Looks good to me!
On Jan 8, 9:43 pm, Antoine Imbert <[email protected]> wrote: > I pushed a new version that moves the hub.verify_token in the callback path. > The main benefit is that the token is now available when a ping is received. > > (There is a config flag to disable this and keep the token as the > hub.verify_token query argument.) > > https://github.com/ant0ine/Plack-App-PubSubHubbub-Subscriber > > This is now closer to the spec changes recently proposed. > Let me know if you want to move forward with the 0.4 spec, I'll be > happy to help. > > Antoine > > On Thu, Dec 15, 2011 at 8:24 PM, Antoine Imbert > > > > > > > > <[email protected]> wrote: > > Hi there, > > > As Julien noticed, I’ve put that on Github yesterday: > > > Plack::App::PubSubHubbub::Subscriber > >https://github.com/ant0ine/Plack-App-PubSubHubbub-Subscriber > > > This is a perl subscriber client based on Plack. I currently use it > > with Starman and I successfully subscribed to a few thousand feeds. > > > Feel free to fork it and to give me your feedback. > > > BTW, here is mine about the recent spec change discussions: > > > The verify_token removal: > > This client supports the verify_token, if it’s removed from the specs, > > I think I’ll just move the token to the callback path, and the API of > > the client will stay the same. Note that removing verify_token > > simplifies the specs, but not the implementation of the subscriber. > > > The arbitrary content: > > I like that a lot, and this client makes no assumption about the > > content, it just gives you the content-type and the raw content. Also, > > if the arbitrary content does not contain the topic, then the > > verify_token included in the callback path makes a lot of sense, it > > allows you to associate a ping with a subscription. > > > I think the verify_token could be removed from the spec, and that > > adding a token or id in the path of each callback url should be > > described as a good practice. That would pave the way for arbitrary > > content pings. > > > Antoine
