esme-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ethan Jewett <esjew...@gmail.com>
Subject Re: ESME-170 - PubSubHubub
Date Sat, 28 Aug 2010 08:21:30 GMT
I guess anything will work to start out and then we can refine it
relatively easily once the basic PuSH functionality is in place.

My preference would be to figure out what to do with any given feed on
our own, without involving the user (so I would prefer no special
action filter for PuSH feeds). Having a special filter for now is
totally fine, but it's a discussion to have before merging the branch.

I agree with you that we would need to poll either the feed or the hub
occasionally to make sure that the hub is still alive and we still
have our subscription active, or use some other mechanism. I think
that the PuSH spec actually includes a requirement for the hub to ping
the subscriber every X amount of time to make sure the subscriber is
still interested. Maybe this will be sufficient -
http://pubsubhubbub.googlecode.com/svn/trunk/pubsubhubbub-core-0.3.html#autorefresh

Ethan

On Sat, Aug 28, 2010 at 9:10 AM,  <in.imtiaz@gmail.com> wrote:
> Actually I think you need to do *some* polling to discover a PuSH  hub's  failure.
That defeats thew purpose of using PubSubHubub in Esme.
>
> So, I guess
>
> Onpush [default every 5 mins]
>
> is useful only because a feed may not support PuSH.
>
> Imtiaz
>
> Imtiaz
>
> Imtiaz Ahmed H E
> Cell +91.98452 84561
> Bangalore, India
>
>
> -----Original Message-----
> From: in.imtiaz@gmail.com
> Date: Sat, 28 Aug 2010 03:09:23
> To: <esme-dev@incubator.apache.org>
> Reply-To: in.imtiaz@gmail.com
> Subject: Re: ESME-170 - PubSubHubub
>
> Maybe case 2 below should be,
>
> Onpush [default every n mins]
>
> Where the polling default is optional and will be done on hub-failure only if specified.
>
> Imtiaz
>
> Imtiaz
>
> Imtiaz Ahmed H E
> Cell +91.98452 84561
> Bangalore, India
>
>
> -----Original Message-----
> From: "Imtiaz Ahmed H E" <in.imtiaz@gmail.com>
> Date: Sat, 28 Aug 2010 08:21:13
> To: <esme-dev@incubator.apache.org>
> Subject: Re: ESME-170 - PubSubHubub
>
> However,
>
> If a feed PuSHes too often Esme may be inundated with messages. So, instead
> of what Ethan concurs with below, should the user have the option to poll
> 'every 5 mins' or so even if the feed supports PuSH ?
> Also, a feed may support PuSH but the hub may breakdown. In that case should
> the user be reverted to say, a default of, 'every 5 mins' ?
>
> Maybe we should have two kinds of filter tests.
>
> 1. every n mins
> 2. onpush default every n mins
>
> PubSubHubub will be used only in 2 above, reverting to every n mins if the
> feed's hub breaks down (fails)..
>
> In 1 above, only polling will be done even if the feed supports PuSH.
>
> Let me know if this sounds okay...
>
> Imtiaz
>
> ----- Original Message -----
> From: "Ethan Jewett" <esjewett@gmail.com>
> To: <esme-dev@incubator.apache.org>
> Sent: Wednesday, August 18, 2010 10:05 PM
> Subject: Re: ESME-170 - PubSubHubub
>
>
>> Hi Imtiaz,
>>
>> I completely agree. If we need to revamp the filter syntax to make
>> this more clear, we can do that after the PuSH functionality is
>> already in place.
>>
>> Thanks!
>> Ethan
>>
>> On Wed, Aug 18, 2010 at 7:38 AM, Imtiaz Ahmed H E <in.imtiaz@gmail.com>
>> wrote:
>>> Ethan,
>>>
>>> After some initial thoughts otherwise about this, I now feel, for an
>>> RSS/ATOM feed that PuSHes ( PuSH is an acronym for PubSubHubub) we should
>>> just ignore any test (filter) whatsoever and update the user's messages
>>> whenever a feed update is PuSHed to ESME.
>>>
>>> Let me know if otherwise...
>>>
>>> Imtiaz
>>>
>>> ----- Original Message ----- From: "Imtiaz Ahmed H E"
>>> <in.imtiaz@gmail.com>
>>> To: <esme-dev@incubator.apache.org>
>>> Sent: Tuesday, August 17, 2010 12:20 PM
>>> Subject: Re: ESME-170 - PubSubHubub
>>>
>>>
>>>> Well, I need to know all the possible kinds of tests (filters
>>>> conditions)
>>>> that make sense for an RSS/ATOM action.
>>>>
>>>> Imtiaz
>>>>
>>>> ----- Original Message ----- From: "Vassil Dichev" <vdichev@apache.org>
>>>> To: <esme-dev@incubator.apache.org>
>>>> Sent: Tuesday, August 17, 2010 8:19 AM
>>>> Subject: Re: ESME-170 - PubSubHubub
>>>>
>>>>
>>>> It's the kind of filter condition which makes the best sense for this
>>>> action, at least. What do you want to do?
>>>>
>>>> Vassil
>>>>
>>>>
>>>> On Tue, Aug 17, 2010 at 5:30 AM, Imtiaz Ahmed H E <in.imtiaz@gmail.com>
>>>> wrote:
>>>>>
>>>>> Probably can figure this out, but would someone oblige and tell me...
>>>>>
>>>>> Is
>>>>>
>>>>> every nn mins rss-url
>>>>>
>>>>> the only form/kind of an rss/atom based action ?
>>>>>
>>>>> Thanks
>>>>> Imtiaz
>>>>>
>>>>> ----- Original Message ----- From: "Ethan Jewett" <esjewett@gmail.com>
>>>>> To: <esme-dev@incubator.apache.org>
>>>>> Sent: Monday, August 16, 2010 4:55 PM
>>>>> Subject: Re: ESME-170 - PubSubHubub
>>>>>
>>>>>
>>>>>> Yes, several. You can either work on that branch directly, or start
>>>>>> over by submitting a patch with a different branch created (or I
can
>>>>>> create one for you - just let me know the name you want). Create
a
>>>>>> branch locally by using the instructions at
>>>>>> http://svnbook.red-bean.com/en/1.1/ch04s02.html
>>>>>>
>>>>>> Because the branch has gotten out of sync with trunk, it is difficult
>>>>>> to diff the branch. If you want to see a diff between the branch
and
>>>>>> trunk you can do (I kid you not):
>>>>>>
>>>>>> svn diff -r985659:980193
>>>>>>
>>>>>> That actually gives a diff between those two revisions, so as soon
as
>>>>>> someone makes a new commit to either trunk or the branch it will
no
>>>>>> longer be valid. To find out what revision a branch is currently
on,
>>>>>> you go into the main directory of that branch and run (or so I read
on
>>>>>> the internet):
>>>>>>
>>>>>> svn log --limit 1
>>>>>>
>>>>>> The more manual option, but the option I prefer, is to browse
>>>>>>
>>>>>>
>>>>>> http://svn.apache.org/viewvc/incubator/esme/branches/tagandconversationfollow/
>>>>>> Based on the log entires, you can see which files I changed, and
the
>>>>>> tool will show you the change history since the branch with diffs,
so
>>>>>> you can see fairly easily what was changed.
>>>>>>
>>>>>> Ethan
>>>>>>
>>>>>> On Mon, Aug 16, 2010 at 12:18 AM, Imtiaz Ahmed H E
>>>>>> <in.imtiaz@gmail.com>
>>>>>> wrote:
>>>>>>>
>>>>>>> Ethan,
>>>>>>>
>>>>>>> Okay.
>>>>>>>
>>>>>>> Need to know:-
>>>>>>>
>>>>>>> Have you committed any changes to the pubsubhubub branch i.e.,
>>>>>>> different
>>>>>>> from what is in trunk ?
>>>>>>>
>>>>>>> Imtiaz
>>>>>>>
>>>>>>> ----- Original Message ----- From: "Ethan Jewett"
>>>>>>> <esjewett@gmail.com>
>>>>>>> To: <esme-dev@incubator.apache.org>
>>>>>>> Sent: Monday, August 16, 2010 1:01 AM
>>>>>>> Subject: Re: ESME-170 - PubSubHubub
>>>>>>>
>>>>>>>
>>>>>>> Feel free to assign to yourself. You can find my small progress
in
>>>>>>> the
>>>>>>> pubsubhubbub branch in svn. As you can see, I had not gotten
far, so
>>>>>>> if you can move it along that'd be great.
>>>>>>>
>>>>>>> If you have interim patches of your work on the branch, just
post it
>>>>>>> to jira and email the lists. I'll try to apply them right away.
>>>>>>>
>>>>>>> Ethan
>>>>>>>
>>>>>>> On Sunday, August 15, 2010, Imtiaz Ahmed H E <in.imtiaz@gmail.com>
>>>>>>> wrote:
>>>>>>>>
>>>>>>>> Ethan,
>>>>>>>>
>>>>>>>> This is currently assigned to you. I thought I'll assign
this to
>>>>>>>> myself
>>>>>>>> and work on it right away...
>>>>>>>>
>>>>>>>> What do you say, if you aren't working on it ?
>>>>>>>>
>>>>>>>> I've been looking up the PubSubHubub stuff pointed to by
Dick in the
>>>>>>>> Jira.
>>>>>>>>
>>>>>>>> The other possibility for me for the release is the
>>>>>>>> container-based-authentication stuff which Dick said recently
in a
>>>>>>>> mail
>>>>>>>> from
>>>>>>>> his vacation will need his spec'ing out on his return which
is of
>>>>>>>> course
>>>>>>>> any
>>>>>>>> day now but the spec will take a few days from today...
>>>>>>>>
>>>>>>>> Imtiaz
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>
>

Mime
View raw message