursday, 8 December 2005 09:02:29 UTC