How Effectively Agents Write D

[Peter Alexander]

On Sunday, 8 November 2026 at 18:41:43 UTC, Paolo Invernizzi 
wrote:
> On Sunday, 8 November 2026 at 12:03:39 UTC, Peter Alexander 
> wrote:
>> On Sunday, 8 November 2026 at 02:21:22 UTC, strawberry9 wrote:
>>> So the question for D in this new age of AI, is: does D make 
>>> simple problems simple, complex problems manageable, and 
>>> dangerous mistakes difficult to make?
>>
>> I think this is a very good question! Would love to explore 
>> this. Another experiment I have in mind is to take a bunch of 
>> large existing projects and have the agents e.g. implement 
>> features, refactor, fix bugs, etc. and then see what kinds of 
>> issues they hit in that setup.
>>
>> Lots of potential experiments to run!
>
> I’m toying with the idea of ​​creating a fork of LDC and simply 
> asking Claude to improve the speed and memory footprint of the 
> CTFE...

I did something similar with DMD. I have a fork that Claude has 
been working on for a while. Has ARM support (only tested on 
Mac), much faster compilation for my main project (>4x without 
opts, >100x optimised) and runtime generated code is just 
slightly behind LDC now. Have not been rigorous with it, but it's 
my daily driver now and is stable at least for my personal 
projects.