[llvm] [llvm-objdump] Honor ELF symbol sizes in selective disassembly (PR #224068)

[James Henderson via llvm-commits]

jh7370 wrote:

Taking this discussion out-of-line. TL;DR: I feel like this and likely several of the other PRs you've written recently are probably violations of LLVM's AI Policy.

Quoting the policy:
> The contributor is always the author and is fully accountable for their contributions. Contributors should be sufficiently confident that the contribution is high enough quality that asking for a review is a good use of scarce maintainer time, and they should be able to answer questions about their work during review.

I'm not confident that you understand the code your AI agent is generating. The fact that you've posted the explanation that the agent wrote rather than writing something in your own words further enhances this, especially as I specifically asked you _not_ to use LLM assistance with the explanation. If you don't understand your own code, you can't take ownership of it and therefore the code is not welcome in LLVM.

Let's put it another way: would you be able to write this PR again from scratch without an AI agent's assistance, now that you've read through it and had your agent explain the original to you? If not, I don't think you understand the code sufficiently to be able to write this patch, making it extractive (see the AI Policy for what I mean by this).

Another minor violation is that your ping comments appear to be entirely generated by your AI agent. This suggests you aren't interacting properly with the PRs and you are mostly just getting an agent to do all the work.

Please don't get me wrong, I'm not suggesting you have bad motivations or anything, but it's important that code that is added is understood. I'm struggling to follow the change without spending significant time staring at it, and if I as one of the lead maintainers in the area cannot follow the change and you can't explain it to me on your own, then it's unlikely anybody else can, making this code unmaintainable.

Setting out the path forward: I think you would be best pausing development on all but one of your active PRs in LLVM and focus on one at a time, making sure you fully understand all the code that is written. In the ideal world, you'd rewrite that one from scratch without your agent's assistance (although using an agent to review your changes and help suggest how you fix any errors - without writing the code itself - would certainly be worth it). This will help you get to know the relevant area of the codebase yourself, which increases your knowledge and your value to the LLVM community.

One final question: what is your motivation for fixing this and the other LLVM issues you've been trying to fix?

Tagging others who have recently been involved in your PRs, for visibility: @MaskRay, @rjmansfield, @oontvoo, @drodriguez, @alexander-shaposhnikov.

https://github.com/llvm/llvm-project/pull/224068