[Standards] Proposed XMPP Extension: References

Kevin Smith kevin.smith at isode.com
Thu Mar 24 10:14:28 UTC 2016


> On 21 Mar 2016, at 14:14, Dave Cridland <dave at cridland.net> wrote:
> 
> 
> 
> On 21 March 2016 at 14:06, Kevin Smith <kevin.smith at isode.com> wrote:
> 
> > On 21 Mar 2016, at 11:48, Daniel Gultsch <daniel at gultsch.de> wrote:
> >
> > An update to the 'arbitrary attributes' feature request. Even more important than the before mentioned content-type, content-length and so forth are probably description and a thumbnail. Information which can be extracted for example by the means of open graph [1] meta tags from a lot of websites.
> >
> > This would allow for a proper link preview like slack, skype and so forth are doing.
> >
> > Kev: When questioning this. Is your concern that you don't want the attributes to be completely arbitrary? Do you want them to be well-defined? (Having a set list of like content-type, description, thumbnail, content-length and so forth?) Or would you rather not have attributes like this in this XEP at all?
> 
> It’s a concern with the attributes being completely arbitrary (how does one know which they need to support? discovery starts being a bit of a nuisance). Non-abritrary data form elements could be reasonable - there’s still a discoverability question, but it might be ok. Would you like to propose something? I’ve not had the cycles since the summit to take much more of a look at References.
> 
> 
> I don't think there's any fundamental difference between arbitrary namespaced child elements and a data form, except that a data form is more restrictive and lends itself well to human interaction. I don't think we have any human interaction here. So why a data form and not just allow namespaced child elements, which can be ignored by recipients that don't understand them?
> 
> (Note that I often dislike data forms in machine-driven protocol work, but this case seems particularly unsuited).

This is entirely fair, and so do I.

/K


More information about the Standards mailing list