Coments on the Google API blog worht to read

#

Greg Linden on December 26, 2005 at 5:39 PM wrote: #

This is pretty cool, Niall. Nice work tracking all this down.

But, is it really true that people can build on this?

This was undocumented by Google and seems to be intended for their own apps. Using this seems risky, I'd think, with high likelihood of getting shut down if you built anything that was available publicly or generated noticeable levels of traffic.
#

[TypeKey Profile Page] mparaz on December 26, 2005 at 6:40 PM wrote: #

My educated guess:

Google will open up the Google Accounts login infrastructure. Once logged in, you can make these calls legitimately.

For non-logged in users, there could be an server-side HTTP equivalent of the Maps API (which is for Javascript rendering)

Nice work, pitting G against Y! in the race for openness!
#

[TypeKey Profile Page] Niall Kennedy on December 26, 2005 at 7:52 PM wrote: #

I think Yahoo! and Google are interested in opening up their feed backend in the same ways they expose maps data and other information. The first step is to expose the data, ask "why not?" and allow other developers to express their interest in using the data.

I sent a note to Scott Gatz, My Yahoo! lead, last week but I have not heard back from anyone at Yahoo! about my previous post. I also sent off an e-mail to Jason Shellen, Google Reader PM, this evening inviting conversation.

Google Maps API came after the JavaScript had been reverse-engineered and posted across the Internet with significant interest. I'd like to start a conversation about feed APIs and generate some interest in the data before a formal program is announced.
#

[TypeKey Profile Page] Chris Wetherell on December 26, 2005 at 10:31 PM wrote: #

Hi, Niall. Nice work on the API analysis! I'm certain Jason would be happy to jump into the conversation - I can as well, if that's all right.

So, to clarify. To date the API has been undocumented by us to date to help prevent those oh-my-gawd-the-url-changed headaches that would've happened if we'd changed something. So developers should note that some of the URLs might still change but that we're getting close to stability and we're happy for anyone to build atop the framework who is interested (and doesn't find managing auth issues daunting) as that's been our goal and intent with the API for Reader.

Other neat stuff I didn't see listed here:

* There's also a way to exclude tags/labels by appending another query to the URL as xt=[tag]. For example, you could exclude read items from some collections of items by appending: ?xt=user/[000...]/state/com.google/read
* Some tags include specific domain information because we've been planning on an approach that's flexible for other people's (read: non-Google) uses.

And I can confirm that you're right about the lens - it's only one of several planned approaches. Based on a series of user studies during early development we've been planning from the start a system that can support several distinct UIs for feed reading that would be developed by us - I actually have two other views on my dev build of Reader - but am more excited that others would be able to build clients that work best for them. One of the most exciting ideas for me for any feed reading API is that we (and perhaps other big companies) could possibly work on the more non-obvious and experimental solutions for syndication transformation since third-party developers are going to be excellent at supplying oft-requested and well-understood modes of interaction.

A short (and incomplete) list of stuff to do includes:
- Passing media enclosure data
- Atom 1.0 compliance (if you notice we're getting close)

Some of the other Reader team might post here, too, as everyone's been directly engaged with the API - it's been a chief part of the discussion, planning, and design for this li'l project. After we finalize the remaining bits we'll definitely post to our blog and document the framework.

- Chris Wetherell
Google Reader Engineer
#

Mark on December 27, 2005 at 5:42 AM wrote: #

For Atom 1.0 compliance, I recommend my own unit tests.
#

ihavenoname on December 28, 2005 at 7:29 AM wrote: #

What happens when you retrieve all unread items? Do they all get marked read? If that's the case then the API is pretty much useless from a synchronization perspective. If I try to sync to my handheld and then only read some of the items (typical usage scenario) and then try to continue reading on the web there will be a problem because items that I have not read in actuality have been marked as read. I don't see any facility for marking individual items read/unread.
#

Laurence Gonsalves on December 28, 2005 at 11:23 PM wrote: #

Items are not marked as read automatically, so it's possible to fetch an item without reading it. You can mark them as read using the same facility that's used for starring. Look for "flag or star a post" above. Just replace "starred" with "read" when passing the "a" parameter.

By the way, the comment "No cookies or session IDs are required to access this member-specific data" is incorrect.
#

[TypeKey Profile Page] Niall Kennedy on December 29, 2005 at 12:07 AM wrote: #

Thanks Laurence. I rerequested all URLs without an SID cookie and Google returned a status code 403 and "permission denied" in the entity.

I updated my original text to reflect this information.
#

jcburns on December 29, 2005 at 8:18 AM wrote: #

The value of my SID cookie from Google.com is a string that is 161 characters long and starts with
DQAAAGgAAABRnt... ...so where is this 20 digit value?
#

[TypeKey Profile Page] Niall Kennedy on December 29, 2005 at 8:37 AM wrote: #

jcburns,
Google assigns each user a 20 digit user ID that would not change over time, and a session ID (SID) you can delete, regenerate, and expire.
#

RichB on January 2, 2006 at 8:56 AM wrote: #

Notice the Google RSS->Atom conversion is recursive. eg:
http://www.google.com/reader/atom/feed/http://www.google.com/reader/atom/feed/http%3A%2F%2Fdel.icio.us%2Frss%2Fpopular%2F
#

Nick Lothian on January 11, 2006 at 10:24 PM wrote: #

You may be interested in some futher reverse-engineering I did, and a Google IG module I build on top of the Reader API:

http://www.mackmo.com/nick/blog/tech/2006/1/5/Google-Reader-Homepage-Module.html
#

[TypeKey Profile Page] Andrei Lopatenko on February 9, 2006 at 2:56 AM wrote: #

2 Chris Wetherell

Hi, Chris
is there any information which kind of Google Reader API is going to be provided by Google?
Will you support a streaming model (event-based)?
I am curious since i like to develop intelligent interfaces and I am RSS-addict. I am quite unhappy about flexibility and search capabilities of many RSS-reader and I want to develop a query interface for Google Reader
#

Steven on February 14, 2006 at 3:42 PM wrote: #

I hope the xt=[tag] functionality makes it into lens soon. The ability to show only things NOT tagged nsfw is a vitally important feature.
#

Mike on February 15, 2006 at 4:56 PM wrote: #

One more comment.... Has anyone figured out how to get a complete list of unread entries for all feeds you are subscribed to?

This
/reader/atom/user[user id]/state/com.google/reading-list

seems to just give me a copy one entry from each feed I am subscribed to. I am looking to get a feed of all entries from all feeds (read or unread).
#

Sérgio Nunes on March 2, 2006 at 3:13 PM wrote: #

Hi,

It seems that URL based authentication works. Try: http://[username]:[password]@[url] Replace [username] and [password] with your Google Reader data and [url] with one of the above urls.

This is great for server side development.

Regards!
#

Kannan on March 8, 2006 at 9:34 PM wrote: #

I am having some problems with google reader, although i must say that the google reader is extremely fast.

1. I want an entire feed to be labelled for example, rugby.com/atom.xml should be labelled as sports and etc. This would mean that I would read my "tech-blogs" labels first and then others.
2. Although it is not very desirable like the above thing, it would be great if we can have rules like gmail to categorize stuff.

Thanks (And as usual Google Rocks)
#

[TypeKey Profile Page] mparaz on March 10, 2006 at 6:54 PM wrote: #

It looks like even retrieval requires a login now. I was hoping to use this backend for an AJAX login-free aggregator.
#

[TypeKey Profile Page] mohan on March 14, 2006 at 11:30 AM wrote: #

Hi,
can you please tell about the user id in the URL
/reader/atom/user/[user id]/pref/com.google/subscriptions

wher can i get it?

thanks
mohan
#

Nitin Nanivadekar on March 31, 2006 at 9:53 PM wrote: #

Thanks a million
i am building an app that user feed reader as an auxiliary function.
This fits my bill perfectly.

-nitin nanivadekar
#

Darren Kulp on June 5, 2006 at 6:13 AM wrote: #

@mohan, jcburns:
View source on your Google Reader page and search for _USER_ID.
#

Jamsi on July 31, 2006 at 1:53 AM wrote: #

I'm looking at creating a feed reader that works with the Google reader API (mainly so I can download a list of items to read on the train ..) but has anyone made such an application?

I can't imagine it to be too hard.
#

Muzaffer Coruh on August 4, 2006 at 4:29 PM wrote: #

Hi,

The number of returned results may be adjusted using the n parameter."

this is really nice but I wonder which parameter is used to call next 20 items...
#

Mark on August 4, 2006 at 5:46 PM wrote: #

This information is great. Thanks so much. I'm having a little trouble understanding how to star or add tags to entries as these links seem to be hidden somewhere deep inside the javascript, css, or otherwise. Can you provide any further information that will help me piece this together?

I'm working on a pre-reader that will filter duplicate news items from different sources as well as other unwanted content, since this doesn't seem to be in lens yet. This page has been a great help. Thanks!
#

Yasha Davidov on November 5, 2006 at 2:39 AM wrote: #

Just few things more:
1. You can access only unread items using ttp://www.google.com/reader/atom/user/[user ID]/state/com.google/unread

2. You can access all items via http://www.google.com/reader/atom/
(i suppose you must be loged in)
#

Yasha Davidov on November 5, 2006 at 12:46 PM wrote: #

It seems that com.google/unread always gives you an empty list.
That was my mistake.

But you can omit user id and replace it with "-". If the user is logged in this will work.
#

Rick on November 22, 2006 at 10:38 AM wrote: #

There is a way to request the number of unread feeds?
#

Venu on November 29, 2006 at 2:43 PM wrote: #

For shared items:

/reader/atom/user/[user id]/state/com.google/broadcast
#

Edwin Khodabakchian on January 22, 2007 at 11:22 PM wrote: #

Hi,
Does anyone know if there is a way to apply multiple tags to a feed as part of a single HTTP request. I have tried the approach of calling multiple times the request which adds a single tag to a feed and I think that I am running into a concurrency bug because I do not get a consistent behavior from the backend.
Thanks,
Edwin