Erik, if you look at an updated version of the gist you posted earlier, I 
think that UpdateCache could readily be replace with LogoutUser (or 
HandleResponseError)

https://gist.github.com/pdamoc/d492ab58023926cd4d4950f12e5e170d

On Wednesday, July 6, 2016 at 10:25:05 AM UTC-10, Erik Lott wrote:
>
> Yeah, the cookie is acting like a global var, which makes me cringe as 
> much as you...
>
>
> On Wednesday, July 6, 2016 at 2:50:40 PM UTC-4, Josh Adams wrote:
>>
>> What we need is a simple way to communicate these events from the API 
>>> module to the root of the application. What if we use an external resource 
>>> like a cookie or local storage to communicate these events? For example, 
>>> when the module detects a successful login, it could also set a cookie with 
>>> value  isLoggedIn = true. If the Api module detects a 401 response, the 
>>> cookie is changed isLoggedIn = false
>>>
>>> Meanwhile, the application root also has access to this isLoggedIn 
>>> cookie value, and it could precede every "update" with an authentication 
>>> check - adding or removing any session data that is stored on the model.
>>>
>>
>> I think this is a good indication of my handwavy concern.  I don't think 
>> this is an event as much as it is data - "I'm logged in as this user".  And 
>> passing data inside your elm application by way of cookies is horribly 
>> reminiscent of using the dom as a database - i.e. that's not what it's for, 
>> if you're using it that way then it indicates a design problem.  The 
>> DOM-as-a-database problem haunted us for 15+ years, and still does for the 
>> unenlightened.
>>
>> Why is it that you need to reach for cookies here?  Cookies are for 
>> remote state.  This is local state. 
>>
>

-- 
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.

Reply via email to