Closed mallardduck closed 1 year ago
I think it's a good idea. Other approach would be to get a relation from WinkAuthor
to a User
automatically, but I think extending it would be better (not sure about using the same table though)
I feel like it might be a good idea to allow developers to change which model is considered the user model via the config file. By default, it could use the default Laravel User
model.
Any extra meta datta about the users could then be stored in a seperate wink_user_meta
table, with a 1-to-1 relationship to the configured user model.
@DivineOmega That's an interesting idea! I like the sounds of that.
@mallardduck Thanks. I think it would be a good way forward.
The only complication I can see is that the current wink_authors
table uses a UUID primary key rather than a numeric auto-increment.
Yeah i wanted to ask that to @themsaid also.
Why not using auto increment ids? They are supported in almost every database engine, itβs easy to read or check against and uses less size
I used UUID as it's easier to reference to something while it's not still persisted in the database. I'm not using it in anyway now but just wanted to leave it flexible for the future.
Maybe if we switch to auto-increment keys would make linking between Wink Author and an existing user easier.
I'd love to make this link easier, however for example if you want to create a Wink user from Wink, but your user model requires much more data than the ones you can provide via Wink, things that can't be left null
, what would you do?
What would be a good example of that kind of data non-nullable but you can't fill on the first create step?
I've always worked with auto-increment ID both in work and in personal projects without any trouble.
I would like to use auto-increment in all the Wink tables, but that's my personal opinion π .
I used UUID as it's easier to reference to something while it's not still persisted in the database.
In which case do you would need to reference something not persisted yet? I can only think about the auto-save feature, but that's compatible with auto-increment ID as well. For example You can check if the post has an id, if not, create it and return the id of the object created, and next auto-save iterations just update that post with the returned id.
I don't know if I'm explaining it well enough since my english is not as good as I'd wanted to be, let me know if I screwed up π
Is the problem we're trying to solve that Wink has a separate Authentication Middleware? So for example if you're logged in as App\User::1 and go to /wink you should be able to access without a second authentication?
If that's the only problem here I think we can make a bridge between the two auth guards. For example link between WinkAuthor and App\User via email, and check if current App\User has a WinkAuthor we let him in without authentication.
@themsaid My original intention was to just pose the question of how best Wink can integrate with Apps that already have defined user and authentication systems.
The idea being that Wink shouldn't make any assumptions about how users should be stored. If a site/app already has Users that own content pieces, it may be best to reuse those users instead of having them duplicated in wink_authors
.
Or maybe an App/Site already has a User system where different user roles extend the base User class. In their system it may be best to define their own WinkAuthor that extends their base User. In systems like this the authentication might only ever use the User model and then trades the instance for role specific ones as needed.
IMHO, Wink would be most versatile and re-usable if it allowed this type of customization out of the box. For the user, it could be as simple as swapping the user in Laravel's main auth config file.
I agree with using / extending Laravel's default auth system, or relating it to WinkAuthor which I still think it's necessary to store wink
fields. Also using id
instead of uuid
for primary keys
I agree, too.
Wink should not use a specific auth and leverage the use of the Laravel's default auth. Wink should only use the ID (or whatever property is used to identify the authenticated user) and use that to relate to posts and other wink specific tables.
Also, and after a little more reflection on my discussion with @Lloople on my PR (#22), about using a specific table to store the wink columns, I can also agree that it's a better way to not mess, directly, with the users' table, on any Laravel application. Although it adds a little bit of complexity it may payout by enabling more complex sites to use this package.
Any reason we couldn't just ship with a default "WinkUser" model which is defined in a config file, An additional "WinkGuard" for auth, which is again defined in a config file,
Make the WinkUser model pretty much empty and move it's relationships etc out to a trait.
We could also have two sets of migrations, one to "update the users table" and one to "create wink_users" table.
Get the update migrations to get the users table name via (new config('wink.user_model'))->getTable()
Which set of migrations we publish would depend on whether wink.user_model points to the default WinkUser model or not.
Users can then, if they wish to use the default Laravel setup simply edit a config file
return [
'user_model' => \App\User::class,
'auth_guard' => 'web'
];
And apply a trait to their user model
class User extends Model
{
use IsWinkUser;
}
@hailwood I'm considering this yes for 1.0
@hailwood That sounds like a very solid idea to me. It's literally exactly on par with what I was imagining too. Such an elegant (sounding) solution.
Any updates on this? I was wondering if this is being worked on in any branch as I'd like to contribute
@idragon81 I just need to free sometime to take a deep look into it :) It has to be done in a way that doesn't break the current method of authentication.
Brilliant - want to use Wink, but would be a bit much to ask someone to log in twice to the same application. Brilliant starting point you have though @themsaid
Used wink for the first time to check it out today, great first pass @themsaid!
+1 for the trait idea under User
IMO. We have an existing platform of a few thousand users with a simple enough User
model, it uses Laravel Spark under the hood, so we're employing the Spark::developer()
system, so possibly as a simple version, something like that to access /wink
?
This probably undoes a lot of your great work in implementing the migration-based system Mohamed, sorry!
Has any headway been made on this front?
I wonder if now that Jetstream is out we could use that as a "base" - more like a control - system to design and implement a contract/trait for the WinkAuthor
. This would provide a concrete/working example of how it's implemented.
And, in theory, could make it easier for other systems (backpack/voyager/etc) to integrate to Wink within a single project more cohesively. I think maybe even it would help with the Spark integrations some users seek too since AFAIK new versions of Spark will use Jetstream too?
Note: I don't have any experience with Wink yet, still planning on setting it up and playing with it later. That said, this is just based on looking over code here in the repo.
As the title suggests, in general I think it would be nice to not have Wink duplicate efforts for Users. In an application I've built with multiple user types/roles I had success with extending the base User class. I was thinking a similar approach may be beneficial here - especially for integrating Wink into Laravel applications with other User based features.
I know that this idea is no simple feat - as Wink would have to be aware (in some way) if it needs to provide a complete
WinkAuthor
or an extendedUser
one based on the existing app. All the while maintaining the same relationships to this user class thatWinkAuthor
currently has. I also know that this would be an easier task if Laravel had a more standardized 'stock' User.So what are the thoughts about this concept, or going this direction?