Hi,
This week I took a deeper look into my fix for this bug. It involves storing the lowering in a new field within CatAssignExp
(called lowering
). It previously ran into a backend error that turned out to be caused by incorrectly inlining the lowering instead of the CatAssignExp
. The lowering could not be inlined as often as the original expression.
With this out of the way, I am now facing another error during CTFE as it cannot interpret the lowering. But similarly to the backend error from earlier, CTFE shouldn't interpret this lowering. Now I am trying to skip this step and interpret the right thing instead.
In addition, the PR for converting _d_newarray{U,iT,T}
to templates was merged and a discussion arose about including it in the changelog. I'm not sure about this, so I created this PR to add a changelog entry about the template _d_newarrayT
. If you have any opinions on this, let's discuss them there.
Thanks,
Teodor