What I really like about this story is that free and unfettered access to the standard/specs, the implementation, the ability to spin up a reproducer, to see all moving parts with X-ray vision, come to a solution and having said solution accepted upstream, all in one go, at your own pace AND being able to share the whole story without needing permission - the power of Open. Not working around bugs in whatever proprietary thing. Mervellieux! Excuse my French. #OpenFTW
Once upon a time I encountered some small issue (not a bug) with Postgres. I was able to get to the very source of it in the C code, understand what's happening, and adjust my code accordingly. It took maybe an hour.
Upon doing that, I shuddered, as I remembered troubleshooting an issue (a real confirmed bug) inside Oracle 9. It was by poking in the dark, trying to trigger a condition that made a server process crash, then trying to help Oracle engineers reproduce that. It took a week.
> I was able to get to the very source of it in the C code, understand what's happening, and adjust my code accordingly.
The ability to do this has saved my bacon on more than one occasion. It's why (outside of work I do for an employer) I won't develop using libraries/frameworks that I don't have the source code to.
Indeed. Adding a significant closed-source component to the tech stack would require a ton of strong and unique advantages to jusufy it. Very, very few things paas this test.
Yup, ages ago, quite a bit of trouble getting Cisco to admit a problem with a router. It should not matter where the traffic is coming from and if anything it would be easier to figure out if I could write code to try to pick the problem out of the firehose. Nope, had to demonstrate it with no code at all before they would look. Finally got enough workstations running dir . /s > nul to cause it to mess up. (I knew it was load related.) And HP wasn't even interested in my file that plotted upside-down. (I suspect even if they had been willing to look they wouldn't have wanted to fix it--there was probably enough buggy hardware out there that they would take the position of declaring it right.)