Re: [PoC] Improve dead tuple storage for lazy vacuum

On Mon, Apr 15, 2024 at 04:12:38PM +0700, John Naylor wrote:
> - Some paths for single-value leaves are not covered:
> 
> https://anarazel.de/postgres/cov/16-vs-HEAD-2024-04-14/src/include/lib/radixtree.h.gcov.html#L904
> https://anarazel.de/postgres/cov/16-vs-HEAD-2024-04-14/src/include/lib/radixtree.h.gcov.html#L954
> https://anarazel.de/postgres/cov/16-vs-HEAD-2024-04-14/src/include/lib/radixtree.h.gcov.html#L2606
> 
> However, these paths do get regression test coverage on 32-bit
> machines. 64-bit builds only have leaves in the TID store, which
> doesn't (currently) delete entries, and doesn't instantiate the tree
> with the debug option.
> 
> - In RT_SET "if (found)" is not covered:
> 
> https://anarazel.de/postgres/cov/16-vs-HEAD-2024-04-14/src/include/lib/radixtree.h.gcov.html#L1768
> 
> That's because we don't yet have code that replaces an existing value
> with a value of a different length.

I saw a SIGSEGV there when using tidstore to write a fix for something else.
Patch attached.