How to detect closed onboarding window? about paypal-sdk-client

[issue-label-bot]

Issue-Label Bot is automatically applying the label question to this issue, with a confidence of 0.61. Please mark this comment with 👍 or 👎 to give our bot feedback!

Links: app homepage, dashboard and code for this bot.

[nixsin]

@JasonTheAdams The platform can use GET /v1/customer/partners/{partner_id}/merchant-integrations/{merchant_id} API documented here to get the onboarding status at any time.

https://developer.paypal.com/docs/api/partner-referrals/v1/#merchant-integration_status

[JasonTheAdams]

Hi @nixsin!

I'm afraid that doesn't actually address the issue. Please imagine the following scenario:

1. User clicks on "Connect to PayPal" button
2. PayPal window displays and the user starts going through process
3. At some point the user decides not to complete the onboarding and closes the PayPal window

At this point, our application is in a weird state. Once the user enters the onboarding flow the application sits in limbo of two possible scenarios:

• The user is in the onboarding flow
• The user has canceled and left the flow

We know if they've completed the flow, but we have no way of discerning (that I'm aware of) whether the user is still filling things out or if they've canceled the flow. So using the endpoint you mentioned isn't really helpful, as we can't act on the status — after all, if it's incomplete that may simply because they haven't finished. But having not finished isn't the same as having canceled.

Does this make sense?

[JasonTheAdams]

Hoping to hear back on this in some capacity. Are you able to provide some insight, @gregjopa?

Thanks! 😄

[mnicpt]

@JasonTheAdams Let me know if this will work for you: https://codepen.io/mnicpt/pen/zYBBXzX?editors=1011

[JasonTheAdams]

@mnicpt Thanks for the workaround. I admit it feels pretty hacky, as it's observing the document to see whether a very specific class is present and changes. If the SDK arbitrarily changed that class name (and why would someone assume it's being strongly referenced by implementors?) then this would stop working. That said, it does seem to do the trick.

Is there a more first-class method somewhere in the pipeline, per chance? I'm fine with this kind of solution for some known length of time, but if it's the indefinite solution then that makes me nervous.

I hope that makes sense!

[mnicpt]

@JasonTheAdams Yeah. Unfortunately, there is not a first-class method anywhere.