Improve compilation latency by reducing recursion limit

[tkf]

An alternative fix to #412.

Compilation benchmark using the example in #412:

FOLDL_RECURSION_LIMIT	w/ Julia 1.6.0-DEV.536	w/ Julia 1.5	w/ Julia 1.4
0	5 sec	8 sec	6 sec
1	6 sec	15 sec	8 sec
2	7 sec	21 sec	10 sec
3	8 sec	30 sec	12 sec
4	9 sec	34 sec	14 sec

FOLDL_RECURSION_LIMIT = Val(0) slows down ["teerf_filter", "noinit"] (and speedups ["filter_sum", "1000", "RandomFloats", "noinit", "base"]):

ID	time ratio	memory ratio
["filter_sum", "1000", "RandomFloats", "noinit", "base"]	0.31 (5%) :white_check_mark:	1.00 (1%)
["teerf_filter", "noinit"]	226.80 (5%) :x:	832.57 (1%) :x:

--- https://github.com/JuliaFolds/Transducers-data/blob/benchmark-results/2020/08/10/070744/result.md

Interestingly, FOLDL_RECURSION_LIMIT = Val(1) improves the performance:

ID	time ratio	memory ratio
["filter_sum", "1000", "RandomFloats", "noinit", "base"]	0.26 (5%) :white_check_mark:	1.00 (1%)

--- https://github.com/JuliaFolds/Transducers-data/blob/benchmark-results/2020/08/10/064257/result.md

Re: ["filter_sum", "1000", "RandomFloats", "noinit", "base"], see #403.

Commit Message

Improve compilation latency by reducing recursion limit (#412)

The tail-call pattern introduced in eb430cf76a5c25aa909858c20424aead21cfecfd (#403) for linear indexing arrays seem to invoke a large compiler latency (see #412 for an example). This patch tries to mitigate the problem by reducing the FOLDL_RECURSION_LIMIT to 1. This still allows two recursive calls. All the benchmarks do not show noticeable degradation (presumably because recursion deeper than this is not exercised in the benchmarks). Note that setting it to 0 (one recursive call) shows 2x slowdown in ["teerf_filter", "noinit"] benchmark.

[codecov[bot]]

Codecov Report

Merging #413 into master will increase coverage by 1.31%. The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #413      +/-   ##
==========================================
+ Coverage   89.52%   90.84%   +1.31%     
==========================================
  Files          25       25              
  Lines        1585     1594       +9     
==========================================
+ Hits         1419     1448      +29     
+ Misses        166      146      -20     

Flag	Coverage Δ
#unittests	90.84% <ø> (+1.31%)	:arrow_up:

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
src/processes.jl	93.03% <ø> (ø)
src/library.jl	93.15% <0.00%> (-0.28%)	:arrow_down:
src/show.jl	89.51% <0.00%> (+0.07%)	:arrow_up:
src/unordered.jl	96.61% <0.00%> (+0.31%)	:arrow_up:
src/progress.jl	89.79% <0.00%> (+1.02%)	:arrow_up:
src/core.jl	85.56% <0.00%> (+1.06%)	:arrow_up:
src/groupby.jl	82.14% <0.00%> (+5.77%)	:arrow_up:
src/reduce.jl	84.55% <0.00%> (+9.14%)	:arrow_up:
src/basics.jl	92.10% <0.00%> (+10.52%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 1f67507...05bbd6a. Read the comment docs.