Hi,

This week I continued with _d_newitem{T,iT,U} and _d_arraycatnTX.

I only implemented a template _d_newitemU and left the introduction of the call to memset(0) / memcpy(T.init) up to the compiler. I am currently having trouble setting vthis for nested structs. It is generally set in e2ir.d and reinventing this wheel in expressionsem.d is a bit trickier or I just don't know where to look.

I've fixed all easily reproducible bugs related to _d_arraycatnTX (PR here). The ones that remain lie inside the compiler, unfortunately. One of them was easy to track. Here 2 strings are concatenated and then the .ptr of the result is used. Because of the missing \0 at the end of strings, this assert was failing. The ctor of Lexer says explicitly that its second argument should be null-terminated. The solution was to manually add a ~ '\0' at the end of the concatenation above. I don't know how this error didn't manifest itself with the older hook. My intuition says that the remaining builkite failures also come from .ptr being used on strings, but I find it difficult to track them down.

Thanks,
Teodor

On Monday, 24 October 2022 at 20:23:05 UTC, Teodor Dutu wrote:

Great work! Keep it up and soon enough we'll have a nice runtime! I have had some problems before when working on the array things