Krystian Guliński

Note that it happens with some Polish letters like ś ć ą, but not with ó for example. Also only appears for me with the .pdf format, others are fine

Yes, this is basically all about parsing the `args...` provided by the user into a format that the end structure (DataFrames or DTable) can take and manipulate the data with...

Just hit the same issue. Thanks for the investigation, had no clue why this was happening Setting the `pool` kwarg to false worked for me as a workaround `d =...

This still happens in other places in the `Dagger` codebase, but it's a Base/InlineStrings thing to resolve. Mostly happens due to having CSV.jl loaded Reopening until fully resolved https://github.com/JuliaLang/julia/pull/43426

Clue for later: Seems like pure distributed stress tests also suffer from slowdown after a while, might not be a Dagger issue after all, but I'll look into that at...

@KristofferC Can we merge and tag this? I confirmed it works fine on my case

Benchmarked it on a simple hello world endpoint with `ab` ## before ``` Requests per second: 54.70 [#/sec] (mean) Time per request: 1827.989 [ms] (mean) Time per request: 18.280 [ms]...

could you try getting an effect like this? when you iterate a gdtable you should be getting dtable type elements so should be fairly easy DataFrames code could be a...

Scaling works fine with processes though each worker has 4 threads in this setup 4 workers time -> 696.21 8 workers time -> 234.21 speedup -> 2.97

reverted https://github.com/JuliaParallel/DTables.jl/pull/45 performance was weird and i was getting hangups locally we need to reevaluate how to implement this correctly and possibly make it opt in instead of active by...