Thanks for getting back to me with some specific examples, that’s really helpful. I’ll dig into them in a moment, but to your point about:
An email list, twitter feed or something that just says “yo guys, check your logs over the next few days” would be awesome.
Part of the problem with the “check your logs” approach is that where we suspect a knock-on effect to the API, we will work to ensure that effect doesn’t happen. If you see a change in API behaviour and it’s not been pre-announced here, it’s either a change to undocumented behaviour, or it’s a change we did not intend to make. In the latter case, we try to be as responsive as possible to reports of changes. In the former case (as with your initial query here), we will look to see if there’s a way we can make the undocumented behaviour robust enough to be restored and fully supported.
Regarding the “email list”, that would be this forum. You’re now a member, and hopefully you’ve set up email alerts so you have a fighting chance of seeing any future announcements we make. I realise we could do a better job of ensuring all developers are signed up to this forum, and we should consider ways of improving that in the future.
Finally, regarding your specific issues:
Please do let us know if you get an idea of when you saw the contacts API behaviour shift. Even if it’s a date when it definitely did work as you describe, it will help us pull code from that date and try to reproduce the behaviour to understand what has changed.
Rate limits: when introducing rate limits, we did reach out to any developers whose integrations were busy enough to have been immediately affected by their introduction. We also announced on here well before introducing them. I realise that’s not much use if you didn’t see the message, or if you started to breach the limits well after the announcement, so I apologise for that. Hopefully those posts help illustrate why we introduced rate limits, and that it wasn’t a decision we took lightly.
Invoices and bank transaction explanations: I’m guessing that’s this issue, which got flagged and fixed up as quickly as we could manage. Reading through that thread, it feels like there was maybe some confusion regarding which channel to raise a “support” query which caused a delay in resolution. This forum does perform multiple duties, but we should look at ways to improve response times here for suspected bugs and changed behaviour
Hopefully the above helps reassure you that we do take support of the API seriously, and I’d encourage you to continue to post questions and report unexpected behaviour here as you find it.
Thanks again for bearing with us, and for using our API.