Ah, yes, finally gitlab will have the same uptime leves as GitHub.

Side note, Jesus Christ on a bicycle, this site has a ridiculous amount of ads. Literally could not read the article at the first try, had to reload and enable reader mode before the 1937391 popups started jumping on my face.

For instance:

> "The caching layer is causing a 400ms overhead on cold requests. Here's the trace."

Cool, but how do you know that? Are you sure you're interpreting the trace correctly? How many measurements have you done? Sure that there isn't a historic reason why the caching layer behaves in this way, or a conflicting requirement that lead to choosing this latency over some other, worse consequence?

In the same way:

> "The current error handling swallows exceptions silently, which is making debugging hell. We should propagate errors to the caller"

Agree in principle, but did you check git blame to see if there was a rationale? Or asked someone else? How much change would this require? Could it break consumers of our code?

Granted, the "polite" messages didn't care any of this information either, but at least they didn't almost preemptively shut down the conversation by laying down what is at best an well informed guess and at worse purely personal opinion as if it's irrefutable truth.

Perhaps instead of debating whether to be (too?) polite or direct, it's better to focus on providing as much information as you can, anticipate follow up questions, and close with clear actionable next steps, something which is also missing from both versions of the messages.

"Well, those are hypothetical examples" cool, another evidence that we're all talking out of our asses about the sex of the angels, here. Maybe if the author didn't try so much to characterize themselves as perfectly infallible information delivery machine, then I wouldn't have nit-picked so much their half-assed hypotheticals.

One (I think?) bug report and one suggestion:

- bug: can't get past the 5th project on mobile, whole screen just go black and I have to reload the page. - suggestion: maybe make cd ~ or something like that bring the user back to the actual initial state, with the login banner and stuff. Just running clear results in an empty shell, and the user has to remember to run help again.

P.S.: I ran the secret command. You son of a gun XD

P.P.S.: Two suggestions, actually: display your CV with pdfjs or something, or make an HTML version. It's annoying to have to download files.

I'm not sure I can trust the author's characterization of Roy, though. I got the impression that they don't like any of the people they interviewed (which, you know, fair), but that doesn't get even close to the depths of hatred towards Roy that they sub-textually exude throughout the article.

If their portrayal is even half accurate, though, that's a perfectly reasonable amount of hate.

Maybe if you read past these paragraph it would have been clearer?