Jonas Schulze

Indeed, if I funnel all the communication through thread 1, it works! ```julia # ... p = D.@spawnat w @sync begin funnel = Channel() T.@spawn stillepost(i, c1, funnel) stillepost(i+0.5, funnel,...

TL;DR: There is a different issue. Above example is fixed by using `addprocs(..., lazy=false)`. I tried to write some tests that send and receive from all combinations of threads, processes...

Latest attempt is JuliaLang/julia#38405

I just watched the [State of Julia 2022](https://youtu.be/N4h46_TCmGc?t=980) which claimed that Distributed would now be thread-safe. However, the snippet above (https://github.com/JuliaLang/Distributed.jl/issues/73) remains broken on 1.8.5 as well as 1.9.1 both...

Should the associativity within `reduce` and `mapreduce` be added to the list of implementation-defined behavior? See discussion in #53871.

Using Julia 1.10 and Plots 1.39 via SSH gave me the same error message excluding the segfault. The images produced looked odd: `yticks` labels were cut off partially, the `xlabel`...

> It would be so much simpler if the cells were stored in the .jl file in the "real" order of how they appear in the document. If that would...

As a tie breaker when creating a topological order, one could use the cell having the alphabetically smallest ~~hash~~ UUID. Assuming that the cell ~~hashes~~ UUIDs do not change (e.g.,...

>> As a tie breaker when creating a topological order, one could use the cell having the alphabetically smallest hash. > > The problem with this approach is that you...

> I don't think that I did too much reordering in the document and still got these huge diffs. Ok, it might be, that the dependencies changed due to the...