Jira issue originally created by user heiglandreas:

When calling the API with an unknown controller the complete error-report is output to the browser. That should not be the case

Reproduce: Point your browser to 'http://api.joind.in/v2.1/foo'
Expected Result: 404 message
Result:

Warning: include(/var/www/api.joind.in/51/src/inc/../controllers/FooController.php): failed to open stream: No such file or directory in /var/www/api.joind.in/51/src/inc/Autoloader.php on line 33

Warning: include(): Failed opening '/var/www/api.joind.in/51/src/inc/../controllers/FooController.php' for inclusion (include_path='.:/usr/local/php/php5.5.12/lib/php') in /var/www/api.joind.in/51/src/inc/Autoloader.php on line 33

Warning: Cannot modify header information - headers already sent by (output started at /var/www/api.joind.in/51/src/inc/Autoloader.php:33) in /var/www/api.joind.in/51/src/public/index.php on line 10

Warning: Cannot modify header information - headers already sent by (output started at /var/www/api.joind.in/51/src/inc/Autoloader.php:33) in /var/www/api.joind.in/51/src/views/HtmlView.php on line 25
0: Unknown controller foo

Jira issue originally created by user akrabat:

For the Akismet check, we need to supply IP address and user agent. Code is here.

We should update this code to check for the X-Forwarded-For and Forwarded headers.

The Forwarded is defined in RFC 7239 and allows for passing originating IP address and host information through a proxy and also allows for extension tokens for other information.

We need to check Forwarded for the 'for' and 'user-agent' tokens, so that a header like this:

Forwarded: for=192.168.0.1;user-agent="Mozilla/5.0 (Windows NT 6.3; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/37.0.2049.0 Safari/537.36"

would provide an IP address and user agent for sending on to Akismet.

Jira issue originally created by user lornajane:

When creating an event, location should be required

Jira issue originally created by user richsage:

Example Frisby code:

frisby.create('Talks at ' <ins> evt.name)
    .get(evt.talks_uri </ins> '?resultsperpage=3')
    .expectStatus(200)
    .expectHeader("content-type", "application/json; charset=utf8")
    .afterJSON(function(evTalks) {
        if(typeof evTalks.talks == 'object') {
            for(var i in evTalks.talks) {
                var talk = evTalks.talks[i];
                checkTalk(talk);
            }
         }
    })

If in the above example on line 5, evTalks returns false due to non-valid JSON being returned, the tests don't fail...and perhaps they should?

Jira issue originally created by user mvriel:

When posting an event the Location header should return the URI of the newly created event but now it returns the URI of the event collection.

The documentation at http://joindin.github.io/joindin-api/events.html also shows this in the example response and my own tests confirm it:

HTTP/1.1 201 Created
Date: Wed, 18 Jun 2014 13:26:11 GMT
Server: Apache/2.2.15 (CentOS)
X-Powered-By: PHP/5.3.3
Location: http://api.dev.joind.in:8080/v2.1/events
Content-Length: 0
Connection: close
Content-Type: text/html; charset=UTF-8

Jira issue originally created by user lornajane:

From phpcodemonkey:
I'd love an RSS feed of the current open CfP submissions? RSS? Yes, I know it's not all REST/Hispter but it's still very useful! :)

Actually there are a few API endpoints that would really benefit from this. In particular the comment collections I think would be good. The API uses output handlers for all requests so one approach might be to add an RSS output handler so the whole API is available this way.

Jira issue originally created by user lornajane:

This issue started life as being about the fact that we don't notify when talk claims are made.

Please read the comment thread for a more detailed description of how to make joind.in much, much better :)

Jira issue originally created by user richsage:

If a comment is posted via the API, it would be nice to show the details of the token used when requesting the comment's "source" property. For example, "joind.in Android app" or similar, rather than just "api-v2".

For reference, this field is available via the comment's verbose URI.

Jira issue originally created by user kevin:

This ones a bit complex; if a user is already logged-in to the web2 site, but then tries to login again (which I was only able to do because I'd introduced a bug into my web2 site) then the API call to /token contains an "Authorization" header, which causes the API code to call "getIdentity" before it follows its normal code path, which causes the "user_uri" field in the API response to be (erroneously) missing the API version from the URL.

The problem is in Request::getOauthModel (line 315 of src/inc/Request.php). The first time it's called, it memoizes $this->oauthModel. The second time it is called, it returns an OAuthModel which does not have the "version" field set, so the URL is not created properly.

I think it's something to do with $this (line 321) not surviving across the memoization properly.

Jira issue originally created by user tocacar:

In the android app, when you click to view an event, buttons are displayed to provide access to talks, comments and tracks.

At the moment, the number of comments is displayed on the button, courtesy of the 'eventcommentscount' value returned by the API.

It would be nice to show the number of talks and tracks on their respective buttons also. Would it be ok to add these values to the EventMapper::getDefaultFields() array? The $sql variable in EventMapper::getEvents() would also have to be edited to include the count of each related collection.

Jira issue originally created by user gargoyle:

patch39.sql added a duration field without a default value (which makes sense). However, seed.sql now fails to run properly.

Jira issue originally created by user clair:

When getting details on the event (for example, on the frontpage) the number of comments is for the event rather than talks - which at first might be a little misleading (especially if the event only has one talk)

Would it be possible to maybe have two figures? Although, I could see how this might be a bit heavy load for the server

Jira issue originally created by user lornajane:

Patch 39 added a duration field without a default. The sample data generation will need updating to add duration when generating sample talks.

Jira issue originally created by user rickogden:

When you list retrieve events from the API, there are many events which don't have a stub. Since the numerical ID of the event is not returned in the API, the stub is the easiest event identifier to use.

Jira issue originally created by user lornajane:

Currently we don't have any forgotten password functionality on the web2 site. We need the frontend on the web2 site but first the API needs to support this. Allow POSTing to an endpoint (maybe /user/resets) to request a password reset be emailed - this needs to point to a URL on web2 that can support checking the unique link that the API sent and store the new password when supplied.

Feel free to pick up just the API part to begin with.

Jira issue originally created by user s_e:

After the Guzzle talk at PHPNW13 and a quick chat with [_{akrabat], [}lornajane] and [~clair] I thought this would be an ideal modular way of calling the API for the responsive site, with a long-term view to having a simple Joind.in API client that could be used in any php-based site wishing to make use of Joind.in data.

I'm happy to be the one to (start) implement(ing) this.

Let me know if you have any comments or suggestions, like or dislike the idea?

Jira issue originally created by user lornajane:

When creating a comment, the response is a 201 with a Location header, but the header points to the collection. It should point to the new resource.

Jira issue originally created by user tocacar:

As I was examining the code, I wondered why the Mapper classes' getVerboseFields method doesn't call getDefaultFields and then merge that array with the additional fields.

I realise that, the way things are now, some of the additional fields added in getVerboseFields are mixed in amongst (thereby affecting the order of) the values in the default fields array. I'm not sure whether this is essential or not though.

Merging the arrays in getVerboseFields would remove the need to maintain duplicate values in each method. It may also be nicer for consumers to get the fields in the same order whether requesting default or verbose (just with verbose's additional fields added at the end). I'm not sure whether or not this does make any practical difference to consumers though.

This is just an observation.

Jira issue originally created by user BarryCarlyon:

As per notes in JOINDIN-361.

When POST-ing to http://api.joind.in/v2.1/events/[eid]/attending/

The API doesn't check to see if the user/event reference already exists. It just adds a reference to the user_attend and 201 CREATED.

I'm not sure if the best solution is to chuck a unique index on the table for the two columns, or to run a query and check during the postAction to the eventMapper model, or implement both.

Jira issue originally created by user lornajane:

If something in the build process fails, the entire build needs to be marked as failed so that we know something went wrong. For example for an exec tag, add the checkReturn="true" parameter. Probably also applies to lint checks.

Jira issue originally created by user tocacar:

This is just another observation as I look at the code.

If the $content variable passed to ApiView::addCount has a next_page link in it, I think you may be able to skip the counting loop and instead return the value of 'resultsperpage' from the pagination link url.

This might be overly complex for not much gain, but I thought I'd mention it as the thought did occur to me.

Jira issue originally created by user clair:

At the moment, if there's exactly 10/20/30 etc results returned, then the "next" button will appear, with the next page showing no results (as there isn't an 11th result)

This non-urgently needs to be fixed, potentially by getting 11 results instead (knowing if there are more) and then only showing the first 10

Jira issue originally created by user erichogue:

When creating the VM, patch-db seems to work fine. But the second time it runs, it fails and leave the db empty.

Here's the output I get when booting the VM a second time:
err: /Stage[main]/Joindin::App/Exec[patch-db]/returns: change from notrun to 0 failed: /vagrant/joindin-api/scripts/patchdb.sh -t /vagrant/joindin-api -d joindin -u joindin -p password -i returned 1 instead of one of [0] at /tmp/vagrant-puppet/modules-0/joindin/manifests/app.pp:8

notice: /Stage[main]/Joindin::App/Exec[seed-data]: Dependency Exec[patch-db] has failures: true

warning: /Stage[main]/Joindin::App/Exec[seed-data]: Skipping because of failed dependencies

Jira issue originally created by user dstockto:

Create re-runnable patch to insert missing languages into db

Jira issue originally created by user richsage:

It would be nice to be able to request events (upcoming, hot, past) that you are marked as attending. This doesn't seem possible presently, and particularly for "Past" events, it would be good to not have to retrieve all events to find eg the 2 you attended.

Jira issue originally created by user dstockto:

Add unit tests for API v2

Jira issue originally created by user clair:

It looks like web2 needs data from the API with the 'urifriendlyname' field, as when connecting to a test API from vagrant I get the following Slim error:

Undefined property: stdClass::$urlfriendlyname

Jira issue originally created by user erichogue:

The new VM for the responsive site does not see the the VM used for the original site and the API. They need to be on the same network so the responsive site can use the API on the dev environment, not the live one.

Jira issue originally created by user dshafik:

It would be nice if conference organizers had the option to require registration for all comments on talks at their conference.

You could further expand this to allow users to require registration for each talk, or all of their talks, but I think it would be best if this were consistent per conference.

If nothing else, it would be interesting to see how many enable this feature, and how it affects comments.

Jira issue originally created by user dstockto:

Create a phing task that makes running codesniffer easy for contributors.

Jira issue originally created by user auroraeosrose:

Often talks have items in addition to slides - maybe a zip file of code examples or a podcast or a video or something - the ability to add additional links to the talk with this information would be great.

Stories…

1. As a Speaker I want to be able to submit a collection of talk links to the API so that they are added to my talk
2. As a System Admin I would like the 'slides*link' posted to the talks endpoint in the 'links' collection to be also stored to the talks table slides*links field in order to support backwards compatibility with Web1
3. As a System Admin I would like the GET /talks endpoint to check that the talks table slide*links match the link in the talk_links table entry for the talk's slides, and update the talk*links table if not, in order that slide links updated via Web1 are replicated to the API responses
   …
4. What should happen if the API talks endpoint receives a request with both slides_link and an entry in the links collection of type slides?

Jira issue originally created by user lornajane:

The gravatar images rely on having a hash of the user's email address, however we don't publish user email addresses via the API. Please add a field for both talk comments and event comments, in verbose mode only, which returns md5(strtolower($email)).

This same hash should be added to the verbose representation of the user resource itself.

Jira issue originally created by user akrabat:

curl -X GET "https://api.joind.in/v2.1/users/1?verbose=yes"  -H "Content-Type: application/json" -H "Accept: application/json" -H "Authorization: Bearer 6be487d170bdc53c" -m 30 -v

Results in:

* Adding handle: conn: 0x7fd18a004000
* Adding handle: send: 0
* Adding handle: recv: 0
* Curl_addHandleToPipeline: length: 1
* - Conn 0 (0x7fd18a004000) send*pipe: 1, recv*pipe: 0
* About to connect() to api.joind.in port 443 (#0)
*   Trying 178.208.42.30...
* Connected to api.joind.in (178.208.42.30) port 443 (#0)
* TLS 1.0 connection using TLS*RSA_WITH_AES_128_CBC*SHA
* Server certificate: api.joind.in
* Server certificate: RapidSSL CA
* Server certificate: GeoTrust Global CA
> GET /v2.1/users/1?verbose=yes HTTP/1.1
> User-Agent: curl/7.30.0
> Host: api.joind.in
> Content-Type: application/json
> Accept: application/json
> Authorization: Bearer 6be487d170bdc53c
> 
< HTTP/1.1 400 Bad Request
< Date: Wed, 25 Jun 2014 12:28:36 GMT
* Server Apache is not blacklisted
< Server: Apache
< X-Powered-By: PHP/5.5.12
< Status: 400
< Vary: Accept-Encoding
< Content-Length: 159
< Connection: close
< Content-Type: text/html
< 
<br />
<b>Fatal error</b>:  Call to a member function render() on a non-object in <b>/var/www/api.joind.in/51/src/public/index.php</b> on line <b>11</b><br />
* Closing connection 0

Jira issue originally created by user clair:

I'd really like to create a kind of "calendar view" for events for the new site (I think this would be easier to use than just a list of events in date order)

Would it be possible (is it possible already?) to perform an API search by date (my use case is to be able to search by month, so for my wishes that's all I need!)

Jira issue originally created by user richsage:

When API data is json*encode'd for subsequent output, we force conversion of strings containing numbers to actual integers, via the JSON_NUMERIC*CHECK flag. This can cause problems if the string is not meant to be parsed as such, resulting in the following error:

Warning: json_encode(): double INF does not conform to the JSON spec, encoded as 0 in php shell code on line 1

I noticed this when a test event had a stub of "4e996", which resulted in the field value being set to 0 and the above error being thrown.

Jira issue originally created by user mvriel:

When I try to submit an event using the /v2.1/events endpoint and provide the cfpstart_date and cfp_enddate I receive the following output from the API:

Catchable fatal error: Object of class DateTime could not be converted to string in /vagrant/joindin-api/src/models/EventMapper.php on line 656 Call Stack: 0.0145 683904 1. {main}() /vagrant/joindin-api/src/public/index.php:0 0.2205 1095472 2. routeV2() /vagrant/joindin-api/src/public/index.php:73 0.2449 1304360 3. EventsController->handle() /vagrant/joindin-api/src/public/index.php:116 0.2452 1304360 4. EventsController->postAction() /vagrant/joindin-api/src/controllers/EventsController.php:9 0.3210 1635672 5. EventMapper->createEvent() /vagrant/joindin-api/src/controllers/EventsController.php:287 0.3210 1637568 6. PDOStatement->execute() /vagrant/joindin-api/src/models/EventMapper.php:656

The values used for both start and end date was 2014-12-31

Jira issue originally created by user lornajane:

The list of talks at an event doesn't seem to be sorted by anything in particular. Should show the sessions in order of start time, so they make sense (and a good agenda view)

Jira issue originally created by user clair:

There's a slug field that the API returns, but for the web2 site we'd like a way to search by this so we can implement slugs in the URL

Jira issue originally created by user clair:

It would be handy if the created date for an event was returned by the API, so that the responsive site can append -1, -2, and so forth to later events that may share the same name

Jira issue originally created by user lornajane:

When talks are imported, they are written into the database as each line is read from the file. If there are any errors, it dies - and so can leave you with half-eaten talks added to your event. Put things into a temporary table and then ask the user to confirm all looks OK before they actually get stored and shown against the event.

Jira issue originally created by user clair:

When visiting http://api.joind.in/, the links don't work correctly. They appear to be pointing to http:/// rather than http://

Jira issue originally created by user lornajane:

Skoop tells me that the pfcongres event doesn't show in the CfP section on the website (event: https://joind.in/event/view/1240). I checked, and it does not show. Since it has the CfP box ticked, the dates are current, and it has a CfP URL, it should show up.

I checked the API and it does not appear there either (although there's another event in the API that isn't on the website! PHP Brazil: http://joind.in/event/view/1354).

Jira issue originally created by user choult:

Warnings are thrown when visiting the following URL:

http://api.joind.in/v2.1/token

( ! ) Warning: Invalid argument supplied for foreach() in /var/www/api.joind.in/68/src/views/HtmlView.php on line 44
Call Stack

Time Memory Function Location

1 0.0001 225352 {main}( ) ../index.php:0
2 0.0015 301504 HtmlView->render( ) ../index.php:105
3 0.0015 301768 HtmlView->printArray( ) ../HtmlView.php:27

Jira issue originally created by user eli-t:

Whether this should be deprecated, removed or set to 0 is up for debate.

Jira issue originally created by user lornajane:

When comments are made via the website, we email speakers to tell them there's a new comment on their talk. The API needs to do the same, ideally with an identical email.

(beware that there's also an open pull request for changing the email joindin/joindin-legacy#675 so check this has been merged first)

Jira issue originally created by user michellesanver:

There is currently only one event organiser by default, it would be nice if one could add multiple organisers from the beginning rather than having to edit the event to do that.

Jira issue originally created by user clair:

For the responsive site, it would be nice to search for both events and talks by title, so that someone can search for a given talk or event in a searchbox

This issue is arisen by JOINDIN-345

Jira issue originally created by user richsage:

At the moment the comments URIs return all public comments and no private ones, irrespective of whether a user is authenticated or not. Therefore no private comments can currently be retrieved via the API

This is more of a placeholder issue to discuss and decide an approach, based on authenticated users and their various types (admin, speaker and similar).

Jira issue originally created by user tocacar:

The EventMapper class's getEventById method contains an $order variable which is not used (and should never be required by this method anyway, as it is not necessary to sort a single result).

Furthermore, the call to $this->getEvents() inside getEventById passes 'null' as the 4th argument (this is where the $order variable would have been passed, if needed). It is not necessary to include null here as the getEvent method has null as a default value for its 4th parameter.

Jira issue originally created by user eli-t:

As a consumer of the API that caches resources locally (talks/comments so far but this may expand), performing a refresh is very expensive; everything must be retrieved and compared.

It would be really helpful to have some sort of sequence number or similar that can be incremented to see when resources have been updated.