Turned out that it was actually my mistake. The notification setup sends the payload as JSON that uses quote marks (”) as the delimiters. Then the data that was supposed to be in one of the fields included quote marks, making the JSON invalid.
// @variablepulserate
Spent the week setting up and testing an automagic notification system for our CI builds to find out that when it is actually used it does not work when it is actually triggered. The CI build reports the notification being sent but it never landed on the receiving side. FFS…
@matigo Well, I wonder who this ”hero” would be so that I could pay them a visit… The infrastructure should make things easier, and this is a great example of things not being easy…
Getting really irritated at work. Our code is hosted on Azure DevOps that seems have intermittent issues with their CI/CD pipeline hosts. The jobs are failing due to the host disconnecting at some point, sometimes it takes only a minute or so while sometimes the job is almost done.
Yes, we know our jobs are not the fastest, but could you just keep the infrastructure up and running so that we don't end up wasting everybody's time by re-running the jobs while keeping our fingers crossed…
@matigo Well put. I have also never seen anyone be able to give answer to that ”what do you want to track” question…
Days like today when all my plans go out of the window pretty much immediately after I actually start checking the first item of my todo list are not fun at all…
@matigo Indeed. This was code that was written back in 2018, and the client-side "bug" we fixed then surfaced the underlying issue from the backend. We could rewrite the client-side logic to fix the issue for good, but that requires a lot of tweaking here and there, so getting it fixed is not that fast to do.
On a related note, I have been rewriting some app logic not directly related to this issue as a small side project1, and if I would get that rewrite done also for this bit of the app the issue would get fixed for good.
We are using HATEOAS for "security reasons", and the links for the various operations are behind an API controller. That then means that we need to make sure that we call the API controller first before we call the actual operation, and that has been manually managed. I am rewriting this to be automagically managed, meaning that the API controller response is refreshed if the link to the actual operation is missing. This would make the developers' life a lot easier.
@variablepulserate Yeah, we have somewhat similar phrase used here. It can snow in June as well, and the last snow far up North can stay on the Northern slopes of the mountains (well, they are just big hills here) until July…