(I hope you meant to CC devl, I'm going to anyway, this should be 
public/logged).

It's a good idea. It's not practical to implement it before 0.7.0, but after 
that it's well worth exploring. I don't suppose you're a javascript guru...?

Whether it's an extension or just a block of javascript, either way a lot of 
it will be in javascript... I dunno if a plugin is appropriate, we can 
probably do what we want with js...

On Friday 28 March 2008 15:01, David Sowder wrote:
> Matthew Toseland wrote:
> > On Friday 28 March 2008 13:13, you wrote:
> >   
> >> Matthew Toseland wrote:
> >>     
> >>> On Friday 28 March 2008 12:45, Obey Arthur Liu wrote:
> >>>   
> >>>       
> >>>> I think what you found was about netscape-style plugins. It seems that
> >>>> firefox-style extensions are a very different beasts :
> >>>> <http://developer.mozilla.org/en/docs/Extensions>
> >>>>         
> >>> That's extensions, not plugins. Extensions are written in javascript. I 
> >>>       
> > don't 
> >   
> >>> know whether they are able to open TCP connections...
> >>>   
> >>>       
> >> If not, then FCP over HTTP, which extensions can presumably do.
> >>     
> >
> > Right, some custom protocol. I agree it should be possible to do this with 
an 
> > extension. I'm not sure if that will adequately solve the 
> > browser-history-snooping problem though? I've been told an extension can 
> > remove stuff from the browser history ... Also won't the performance be 
poor, 
> > since we'd have to poll constantly? Or could we have a single connection 
> > which fetched a text file which included status updates, and handle them 
live 
> > as they come in?
> >   
> The performance wouldn't be much different than normal FCP.  It would be 
> a lot like AJAX from a performance perspective.  A single connection 
> could probably be used.  We could use once a second or so polling for 
> anything new on the FCP virtual session (the session would be based on a 
> UUID or something tying the polling HTTP request together as an FCP 
> session rather than a single TCP socket as we do elsewhere).  The FCP 
> over HTTP stuff would probably be a relatively simple wrapper class 
> around FCP on the node's side.  The "hard part" would be the FCP 
> implementation in Javascript since it doesn't already exist.  That 
> implementation could be written generically and then the HTTP stuff 
> written as a "backend transport" such that the Javascript FCP library 
> could be reused elsewhere as well.
> > This might be a good SoC project. I've mentioned it on the SoC ideas page.
> >
> > It would be really nice if each activelink on FAI showed a loading 
animation 
> > with a progress bar until it actually loads...
> >   
> 
> 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: 
<https://emu.freenetproject.org/pipermail/devl/attachments/20080328/ce50d7ae/attachment.pgp>

Reply via email to