Ecosyste.ms: Timeline
Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.
jishnub created a comment on an issue on JuliaLang/julia
What works currently is obtaining the `ref` using `cconvert` and then fetching its parent. ```julia julia> A = zeros(2); julia> parent(Base.cconvert(Ptr, A)) 2-element Memory{Float64}: 0.0 0.0 ```
nsajko created a comment on an issue on JuliaLang/julia
> AbstractAlgebra.jl from Nemo implements this exact function The implementation has since been improved, xref https://github.com/Nemocas/AbstractAlgebra.jl/pull/1871: https://github.com/Nemocas/...
giordano closed a pull request on JuliaLang/julia
Restore support for checking for UndefVarError variable name in at-test_throws
Fix https://github.com/JuliaLang/julia/issues/54082 Arguably this was a breaking change (as a consequence of https://github.com/JuliaLang/julia/pull/51979). But regardless, it seems like useful...
Vobarkun created a comment on a pull request on JuliaLang/julia
Yes, this appears to fix it. The features are now detected correctly also for me and the cache file gets properly rejected.
jishnub pushed 1 commit to master JuliaLang/julia
- rename: invalid -> incompatible cache header (#56240) Falling back to the older serial precompilation process is bas... aa51abe
jishnub closed a pull request on JuliaLang/julia
rename: invalid -> incompatible cache header
Falling back to the older serial precompilation process is basically a bug (except for if a manifest hasn't been resolved) so https://github.com/JuliaLang/julia/pull/52619 added more info on why it...
d-netto pushed 3 commits to dcn-all-task-profiler JuliaLang/julia
d-netto created a comment on a pull request on JuliaLang/julia
@vtjnash: IIRC your comments stem from the fact that a GC may get interleaved with `jl_get_all_tasks_arraylist` and invalidate the contents of `live_tasks` -- both by freeing unreachable tasks and ...
vtjnash created a review comment on a pull request on JuliaLang/julia
Using `mtarraylist_get` is UB outside of a gc-unsafe region, since the RCU period could expire at any time on this memory
vtjnash created a review comment on a pull request on JuliaLang/julia
Accessing the contents of this list would be UB as soon as you release this mutex, since you aren't in a gc-unsafe region. This mutex shouldn't really be necessary, since if you want to access any ...
vtjnash created a review comment on a pull request on JuliaLang/julia
```suggestion GC.@preserve dest src memmove(pointer(dest), pointer(src), aligned_sizeof(T) * n) ```
vtjnash created a review comment on a pull request on JuliaLang/julia
Although that happens in parallel with the runtime, so I think that still might be okay, for something the user has to request pretty explicitly
vtjnash created a review comment on a pull request on JuliaLang/julia
Ah, gotcha, that does make sense. Symbolizing something big (such as the LLVM debug info when asserts are on) can take quite some time.
vtjnash created a comment on a pull request on JuliaLang/julia
Yep, we've eliminated most calls to `remove_linenums!` instead (except from `@inferred` which hasn't been fixed yet to remove that)
oscardssmith created a comment on a pull request on JuliaLang/julia
I think it's more that a lot of code somewhat reasonably believed that Array was a fixed point of `parent`. the real problem is that parent is a bad API, but there's not much to do about that now.
oscardssmith created a comment on a pull request on JuliaLang/julia
don't we still need to turn this on by default?
vtjnash created a comment on a pull request on JuliaLang/julia
Oh, what, that seems a ridiculously bad assumption. That is literally expected to only be true in rare cases, and not commonly true most of the time: ```julia julia> V = view(A, 1:1, :) 1×2 view...
oscardssmith created a comment on a pull request on JuliaLang/julia
looks like there are a few 8x (vectorization probably) failures here.
vtjnash pushed 1 commit to master JuliaLang/julia
- Split up the one big codegen lock into per-function locks and dependency edge tracking (#56179) Disjoint content can... cd99cfc
vtjnash closed a pull request on JuliaLang/julia
Split up the one big codegen lock into per-function locks and dependency edge tracking
Disjoint content can be LLVM optimized in parallel now, since codegen no longer has any ability to handle recursion, and compilation should even be able to run in parallel with the GC also. Individ...
oscardssmith created a comment on a pull request on JuliaLang/julia
looks like more negative than positive, but lots of the negative is ~5%
vtjnash created a comment on a pull request on JuliaLang/julia
Looks like it gave LLVM a little bit of trouble on a few benchmarks, but overall seemed potentially worth doing this for the simplification of everything. I am not sure that overall it matters that...