Lettuce features and scenariis may be a little bit of the track. I lack a good example of Web API testing.
However I really like the idea !

I wonder if we should actually be more readable in the scenarii file, for e.g :

When I send a bad model
When I send a incomplete model

And something like:

@step("When I send a (bad|incomplete) model")
def i_send_a_model(aspect):
    if aspect == 'bad':
         model = "{uierui:gjigjr}"
    elif aspect == "incomplete":
         model = '{"title" : "yeah"}'
    ...
    ...

I asked @stephane for a quick review :) Let's see!

http://remotestorage.io/get/ ?

After discussion with @AntoineCezar and debates in issue #9 it seems relevant to use JSON schema for model definitions, instead of our own formalism.

A basic investigation needs to address the following questions:

• Should it replace or supplement the current definition formalism ?
• Does it support current and planned field types ? #51 #33 #29
• Is it extensible and simple enough to add custom notions ? #11
• Could it address the need of defining and validating relationships between models ? #32
• Is it redundant with SPORE ?

More generally, should JSON schema validation come as a Cornice extension or as internal Daybed mechanics ?

currently we are returning a list of fields, we could also return more information about them.

A way to register available fields types.

It should :

• populate the list of supported values for the type field validator
• add extra supported items in field definitions (e.g. choices, dateformat)
• provide a validator for related fields of posted data (e.g. month<=12, abs(lat)<=90)

See :

http://daybed.lolnet.org/data/mushroomspots
http://daybed.lolnet.org/data/fefe

List of integer, string, etc. or list of "sub-model" for inline records

The actual token system bring behaviors that does'nt look good to me.
When I POST data I excpect to get the full object with additionnal information like for exemple a "created_at".
In a POST based create definition I recive a token and have to follow a redirect to get the full object.
So unless there is a better way to pass tokens I suggest droping them for the moment.

At model creation, it could be possible to define groups of fields:

SCHEMA = {
    name: "My super event",
    description: "",
    fields: [
        {   type: "string",
            name: "title",
            description: "blah",
        },
        {   name: "category",
            type: "enum",
            description: "blah",
            choices: ['sport', 'culture']
        },
        {   name: "street",
            type: "string",
        },
        {   name: "city",
            type: "string",
        },
    ],
    groups: [
        {  title: "Home",
           description:  "Your address helps us find you",
           fields: ["street", "city"]
        },
    ]
}

This would allow the UI to nicely layout the form for this model.

Since groups would be optional, they do not interfere with existing models, or those who don't need UI forms.

Will be useful for remote applications after merging what lays in dev branch

http://www.elasticsearch.org does exactly what we need in terms of faceting (i.e. filters, drill down, ...)

Plus, it has a plugin to be fed from CouchDB : http://www.elasticsearch.org/guide/reference/river/couchdb.html

However I wonder how easy it could be deployed (ex: Heroku, Travis,...)

Accept HTTP directive may not be enough to format records.

For example, getting the records of a model which contain a geometry as JSON may have two formattings :

• a basic JSON (as currently : [ {...}, {...} ])
• a GeoJSON ( { "type": "FeatureCollection", "features": [{...}, {...}]), that integrates nicely in JS map widgets like Leaflet.

Both are valid JSON. We might need a ?format=XXX parameter to specify which layout it should have.

For example (does not have to be all of them) :

• Interval : value with min & max
• Email
• URL
• Phone
• Date
  ...

I found myself trying to find a name for my definition even if I didn't needed one. That could be useful to have daybed deciding directly this kind of things for us. I like heroku which is choosing some names that "could" be real worlds, rather than using hashed of some sort.

https://github.com/jkbr/httpie

that would probably be prettier than curl

http://jsoneditoronline.org/

• Draft a roadmap
• Explain what it does not and will never do
• Explicit missing parts supposed to become external projects (UI form builder, form & view, maps, mobile apps)

The documentation is outdated.
We have to document everything.

POST ``{"mushroom":"frfr","area":"111"}`

{"status": "error", "errors": [{"location": "body", "name": "", "description": "\"111\" is not iterable"}]}

It says "make setup_venv" instead of "make install"

Vega is a declarative grammar for data visualizations

https://github.com/trifacta/vega/wiki/Tutorial

(previous title was "remove the notion of token")

Suggested by @AntoineCezar

Daybed should be as standard as possible in terms of REST and CRUD.

If any ACL feature (see #8) has to be implemented, it may be relevant to implement it in a separated frontend rather than in daybed core.

Il faut valider avec https://github.com/SPORE/specifications/blob/master/spore_validation.rx en utilisant https://github.com/rjbs/rx

when resources don't exist, we should return the right CORS headers. I'm not sure we're doing this ATM.

Idea by @AntoineCezar

Browser ID is a really good idea to use daybed as as javascript UI backend.

Basically we can then use the email as the account id for models and data management.

Allow users to define their own validation function in the posted JSON model definition :

fields: [
        {
            type: "int",
            name: "age",
            description: "Your mother age",
            validation: "if (value < 15) { 
                                return 'Seems a bit young.';
                             }
                             else (value > 120) {
                                return 'Can't be so old.';
                             }
                             return null;",
        },

Most appropriate language seems to be Javascript (since JSON).

Javascript syntax must be checked during model definition.

Validation function should be executed on model post. In my opinion, server side (thus in python) :

http://renesd.blogspot.fr/search/label/shitjs

https://bitbucket.org/pypy/lang-js/src/de89ec32a7dc/js/javascript-interpreter.txt

http://code.google.com/p/pyv8/ : extra dep :(

We could envision a way to refer a model as a relationship :

Definition could look like:

{
"title": "Person",
"description": "An individual with a soul",
"fields": [
    {
        "name": "fullname",
        "type": "string",
    }, {
        "name": "brothers",
        "type": "many-to-many",
        "model": "Person"
    }

And an entry like :

{
    "name":  "Howard",
    "brothers": [54321, 9876]
}

Validating this entry would mean checking that provided brother ids exist.

With this approach ?

• Give UUID at record insertion (POST /model)
• Hide records ids from public (GET /model)
• Update a record with PUT /model/<UUID>

Means users should keep track of their uuids themselves, not fantastic.

Other idea (brainstorm)...

• Provide password at record insertion (POST /model?lock=1234). If no password, record is public.
• Show records ids in GET requests
• Update records providing password

Actually, same approach could be applied to model definitions...

Will imply running daybed behind https...

POST invalid point data gives 500 error :

  File "/home/mathieu/Code/daybed/src/daybed/validators.py", line 35, in schema_validator
    return validator(request, schema)
  File "/home/mathieu/Code/daybed/src/daybed/validators.py", line 20, in validator
    for field, error in error.asdict().items():
  File "/home/mathieu/Code/daybed/src/local/lib/python2.7/site-packages/colander/__init__.py", line 165, in asdict
    keyname = exc._keyname()
  File "/home/mathieu/Code/daybed/src/local/lib/python2.7/site-packages/colander/__init__.py", line 153, in _keyname
    return str(self.node.name)
AttributeError: 'PointType' object has no attribute 'name'

The format description we're using to define our validation models could be useful for others, so let's split this up! (maybe in the documentation ?)

When running tests, the "daybed" database should already be created otherwise an error will be thrown. This should be done automatically to avoid additional user hassle when users are getting started.

Does not seem to be the case:
https://github.com/spiral-project/daybed/blob/master/daybed/views/data.py#L45-L49

http://en.wikipedia.org/wiki/XForms#ODKCollect

Looks like Android applications can be automatically built upon this spec...

Be careful though, could be bloated (Java inside).

I'm on ArchLinux and the default python is python3.
Virtualenv should explicitly use python2 with the --python=python2 option.

see Cornice error Cornices/cornice#55

That could be useful to return a random element from a list of resources.

PATCH allows us to only change some values inside an item, which is rather cool.

I want that :-) Less traffic, less headaches.

• find a way to specify expected language

That can store and validate any type of supported geometry value

• Point
• Line
• Polygon
• Posted values will be coordinates list, like those stored in GeoJSON format
• Will be rendered as GeoJSON individually
• Collection of records will be rendered as GeoJSON only once issue #12 is implemented

Request method: POST
Request URL: http://localhost:8000/data/mushroomspot
Request payload: {"mushroom":"tete","area":"[20.390625,7.536764322084078]"}

ERROR [waitress][Dummy-1] Exception when serving /data/mushroomspot
Traceback (most recent call last):
  File "env/local/lib/python2.7/site-packages/waitress/channel.py", line 329, in service
    task.service()
  File "env/local/lib/python2.7/site-packages/waitress/task.py", line 173, in service
    self.execute()
  File "env/local/lib/python2.7/site-packages/waitress/task.py", line 380, in execute
    app_iter = self.channel.server.application(env, start_response)
  File "env/local/lib/python2.7/site-packages/pyramid-1.4-py2.7.egg/pyramid/router.py", line 252, in __call__
    return response(request.environ, start_response)
  File "env/local/lib/python2.7/site-packages/WebOb-1.2.3-py2.7.egg/webob/response.py", line 1021, in __call__
    start_response(self.status, headerlist)
  File "env/local/lib/python2.7/site-packages/waitress/task.py", line 364, in start_response
    'Header value %r is not a string in %r' % (v, (k, v))
AssertionError: Header value u'http://localhost:8000/data/59fcf69f72069e8bf8d917a155001179' is not a string in ('location', u'http://localhost:8000/data/59fcf69f72069e8bf8d917a155001179')