Re: Evolving AppCache discussions

On Sep 25, 2012, at 9:08 AM, Robin Berjon <robin@w3.org> wrote:

> 
> I believe that the discussion on the logistical details of how to organise this should not block talking about other specifics.
> 
> My first question is how do you see this work taking place at the editorial level? By this I mean that it could be placed in the HTML.next draft as a series of changes to the current model, or the current model could be ripped out (in which case we would ship 5.0 without it, or at the very least mark it as at risk — it's rather broken anyway) and put in an extension spec.
> 
> The AppCache feature is rather intricately embedded into the current spec and therefore somewhat difficult to extricate, but as we all know things that are impossible just take longer. There may be a middle ground in which some of AppCache's interactions may remain in the spec as hooks while the concrete mechanism is left to the extension spec (not a wonderful approach, but I want to list the options).
> 
> The pros of having a separate spec are that it can ship on its own timeline, and can perhaps more easily be the subject of focused work by the group you are proposing to assemble around a separate list. The cons are extra editorial work in the short term (which I reckon we can figure out) and potential mismatches going forward (which I would hope we can avoid by virtue of being implemented by exactly the same people).

Speaking as an implementor:

AppCache has shipped in multiple browsers, and is in active use by some sites, despite its known issues. I am not sure it can be dropped from implementations at this point, or changed incompatibly. Therefore, I personally would prefer to see new work as a compatible extension (preferably) or side-by-side alternative (worst case) to the current AppCache. One possible way to do make it a backwards compatible extension is to define a v2 manifest format. Then sites using the current manifest format would continue to work fine. Either of these approaches would let us avoid ripping anything out of HTML 5.0. I do not see how any of the issues raised fundamentally require breaking backwards compatibility.


I have also heard offlist that some may find Web Apps WG a better venue for this work. I note that Web Apps WG has worked on an AppCache extension in the past, though the work was abandoned: <http://dev.w3.org/2006/webapi/DataCache/>. I think it would be reasonable to do such work in Web Apps. I think it would be reasonable to do it here too.


Regards,
Maciej

Received on Tuesday, 25 September 2012 17:50:01 UTC