Sukera

Thank you! I'm not too familiar with the impact a higher precision could have here - naively I didn't think it would impact things, so maybe it "just" uncovers other...

Any update on this?

I think a problem similar (identical?) to this came up on [zulip](https://julialang.zulipchat.com/#narrow/stream/225540-gripes/topic/IDE.20support/near/258146411) today: This successfully shows completion:  While this breaks:  Is there a concrete place I could begin...

Huh, just noticed this: https://github.com/JuliaGPU/GPUCompiler.jl/blob/d9b8f4747cf789139fb2586ee67ced93967a8237/src/rtlib.jl#L152-L160 I suppose that's already the same thing, except it doesn't take care of the additional versioning scheme with 1.9+? Should I delete this or close...

The failure on julia master seems to be due to https://github.com/JuliaGPU/GPUCompiler.jl/issues/506

> Yeah, I'm not sure why. When trying in isolation, reading invalid bitcode does throw a proper error. I'm not sure why it suddenly exits there sometimes... Well, I can...

I'm torn on this - on one hand, the current printing annoyed me in the past when inspecting generated LLVM IR of a live session, on the other hand I...

> Why is having a copy-pasteable Ptr representation actively bad? It gives the illusion that the value behind that pointer is alive, when that's not necessarily the case. Case in...

> The reasoning for this is to make repr valid / 2-arg show abide by this part of the show docstring: > What you're talking about "we should show humans...

> This said, I do think parsibility is worth prioritising, for cases like when a test fails and you want to copy-paste the value into the REPL to inspect the...