Re: [ppsp] Firefox ppsp URL support

Hi Dave,

Good news that there is some motion on this. I think I can contribute.  
I've had some experience implementing protocol handlers as a Firefox 
extension.  I wrote one to handle the NetInf ni protocol which has a 
hash-based URI similar to ppsp, and helped write one for the ndn 
protocol which has some more complex TCP communication behind the 
scenes.  These are able to render an entire web site in the space of 
documents referred to by URIs with the special protocol.

> I would imagine it will need something like http://www.w3.org/TR/app-uri/

This uses a universal app: URI with an application-specific UUID which 
has to be in every URI, for example app:<UUID>/<hash>. You prefer this 
to using the ppsp: protocol directly, for example ppsp:<hash> ?  This is 
how the examples look on ppsp.me:
http://www.ppsp.me/start.html

> I’m not sure how mime types etc would be transported

The NetInf people have thought about this when they wrote the "Naming 
Things with Hashes" RFC: http://tools.ietf.org/html/rfc6920 . They 
propose putting the content type in the URI, for example 
ppsp://2b2fe5f1462e5b7ac4d70fa081e0169160b2d3a6?ct=video/mpeg

Thanks,
- Jeff

On 2014-03-10 14:07, Dave Cottlehuber wrote:
> On 08. März 2014 at 17:38:46, Jeff Thompson (jeff@thefirst.org) wrote:
>>
>> Hello Dave,
>>   
>> A related question. What are the plans for Firefox to support ppsp
>> URLs, so that an HTML page can have, for example
>>   
>> and Firefox (or an extension) will use PPSP to fetch and display an
>> image (or other content type)?
>>   
>> Thank you,
>> - Jeff Thompson
> Hey Jeff,
>
> I would love to see this too. I had a brief chat with some IETF folk about the right way to organise this; it requires a bit of coordination across Mozilla, IANA, and a few other places. I’d love to have some help on this, atm I am still working on the Erlang peer before I get stuck into the javascript one. If there’s interest in doing this I am happy to give input based on what I’ve learned so far.
>
> One of the other crazy ideas I had was that, as the leaf nodes don’t have to be identical chunk size, one could in theory cram an entire web site into a hash tree simply by putting the files into the appropriate lower level chunks with some structuring info, for example, I’m not sure how mime types etc would be transported, and there are a few other hitches, but in *theory* it’s all achievable :).
>
> I would imagine it will need something like http://www.w3.org/TR/app-uri/ and these are defined by http://www.w3.org/TR/uri-clarification/ which refers to http://tools.ietf.org/html/rfc2717 and http://tools.ietf.org/html/rfc2718 ; I have had some offers of assistance in working through all this, with a recommendation that getting a few of the core browser developers onside with the proposal before doing all the paperwork.
>
> Currently my priority is to finish the Erlang PPSP code first :-).
>
> A+
> --
> Dave Cottlehuber
> Sent from my PDP11
>
> _______________________________________________
> ppsp mailing list
> ppsp@ietf.org
> https://www.ietf.org/mailman/listinfo/ppsp