Benjamin Bannier

Unassigning myself since ATM there is no clear path forward which doesn't involve a lot of work like implementing a standard libary-quality smart pointer library.

If we end up adding this we should add it to the Spicy plugins as well. Whenever a Spicy grammar hits a call to [`spicy::accept_input`](https://docs.zeek.org/projects/spicy/en/latest/programming/library.html#spicy-accept-input) we trigger code which is...

We have #1444 for removing dead stores but didn't have a benchmark showing it was useful. Would you be open to compare `topic/bbannier/dead_stores` against the baseline ec87b43037dba50648cb93be8940a4db23658905?

I believe I already looked into these. The `#pragma once in main file` diagnostic is also described on [SO](https://stackoverflow.com/questions/71184408/how-can-i-avoid-pragma-once-in-main-file-in-gcc-when-using-precompiled-header) and probably would require some sort of workaround. The other warnings...

The issue here seems to be that the parsing with `&chunked` doesn't directly populate `xs` with a value, but this only occurs after the chunked parsing is done.

> Just to make sure this didn't go under, there's no issue if I remove the `&convert`: Yep, that's the normal assignment of the value after parsing is done. We...

There are actually more things wrong with chunked parsing. After a field is done parsing in addition to potentially setting the unit field we currently also update offsets, https://github.com/zeek/spicy/blob/daa2e3ce48a92dcf6662df724a8bb1c8d5f8df4b/spicy/toolchain/src/compiler/codegen/parser-builder.cc#L733-L834 ```ruby...

@rsmmr `$$` refers to the converted value. Like you wrote, whether that makes sense for the user is up to them, e.g., if `&convert` does a base64 conversion the results...

@MP-Corelight, we have a PCAP here and an idea how to fix the issue, don't we have enough information to fix this even without OP?

Recreated by OP as #4543.