They ignored me for a couple weeks, but I kept asking and eventually someone answered and said yeah you'll get the 504 Gateway Timeout if you try to update the account balance to the same value that it already has.

OK, that makes sense to you? What if there was an actual gateway timeout?

You would also get a 504 Gateway Timeout in that case.

They apparently have no plans to modify this service's behavior.

I let it go, I just moved on with my life. I did not find myself awed or stunned with the mountains of work and talent and creativity and foresight and intelligence and luck that went into it.

So, dicks and idiot middle managers. I think you made his point.

Someone could have documented this and/or simply answered a damn question instead of wasting weeks of opportunity cost. Too many dicks and idiots on that website, it seems.

"who honestly cares if it gets fixed" - everyone that comes after this and has to deal with it and waste weeks of their life trying to pry some required nugget of info from a passive aggressive group of maintainers of stuff which actively ignores industry standards?

And yeah, I can put myself in those shoes, as I've been on both sides of this. Treating newcomers in your communication channel (slack, etc) with the same level of urgency YOU would want (or need) when you ask for help from some other team is a bare minimum.

A 504 is a gateway timeout due to the server failing to complete the action, hell, a 400-level for "you fucked up" would be preferable.

  let x = 2 // runs
  x = 3;    // runs
  x = 3;    // runtime error, you made a mistake

In the real world, updating your balance to the same number should be a 200, however in some weird financial world, it may be a 400 because you may not be allowed to set the same balance due to some financial rule. A 500 is a terrible choice unless setting the balance caused an error at your code/infrastructure level.

A shitty environment can't foster good work relations.

End users of my colleague's application were now saying they were seeing inactive client data coming up in their app, and it was causing confusion.

During the time trying to figure out the new behavior, the main client finally got something in writing about "here's how it works" from the data vendor...

* Rule 1...

* Rule 2...

* Rule 3: If a client record has not had any data updated, we will only send over the client name and header info, but no additional data records will be sent over.

* Rule 4....

* Rule 5: If a client record is no longer active in the system (for example, has been deleted), we will only send over the client name and header info, but no additional data records will be sent over.

Questioning back to the original data provider - if we get a client name and header info, but no additional data records - what does that mean? Does it mean the client is inactive/deleted, or simply that they've had no data change since yesterday?

Absolutely no coherent response, and I doubt there will be one. The defense of "no no, we have to keep stupid behaviour around because people rely on it!" doesn't even stand - this change came about in the last couple weeks, after having worked this way for more than a year.

It's both maddening and frightening how fragile so much daily data exchange is.

That would still be fine... if the data was static. But it's not - and changes are communicated; but not to the relevant tech teams.