More discrepancies in data formats

When new data is added, could we please take a look to see if there is already an existing format that we could clone?

Right now, guild achievement data is exported, but:

(1)  it does not include a total count or a total score, like character's achievements do

(2)  It uses "completedtimestamp" rather than "completed_timestamp" as character's list does.

So, once again, new data has been added without checking to see if it should follow a format that already exists.  I'm thrilled the data is there, but it would be nice to not make the API into more of a mess than it already is by just throwing stuff against the wall and hoping it works.

 

Also, if the achievements were:

<achievement_494832432 id="494832432" ...>

instead of:

<achievement id="494832432"...>

wouldn't that provide us some additional searching and sorting ability?

I noticed on Planetside 2 that all the weapons are <n_948 id="948"...> instead of <weapon id="948">.

 

While data APIs should be defined to make it easy to parse them, I don't at all like using the tag name (in XML) to mean both "the type" and "the id".

I would rather see:

<achievements>

  <achievement id="12345">

    <count>5</count>

    <completed_timestamp>0</completed_timestamp>

  </achievement>

</achievements>

This is a very standard way to represent the data.  How it is searched for should be somewhat orthogonal to how it is returned.

In json:  achievements: { "12345": { count: 5, completed_timestamp: 0 } }

Nice and compact, and it uses the id as the key into a hash, so individual items can quickly be used.  Chances are this is the format you want in your presentation layer as well.

 

Unfortunately with that format, searching would be essentially impossible. No doubt that's a good XML structure for local data, but it would cripple a REST API like this one. All SOE fansites would have to resort to scraping the entire database and storing a local copy to do anything with the data. When you search for items and characters on EQ2U, you are actually realtime searching Census and then we parse the results.