Re: follow up on service workers for publishing platform

> On 1 Dec 2015, at 18:03, Daniel Weck <daniel.weck@gmail.com> wrote:
> 
> On Tue, Dec 1, 2015 at 4:50 PM, Ivan Herman <ivan@w3.org> wrote:
>> 
>> I understand this for Acme. However, thinking it further in direction of a
>> more sophisticated reader: is it necessary to store all the file references?
>> After all, I would expect the Service Worker may find out on the fly that a
>> resource is to be cached.
>> 
>> This may be important for the production site. This means I do not have to
>> list the references of all the images, js and css files, etc, just the 'top
>> level' files to trigger the process and those could be cached.
>> 
>> Of course, there is a danger that the reader would cache too much, eg,
>> remote references that the content refers to. So maybe the manifest could
>> contain some sort of URI patterns, saying to the reader: "if the URI matches
>> one of these patterns, cache it".
>> 
> 
> I would assume that all of the reading system's app resources would be
> cached "immediately" (i.e. as early as possible in the bootstrap
> process), so that the reading system itself is available offline.

Absolutely.

> However, publication content would be cached according to a particular
> strategy (on-demand vs. preload, LRU eviction, etc.), in terms of
> offline-ing multiple publications, but also in terms of offline-ing
> resources within a given publication (partial fetch vs. full cache).

Right. But, additionally, I think providing some level of content provider option may be necessary (inclusion patterns, exclusion patterns, etc.).

We used this example before: a publication may refer to very large font files. In some cases, those fonts are for fanciness; ie, it is perfectly o.k. if the reader does not cache those and, in case it is offline, falls back to some system fonts. In other cases those fonts are essential because, for example, they display mathematical symbols or musical notes: in this case the font must be cached, too, for proper offline use. This cannot really be covered algorithmically; it is up to the creator of the publication to control this.

Ivan


> Dan


----
Ivan Herman, W3C
Digital Publishing Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
ORCID ID: http://orcid.org/0000-0003-0782-2704