On Meta's development platform

Pain.

As part of a current project of mine, I had to use Meta's APIs in order to interface with some of their apps. I had never used them before, but since this project involved the processing of direct messages from Meta, I was left with no choice but to learn how to use them.

The APIs themselves seem great on paper. Processing direct messages from Meta's platforms requires the use of their webhooks, which send requests to endpoints implemented by the programmer. This approach is good, as it helps in both avoiding the frequent polling of Meta's endpoints, and in the implementation of serverless architectures. As I used it, though, I couldn't help but get a little frustrated.

Meta gracefully offers the developer an option to test-call whatever endpoint was entered in their platform. Their tests, actually, work great: it takes little time to receive and log whatever message they send into your endpoints, which helps you discard the possibility of a failure in communications between Meta and your program.

These tests passed. Every single test I ran using these buttons passed. So I tried to send an actual message to the Instagram account I linked to Meta's development platform. I sent dozens of messages, in fact, but aside from the periodic chronological logs that my program printed for me, nothing ever showed up. To those who are familiar with this platform, you may read this and blame it on user error. I assumed I had made a mistake, too, so I turned to the Internet for guidance.

Lo and behold, you have to turn on webhooks for the individual platform accounts you may want to test. Honest mistake. So I clicked on the button that turns these webhooks on, to a shocking surprise: "An error has occurred".

There is one thing that I believe all programmers share a common hatred for. Vagueness in logs. Some people take advantage of this hate with a good reason, maybe because they don't wish to guide potential attackers on exploiting their systems, for instance. I'm no attacker, though. I resorted to GenAI to perhaps fix this problem, which it did by turning on webhooks using Meta's API (which is a thing, apparently).

AI voodoo aside, that didn't fix the problem. I was getting tired, also because trying to test this program using anything other than Meta's test button proved quite difficult, as well. I attempted to log into my testing account on desktop, which resulted in an infinite loop of entering credentials, and solving captchas, to absolutely no error message. It'd send you back to the login page with no explanation. If I hate something more than vague error messages, it's perhaps, not getting error messages at all, but I digress.

Late at night, you could call my feud with Meta another improvised hackathon. I tried to enter the development platform once again, so I could take a picture for this post, but I failed, since it found "suspicious activity" in my account. I can't stand this, and yet, I go back to it, because there's no other choice in this project.

That's life, isn't it?