gmane.comp.web.pinboard.devel

[pinboard-dev] Don’t remove the URL from imported tweet text

Michal Migurski — 2013-05-19T16:18:48

I import tweets and favorites from Twitter to Pinboard. Pinboard removes 
the URL from the flow of the text, which typically results in a choppy, 
senseless description. Sample tweet: “I just visited example.com and it was 
great." Resulting Pinboard description for example.com: “I just visited 
 and it was great”. It’s be better to leave the text of the tweet as-typed, 
URL and all.

Grouped Bookmarks (or More Powerful Tab Sets)

Zhang Jun — 2013-05-16T04:32:06

Here's my *problem*:

Sometimes when I research a topic, I turn into a maniac and open a lot of 
URLs. To save them to Pinboard, currently I have 2 available options:

1) Save them as a tab set, BUT I can't add notes or tags or search them 
with a keyword (I think tab sets are only suited for cases when you don't 
have time to close read and not suited for research scenarios such as mine);

2) Save them individually as normal bookmarks with a common tag signifying 
the research topic, BUT it is very tedious and if I want to refer to them 
later, I have to select the assigned tag to get a full view of all these 
bookmarks (and their notes).

My *suggestion*:

Along with bookmarks, notes, and tab sets, let's have a 4th category:*grouped bookmarks
*, where one can batch save urls just like tab sets but has the added 
option to assign notes and tags to each url in the set. 

Grouped bookmarks should display in the* normal stream* of bookmarks, 
unlike tab sets tugged away separately. A button can be added to *
c

Auto suggest considering other public bookmarks of the same address/site

Carlos K. — 2013-05-15T21:11:56

Hello.

I'm strongly considering joining Pinboard as it seems a better option to me 
when compared to Delicious (or whatever it became during the last few 
years).

I read the whole features overview page and noticed the auto-suggest tags 
feature seems to be based solely on my own bookmarks. I also searched 
through this forum but couldn't find any similar suggestions. So, here it 
goes...

One feature that used to work very well on Delicious (some good few years 
ago though) was that it also analyzed *public* bookmarks from other users 
to come up with suggestions when adding/editing your own. I assumed it 
based suggestions on identical bookmarks and possibly from the same domain 
as well.

I understand Pinboard is WAY more concerned with privacy but it seems to me 
that that in this case there's no issue at all as users are able to mark 
both bookmarks and tags as private. This feature is one of the very reasons 
why "social bookmarking" has benefits when compared to private.

Anyway, just wanted to ask

Re: t.co links truncated from twitter

Kat — 2013-05-11T09:52:28

Hm, not sure what happened to my first message.  Anyhow it looks like it is 
still happening, here's an example from 
today: https://pinboard.in/u:wirehead/b:8a54d10879a8

--Kat

On Wednesday, May 1, 2013 6:17:26 PM UTC-7, maciej wrote:

related tags in search

vayan — 2013-05-08T14:11:58

When I search for a tag I don't get the "related tags" in the sidebar to 
refine my search. It could be cool.

Cheers 

- vayan

Re: t.co links truncated from twitter

Kat — 2013-05-10T21:48:30

Is this fix live yet?  I've noticed it's still happening.  I don't have an 
example handy since I've been trying to fix them as I see them, but I will 
watch for one and post it for examination.

FWIW, this also happens with my auto-post of RTs to Facebook, so I wonder 
whether it's not a problem on the Twitter end.

As a workaround I've been fixing the broken links manually, but oddly I 
have a couple of broken links I can't edit.  I click "edit", change the URL 
to the correct one, and hit save... but the URL keeps going back to the 
broken t.co link.  It's weird because this has certainly worked on some (or 
most) of them.

Example broken/unfixable links: 
https://pinboard.in/u:wirehead/t:broken-links/

Thanks!

--Kat

On Wednesday, May 1, 2013 6:17:26 PM UTC-7, maciej wrote:

'Suggested tags' includes deleted tags

Meredith Shepherd — 2013-05-04T04:04:17

A couple of people have posted saying they'd like to see other people's 
tags in the 'suggested tags' list, but I'm having the opposite problem: I'm 
getting suggestions for 'recommended' tags that other people are using, but 
that I'm not using anymore.  I recently changed my previously inconsistent 
tagging system for easier organization (for instance, "f:dccomics" and 
"dcu" are now renamed to "fandom:dcu"), and now I'm finding that every tag 
I've used in the past is eligible to show up in my suggested tags list.  
So, now, when I go to bookmark something Batman related, my suggested tags 
list will look like: "batman c:batman char:batman character:batman dcu 
f:dcu fandom:dccomics fandom:dcu ...", even though most of those tags are 
no longer in my tag list.  
I can see the advantages of remembered tag suggestions for, say, someone 
who uses their pinboard for to-do items and deletes bookmarks often, but 
for my part, I'd like to be able to tell pinboard to 'forget' my deleted 
tags.  Is there a workaro

Re: Draft spec for API v2 - please comment

maciej — 2013-05-04T22:23:08

Thanks to everyone who commented on the draft API spec.  Here are the
changes I'll make based on your feedback:

- allow HTTP Basic Auth with the API token
- add an API call for fetching API token with a username/password pair
or website auth cookie
- remove X- from HTTP header names
- all API errors will be reflected in the HTTP status code, using 400
Bad Request as a generic catchall
- success field in the JSON envelope will have a value of 'true' or
'false' rather than 'ok' and 'not_ok'
- add pubkey and signature to API docs

Some things I'm still mulling over:

- CORS or JSONP support
- What to name the throttling headers.

I'll post some actual proposed API calls next. Of course discussion is
still welcome on any aspect of the API reboot.

Adding a note with an empty "note" textarea

Olga Sayenko — 2013-05-02T08:32:32

Sometimes the title and tags is enough, but now you get an error if "note" 
textarea is empty.

Also, just want to let you know that I absolutely love your service - even 
better than the original delicious. Thank you very very much.


Olga

Re: t.co links truncated from twitter

maciej — 2013-05-02T01:17:26

I found a bug that was affecting retweets.  Basically they would be
truncated at 140 characters even though the 'RT < at >user:' prefix made
them longer.  This tended to cut off any trailing links.  I've fixed
the error; please let me know if you see it happen again.

On May 1, 8:17 am, sheila miguez <she...< at >pobox.com> wrote:

t.co links truncated from twitter

sheila miguez — 2013-05-01T15:17:06

Hi all,

I mentioned this a while back, and was wondering if anyone is working
on it. My retweets are intermittently truncated.

I retweeted this to bookmark it.
https://twitter.com/ChicagoCDO/status/329605205647835137

It was bookmarked as t.co.
https://pinboard.in/u:sky/b:93ff25b767f2

Since I often miss that this has happened, when I go back to find the
real link,  I may not find the original tweet. This makes me think it
might be useful to bookmark the permalink to the tweet as well.

I prefer first to have the untruncated links if possible.


--
sheila

Re: Draft spec for API v2 - please comment

maciej — 2013-04-30T18:54:21

I don't see a reason not to let people trade auth cookies or their 
credentials for the API token.  I'm sympathetic to your plight.

As far as oauth goes, I came close to implementing it, but got scared off 
by the complexity and the unfortunate resemblance to phishing.

On Monday, April 29, 2013 10:45:39 AM UTC-7, Stephen Darlington wrote:

Re: Draft spec for API v2 - please comment

maciej — 2013-04-30T18:50:24

For some reason I'm deeply ambivalent about making the API friendly for use 
from within web apps. I guess I need to look deep within myself and figure 
out why I have such a server-side bias.

On Monday, April 29, 2013 9:35:56 AM UTC-7, Les Orchard wrote:

Re: Pinboard for private teams?

maciej — 2013-04-30T18:48:57

There's nothing to report.  It's still a feature I would like to implement, 
but I don't know when I'll have an announcement to make about it.

On Tuesday, April 30, 2013 4:01:33 AM UTC-7, Simon Willison wrote:

Pinboard for private teams?

Simon Willison — 2013-04-30T11:01:33

I'd love to be able to use Pinboard to privately share links within my 
company. In particular:

- Have a team consisting of a group of individual users
- Share bookmarks public, private or with a specific team (or several teams)
- Be able to browse everything available to a team by tag
- See a feed of new links that have been added to teams I am a member of

I found a thread from back in 2011 mentioning 
this https://groups.google.com/d/msg/pinboard-dev/1Gdx-uhs6mI/dga_ijNoVWEJ 
- any news since then?

Thanks,

Simon Willison

Re: Draft spec for API v2 - please comment

kijin — 2013-04-30T04:08:25

API v1 has a call, /user/api_token, that returns a user's API token. Since 
API v1 uses the username and password for authentication, I suppose you 
could use it to obtain a user's API token, and then make API v2 requests 
with the token. Of course, this assumes that API v1 will remain operational 
after API v2 is released.

Kijin


On Tuesday, April 30, 2013 2:45:39 AM UTC+9, Stephen Darlington wrote:

Re: Draft spec for API v2 - please comment

Stephen Darlington — 2013-04-29T17:45:39

Will there be a way in the API to "convert" a username and password to an API token? 

I ask because no matter what you put on-screen, users will enter their username and password. The only other API I've used with the same feature is bit.ly's, and I've lost count of the number of support emails telling me that my app is broken. (It isn't, at least not in this respect.)

Can't believe I'm even suggesting this, but have you considered OAuth or at least something like it? You could have a "flow" that apps follow -- open a web page, log in, get the API key -- that would make it much less error-prone.

Cheers,
Stephen

On 25 Apr 2013, at 20:52, maciej <mceglowski< at >gmail.com> wrote:


------------------------------------------------------------------------
                    Stephen Darlington (www.zx81.org.uk)
"I blame sex and paper for most of our current problems"
------------------------------------------------------------------------

Re: Draft spec for API v2 - please comment

Dan Loewenherz — 2013-04-29T16:42:46

I'm personally a fan of the way the Stripe API utilizes HTTP status codes,
not that I really consider myself a stickler for these things.

https://stripe.com/docs/api#errors

200 OK - Everything worked as expected.

400 Bad Request - Often missing a required parameter.

401 Unauthorized - No valid API key provided.

402 Request Failed - Parameters were valid but request failed.

404 Not Found - The requested item doesn't exist.

500, 502, 503, 504 Server errors - something went wrong on Stripe's end.




On Mon, Apr 29, 2013 at 9:35 AM, Les Orchard <me< at >lmorchard.com> wrote:

Re: Draft spec for API v2 - please comment

Les Orchard — 2013-04-29T16:35:56

 > one covers general API features; I'll post the actual proposed method
 > calls separately.
 >
 > Please let me know your thoughts, good or bad, and any suggestions:
 >
 > http://static.pinboard.in/api2_draft.pdf

One more thought: I'd love to see CORS enabled for this API, so I can 
use it from HTML5 web apps. (eg. in a browser, on a FirefoxOS phone, etc)

Sending a `Access-Control-Allow-Origin: *` header will cover GETs, but 
POST/PUT/DELETE requires a slightly more complicated dance:

https://developer.mozilla.org/en-US/docs/HTTP/Access_control_CORS#Preflighted_requests

Re: Re: Draft spec for API v2 - please comment

Les Orchard — 2013-04-29T16:32:21

 > when there's not an obvious mapping.  In the example you give (422
 > Unprocessable Entity) there would still need to be a meaningful
 > error message in the body of the response to give the client any clue
 > about what specifically had gone wrong.

HTTP error code scheme lends itself to shoehorning - at least, with 
respect to statuses 200, 400, and 500.

200 OK is the catch-all for success; 400 Bad Request is the catch-all 
for something the client did wrong; and 500 is the catch-all for a 
server failure. If you can't nail the status down to a more specific 
code within those ranges, it's up to the client to consult the response 
body (as you mention about JSON error codes).

But, at least 2xx/3xx/4xx/5xx gives an initial branching point for 
response handling, both for the end client and intermediaries. For 
example, the status influences caching + conditional GET - 200 OK for 
everything means caches (in-app or on the network) happily hang on to errors

Re: Draft spec for API v2 - please comment

Les Orchard — 2013-04-29T16:31:33

 >
 > http://static.pinboard.in/api2_draft.pdf

Why not use HTTP Basic auth for the user + token authentication?

You could even combine the app ID into basic auth & parse it out on the 
server side - eg. base64(appID+user:token)

That might make things easier on the client, eg. no custom X-* headers. 
I know it can do funky things in browsers - ie. entice a user to enter 
user/pass in a phishing attempt (which happened on del.icio.us) But that 
might be less a concern since the user + token are not the same as user 
+ password