[bawolff]

I'm kind of doubtful of this. Just to take print debugging as an example. While there is definitely a time and place for debuggers, i think print debugging can be shockingly efficient in many circumstances.

Besides, typing speed is usually not the limiting factor in programming speed. I think overall efficiency needs to take a broader look.

[sagarm]

I default to printf debugging myself, but it can be an indicator that someone doesn't know their tools.

[erik_seaberg]

Breakpoints are likely to screw up a lot of concurrent user requests, so in prod we debug microservices by comparing pushed logs.

[llm_trw]

The typical bell curve meme: https://imgflip.com/i/9a9gm7

The fun begins when you can't debug by printf because the buffer is shared by multiple processes that change it while the printf command is running.

[erik_seaberg]

Yeah, network logging has latency but it does pack the messages into structs and demux them. The remaining debugger problems are:

- Do you know which service tier has the issue?

- Can you halt one thread that sees the issue, not all of them on an instance?

- When the request times out, do you have to start over and make another prod user sad?