cherleey wrote:
Thank you for the review, and I apologize for the state of this contribution.
I have to be honest about my situation first. This patch is an output of a
project where I have been using AI agents to work on top of llvm-project. The
diagnosis, the patch itself, and the measurements were produced by AI. Because
of that, my own understanding of this change is not deep enough to write the
description and comments by hand — and my English is not good enough for it
either. So this comment and the updated descriptions are, again, written with
AI. I could not do the part you asked for by hand, and I am sorry about that.
If this is not acceptable, I fully understand, and I will follow your decision
on how or whether to proceed.
What I could do is the split you asked for:
- This PR now contains only the **deferred PdbIndex creation**
(`SymbolFileNativePDB.{cpp,h}`, +92/−15). The branch was force-pushed
accordingly.
- The **lazy TPI/IPI stream loading** is now a separate PR: #211268
(`PdbIndex.{cpp,h}`, +49/−8), stacked on top of this PR — the lazy
`tpi()`/`ipi()` accessors rely on the abilities check added here to guarantee
stream presence, so this PR needs to land first.
- The two commits together are byte-identical to the previous head of this PR.
- One caveat: the memory numbers were measured with both changes applied
together; per-part numbers do not exist yet.
https://github.com/llvm/llvm-project/pull/211178
_______________________________________________
lldb-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
