Christoffer Lerno
Christoffer Lerno
Yeah, this is something I've been pondering. I think that you should probably have a non-copying version and a copying and they should be distinctly named even.
Can you add a test for it and add a note in the releasenotes?
There is a problem with non-retained keys, and that is returning hash maps. So imagine you want to pass a string-value hash map to a user of your library. The...
From what I read, setting it to double should indeed set compiler.platform.riscv.flen = 8, perhaps there is some additional flag that has to be packed into the LLVM IR for...
From what I can see, these are the possible ABI types: ``` .Case("ilp32", ABI_ILP32) .Case("ilp32f", ABI_ILP32F) .Case("ilp32d", ABI_ILP32D) .Case("ilp32e", ABI_ILP32E) .Case("lp64", ABI_LP64) .Case("lp64f", ABI_LP64F) .Case("lp64d", ABI_LP64D) .Case("lp64e", ABI_LP64E) ``` The...
Ok, so hang on, let's see if we can break this down: 1. floats: none on riscv32 => ilp32 riscv64 => lp64 2. float => ilp32f / lp64f 3. double...
(We don't have a CI for Riscv, so everything is a bit experimental BTW, it would be great to have something, all we have is a baremetal riscv test)
So what would you propose?
Ok, I was thinking of working on this. This looks like we need something like a list of RISCV capabilites, right? For x64 I do this by grouping them: none,...
Ok, I've fixed so that flags can be added manually using --cpu-flags a default for cpu type will be added.