typos

[spaette]

delimit
negligible
potatoes
satisfied

The basis of the decision to use the string fais in the benchmark subdirectory files might welcome a clarification.

$ grep -nr delimite ConcurrentCollections.jl
ConcurrentCollections.jl/src/dict.jl:423:avoid stepping into other base cases by using an empty slot to delimite the base
$ grep -nr neglegible ConcurrentCollections.jl
ConcurrentCollections.jl/src/dlcrq.jl:222:# It looks like some non-neglegible amount of time is spent in
ConcurrentCollections.jl/src/dict.jl:24:# seems that the effect is neglegible. The performance is much better than
$ grep -nr potatos ConcurrentCollections.jl
ConcurrentCollections.jl/benchmark/hotpotato/run.jl:9:    potatos = []
ConcurrentCollections.jl/benchmark/hotpotato/run.jl:17:            push!(potatos, (; trialid, result, ntasks, impl = :dlcrq))
ConcurrentCollections.jl/benchmark/hotpotato/run.jl:20:            push!(potatos, (; trialid, result, ntasks, impl = :base))
ConcurrentCollections.jl/benchmark/hotpotato/run.jl:27:    return (; potatos, fais, repeat, duration)
ConcurrentCollections.jl/benchmark/hotpotato/plot.jl:9:potatos = map(rawdata["potatos"]) do info
ConcurrentCollections.jl/benchmark/hotpotato/plot.jl:39:datadf = vcat(DataFrame(potatos), DataFrame(fais); cols = :union)
$ grep -nr satisifed ConcurrentCollections.jl
ConcurrentCollections.jl/src/dlcrq.jl:62:            @goto satisifed
ConcurrentCollections.jl/src/dlcrq.jl:77:    @label satisifed
$ 

Was the author's intention is no longer used (without adding other mechanisms like ref counting)?

$ ed -s ConcurrentCollections.jl/src/dlcrq.jl <<<'222,229p'
# It looks like some non-neglegible amount of time is spent in
# `Vector{Waiter{T}}(undef, _)`, presumably zeroing out the elements.  A tricky
# part of this is detecting that the CRQ is no longer used. Since the CRQ may be
# closed just after the FAI and before the CAS on the slot is complete, (naively
# thinking) linear scan the `.ring` array is required for checking that the CRQ
# is not longer used (without adding other mechanisms like ref conting). So,
# ATM, only the newcrq that lost CAS is cached (which still helps the
# performance).
$ 

Was the author's intention mutated here?

$ ed -s ConcurrentCollections.jl/src/dlcrq.jl <<<'117,127p'
function trash!(w::Waiter)
    if assertion_enabled()
        old, success = @atomicreplace w.state WAITER_INIT => WAITER_TRASHED
        if !success
            error("unshared waiter mutaed: state: $old")
        end
    else
        @atomic :monotonic w.state = WAITER_TRASHED
    end
    w.task = nothing
end
$

The following shell script is authorised for use and/or modification by the repo Maintainers.

$ cat typos.sh
#!/bin/sh

# sed -i "s/conting/counting/g" ConcurrentCollections.jl/src/dlcrq.jl
sed -i "s/delimite/delimit/g" ConcurrentCollections.jl/src/dict.jl
sed -i "s/neglegible/negligible/g" ConcurrentCollections.jl/src/dlcrq.jl
# sed -i "s/mutaed/mutated/g" ConcurrentCollections.jl/src/dlcrq.jl
sed -i "s/neglegible/negligible/g" ConcurrentCollections.jl/src/dict.jl
sed -i "s/potatos/potatoes/g" ConcurrentCollections.jl/benchmark/hotpotato/run.jl
sed -i "s/potatos/potatoes/g" ConcurrentCollections.jl/benchmark/hotpotato/plot.jl
sed -i "s/satisifed/satisfied/g" ConcurrentCollections.jl/src/dlcrq.jl
$ 

[spaette]

closed due to age