On Thursday, September 29, 2016 at 4:25:25 PM UTC+1, Rupert Smith wrote:
>
> Some scnearios that would need to be covered:
>>
>
I also have a little problem with the 'roles' encode to solve. At the
moment I have:
, ( "roles"
, case model.roles of
Just roles ->
roles |> List.map roleEncoder |> Encode.list
Nothing ->
Encode.null
)
which will mean 'null' is output when the roles are Nothing. Which is not
what I really want, as an explicit null will be interpreted by the server
as meaning 'set the roles to null', whereas what I really want is no roles
field set at all which the server will interpret as 'don't change the
roles, they are not being defined by this request'.
But Encode.object takes a list of (String, Value) -> Value and its easiest
for me to write this as an entry in a list rather than build up the list
more conditionally. I suppose I could output a list of Maybe Value and then
use the filter to remove Nothings from that list to arrive at a list of
Value just for fields that are not Nothing.
I suppose what I am trying to point out, is that auto generating encoders
and decoders requires dealing with a lot of different conventions depending
on how you want it done. If I was using something like jackson in Java, the
behaviour would be controlled by a set of flags when building the parsers,
like "allowNoQuotes", "allowMissingFields", and so on. I think any attempt
at auto generating would need to look at a good number of common variations
in how json is treated and provide options to deal with them, in order to
not be overly restrictive and useful.
--
You received this message because you are subscribed to the Google Groups "Elm
Discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
