Try JSON3

[mmiller-max]

I think JSON3 could bring some speed improvements, and we could possibly make use of StructTypes to help read OSM graph straight into an OSMGraph, maybe

[mmiller-max]

Using a 1km map centred on Melbourne, JSON looks quickest and has fewer allocations. It is twice as quick as xml so I think we should be recommending this in docs/tutorial @captchanjack ?

Graph download in two formats:

julia> data = download_osm_network(:place_name,
                                   place_name="melbourne, australia",
                                   network_type=:drive,download_format=:json,
                                   save_to_file_location="melbourne_drive_network.json");
julia> data = download_osm_network(:place_name,
                                   place_name="melbourne, australia",
                                   network_type=:drive,download_format=:osm,
                                   save_to_file_location="melbourne_drive_network.osm");

JSON benchmark

julia> @benchmark graph_from_file($("melbourne_1k.json"))
BenchmarkTools.Trial: 1075 samples with 1 evaluation.
 Range (min … max):  4.066 ms … 51.116 ms  ┊ GC (min … max): 0.00% … 88.51%
 Time  (median):     4.289 ms              ┊ GC (median):    0.00%
 Time  (mean ± σ):   4.640 ms ±  3.445 ms  ┊ GC (mean ± σ):  5.38% ±  6.61%

    ▆█▆▂                                                      
  ▆███████▅▅▅▆▆█▇▅▆▆▇▄▃▃▃▃▂▃▃▃▃▂▂▂▂▂▂▂▂▂▂▂▂▁▂▂▁▂▂▂▁▁▁▂▁▂▁▁▁▂ ▃
  470 ms         Histogram: frequency by time        5.79 ms <

 Memory estimate: 4.20 MiB, allocs estimate: 43720.

JSON3 benchmark (deserializer = (file) -> JSON3.read(open(file), Dict))

julia> @benchmark graph_from_file($("melbourne_1k.json"))
BenchmarkTools.Trial: 1046 samples with 1 evaluation.
 Range (min … max):  4.135 ms … 52.044 ms  ┊ GC (min … max): 0.00% … 84.64%
 Time  (median):     4.445 ms              ┊ GC (median):    0.00%
 Time  (mean ± σ):   4.768 ms ±  3.524 ms  ┊ GC (mean ± σ):  5.22% ±  6.43%

    ▅▇█▅▂     ▂ ▁▁▁                                           
  ▄▇███████▅▆██▇███▇▅▅▄▄▄▃▅▃▄▃▄▂▃▃▂▃▂▂▂▂▃▂▂▂▂▁▂▂▂▁▂▁▁▁▁▁▂▁▁▂ ▄
  4.14 ms        Histogram: frequency by time        5.92 ms <

 Memory estimate: 4.50 MiB, allocs estimate: 56697.

OSM/XML benchmark:

julia> @benchmark graph_from_file($("melbourne_1k.osm"))
BenchmarkTools.Trial: 495 samples with 1 evaluation.
 Range (min … max):   8.902 ms … 58.372 ms  ┊ GC (min … max): 0.00% … 82.90%
 Time  (median):      9.318 ms              ┊ GC (median):    0.00%
 Time  (mean ± σ):   10.095 ms ±  4.554 ms  ┊ GC (mean ± σ):  3.85% ±  7.46%

  ▆▇█▇▃▁                                                       
  ██████▇▅▄▇▅▆▇▆▁▅▁▄▄▅▁▄▆▁▁▄▁▁▁▁▄▁▅▁▁▁▄▄▄▁▄▆▆▁▄▄▁▁▄▁▁▁▁▁▁▁▁▁▅ ▇
  8.9 ms       Histogram: log(frequency) by time      19.1 ms <

 Memory estimate: 6.58 MiB, allocs estimate: 75138.