Google's take on memory safety

[Lance Bachmeier]

On Saturday, 16 March 2024 at 18:47:40 UTC, harakim wrote:

> That seemed like a great goal, but obviously that couldn't be 
> it because D would be a lot more popular. Now you say this is 
> true!?

The thing is, it is only very recently that it works well enough 
to be a practical way to incorporate C code into your project.

Initially it just compiled C code. The only problem is that 
there's very little C code in the wild. It's a mix of C, the 
preprocessor, and compiler extensions. Over time, that was 
addressed, but the one remaining (big) limitation was that it 
couldn't deal with function-like macros. Walter recently remedied 
that, so for the first time you can grab a 10,000-line .c file 
and have an expectation that it'll compile cleanly. The only 
issue I've had is a minor bug that has an easy workaround. Other 
than that, tens of thousands of lines of C code has compiled for 
me. Note that right now you need to use the DMD release candidate 
for 2.108.

> This is such an enormous benefit! This should be a huge driver 
> for D. With this functionality, I could easily see it being the 
> language of the year on tiobe, for example.

It's hard to predict usage, but it's a darn good tool for working 
with a legacy C codebase. With module support, you can even add 
import statements to your C files to import D functions. I've 
used that in a couple cases to eliminate dependencies.