Shuhei Kadowaki
Shuhei Kadowaki
Thanks for the update! I had been delaying the LCT update, so this is very helpful. > The overall trend is less > reliance on `norequire` as a way of...
Thanks! I can look into these sometime this week. I am also planning to take a look on your PRs on LCU side again too.
Yes, this false positive is very annoying. I will try to fix it sometime in this week.
We can wrap `length(list) >= 1` into a separate function, and then we can propagate type information of `list::Vector{Int}` to the `then` branch: ```julia julia> code_typed((ItrList,); optimize=false) do ilist list...
> But here is another problem that JET's analysis fails to propagate that information. I found it is due to the implementation of the type lattice that JET uses, so...
So, it appears that Revise is capable of changing the line number information of a method that has been moved without redefining it? Since JET records and caches this line...
There is no proper support for 1.6 so that is probably the best way to workaround the compat problem.
It should work. Can you provide a concrete example where it does not work?
Thanks, I will take a look later.
Okay, to avoid these issues, you can extract the variable into a local variable. In the first case, for example, the false positive error will not be reported if you...