For as long as I can remember, the API to use web services on groups has been "Coming Soon". When can we expect to see that API implemented, at the very least to download our group bookmarks? I'm going to be uncivil and tag this as a bug.

I have no choice but to label this problem a bug. Sure, it could be inaccurate use of the export API, but the documentation suggests otherwise.

I am using rows=x to signal how many rows I want to download.
I am using start=x to signal the row from which downloading starts, using a 0-based counter, row 0 being the first row. That seems to work fine when exporting all 400+ bookmarks in quantities of 10 rows, stopping when either a null is returned in the query or that less than 10 rows came back from a query.

Jump ahead to using the API again later to refresh, after more bookmarks have been added. More rows. I use the row count from the previous export (e.g. 499) as my new start row. What Diigo returns is something exported earlier from a lower row.

Specifically, I asked for export with start=499
What I got back was something I exported the day before at row 496

I'd really like to hear from the Diigo programmers on either the nature of a new bug, or an explanation of how to use the export API in a way that prevents this from happening.

I have a new theory. It appears that the explanation for this is that Diigo treats rows as a stack, last one in on top (row 0). That explains why adding a couple of new bookmarks pushed old ones further down. So, my new theory is that I must put in the query sort = updated_at which, I suspect, would sort the rows on update date always giving me (I'm just guessing here; the documentation doesn't suggest otherwise) that last one in will be last one out.

Sorting returns last first. That's not helpful. This becomes an open question: how does one update an export without seeing previously-exported bookmarks?

I do an export of all my bookmarks until none are left. Then I add another bookmark and return to export what's new. I show in the first export that I exported a count of 497 rows. So, I begin my next export as follows:
bookmarks?users=<...>&rows=50&start=497

What is returned is not the lone bookmark, but rather several, starting from, according to the first export, row 449 and progressing forward from there. I am not sure what this suggests, other than the distinct possibility that I'm not using the API correctly.

Thoughts?
Thanks, Jack

More information with a more controlled test.
Reduced rows to 10 at a time.
When it hits thinks it is retrieving row 498, row 490 is retrieved (again), and so on for 8 more rows.

Here is what I see. In theory, I am asking for 10 rows at a time, but a query started at an interval that is not on a multiple of 10. The previous query started at 490. It appears that only 8 bookmarks were returned, so the next query started at 498, which, it would seem, is out of bound. I should have received an empty result, but didn't. Instead,Diigo appears to have begun recycling the previous query.

Jack

When I use the API to read list my bookmarks, I need to understand the following:
1- will the bookmark include all users who have tagged the same resource or must me?
2- will the bookmark reflect all tags or just mine?
3- will the bookmark reflect all comments and annotations or just mine?

If the answer is no to any or all of those, is there a way to export beyond just my own bookmarks?

Thanks in advance

Exporting using the API worked fine a few weeks back; now all I can get is the 502 Bad Gateway message. Get it both from FF 3+ address bar and from software I wrote that worked fine earlier. What happened?

Magic, I suppose. Gotta wonder why it was down.
Just noticed something changed. When I first exported my data, I had date fields that look like this:

"2008/07/24 18:28:09 +0000"

Today, I got date fields that look like this:

2008-08-24T01:29:20Z
which broke my code.
Can I assume that will be the final way dates are formatted?

Many thanks

Ken Wei wrote:
> it works now
>
> Jack Park wrote:
> > Exporting using the API worked fine a few weeks back; now all I can get is the 502 Bad Gateway message. Get it both from FF 3+ address bar and from software I wrote that worked fine earlier. What happened?

Swell!
This is somewhat disconcerting: I'm getting exceptional service here. Will it always be like that? I already patched my code to handle the, um, new date format. Guess I'll have to patch it back?

Ken Wei wrote:
> fixed, back to format: "2008/07/24 18:28:09 +0000"
>
> Jack Park wrote:
> > Magic, I suppose. Gotta wonder why it was down.
> > Just noticed something changed. When I first exported my data, I had date fields that look like this:
> >
> > "2008/07/24 18:28:09 +0000"
> >
> > Today, I got date fields that look like this:
> >
> > 2008-08-24T01:29:20Z
> > which broke my code.
> > Can I assume that will be the final way dates are formatted?
> >
> > Many thanks
> >
> > Ken Wei wrote:
> > > it works now
> > >
> > > Jack Park wrote:
> > > > Exporting using the API worked fine a few weeks back; now all I can get is the 502 Bad Gateway message. Get it both from FF 3+ address bar and from software I wrote that worked fine earlier. What happened?

I'm attempting to authenticate with the Apache HTTPClient and keep getting a 401 not authorized. Any hints? I can run the api from my browser just fine.
Thanks, Jack