[chongli]

It's one thing to criticize somebody over a missing feature or a bug. It doesn't take an engineer to author a valid complaint when a car door handle falls off. It's another thing entirely to criticize an entire methodology or to try to dictate exactly how something should be done.

[viraptor]

I don't think this generalises that well. We're learning some general truths that should be known to many people with enough time. You can certainly criticise whole architecture and methodology based on those ideas. For some ideas it took a long time before they became something obvious - we may be at a time when the scales are slowly tipping.

For example imagine telling someone in 2000 that they should be running an automated continuous integration pipeline rather than purely manual testing. Now it's a norm. Then, we were still discussing whether Extreme Programming was a waste of time. But we also didn't have the tools to support the CI properly.

Memory safe compiled languages seem to me like the same thing. We had things like Cyclone, OOC, and other experiments which barely anyone knew. Now we have Rust which is still hard to use, but it's gaining ground. In a decade, I hope to be pointing out to this post when someone complains that you can't criticize an entire methodology. "Remember when people said the same thing about enforced memory safety?"

Curl will not be rewritten of course. But I'm happy people push this idea. The next curl will not be written in C.

[josteink]

> It doesn't take an engineer to author a valid complaint when a car door handle falls off. It's another thing entirely to criticize an entire methodology or to try to dictate exactly how something should be done.

But if a car manufacturer consistently uses a production methodology which causes door handles to keep falling off, again and again, even after they've been re-glued on for the Nth time...

Then I think it's fair game to say that you think there's something wrong with the methodology. People suggesting alternate approaches without putting in the actual work involved may very well be "backseat drivers of the Internet"-type commenters, but that doesn't mean they're not on to something.

[chongli]

If my comment could be construed in any way as an attack on people for suggesting others use different languages such as Rust, it was not intended. I'm a fan of Rust myself and I would recommend it to anyone who wants to start a new project.

My comment was meant to pertain to the suggestion floating around that Daniel ought to undertake a total rewrite of curl in Rust. That's an entirely different ball of wax. It's like telling the car maker to tear down and build a new factory because the door handles keep falling off.

[bluGill]

>But if a car manufacturer consistently uses a production methodology which causes door handles to keep falling off, again and again, even after they've been re-glued on for the Nth time...

We know the characteristics of glue, and have switched to better glues. All our assembly jigs work with glue. When the handle falls off it is easy and cheap to glue it back on.

If we switch to screws we need to redesign the parts and the jigs; and reorder the entry assembly line. [Not realized by the engineers at the time is if they use screws dissimilar metals of the screws corrode and result in the handle falling off as well - and it is a much more expensive fix because you have to replace the handle and the part it attaches to as both have nothing to screw into)