[fix] accept tabs as tokens after : and -

[biojppm]

No description provided.

[codecov]

Codecov Report

Merging #211 (d7d0190) into master (10d68ca) will increase coverage by 0.06%.
The diff coverage is 97.84%.

@@            Coverage Diff             @@
##           master     #211      +/-   ##
==========================================
+ Coverage   97.00%   97.06%   +0.06%     
==========================================
  Files          77       77              
  Lines       16672    16664       -8     
==========================================
+ Hits        16172    16175       +3     
+ Misses        500      489      -11     

Impacted Files	Coverage Δ
src/c4/yml/detail/parser_dbg.hpp	29.62% <ø> (ø)
src/c4/yml/parse.hpp	100.00% <ø> (ø)
test/test_seq_of_map.cpp	100.00% <ø> (ø)
test/test_simple_map.cpp	98.30% <ø> (ø)
test/test_simple_seq.cpp	97.00% <ø> (ø)
test/test_suite/test_suite_parts.cpp	100.00% <ø> (ø)
src/c4/yml/parse.cpp	95.63% <97.84%> (+0.43%)	⬆️

---

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 10d68ca...d7d0190. Read the comment docs.

[biojppm]

This feature is too costly for the benefits it offers. It costs somewhere between 10-15% in throughput/cpu time because of the added branching.

Here is a benchmark for parsing a file which is not even token-heavy.

First, before the feature:

---------------------------------------------------------------------------------
Benchmark                    Time             CPU   Iterations UserCounters...
---------------------------------------------------------------------------------
ryml_inplace_reuse       12301 ns        12308 ns        56592 bytes_per_second=165.04M/s items_per_second=81.2476k/s
ryml_arena_reuse         12445 ns        12453 ns        55843 bytes_per_second=163.12M/s items_per_second=80.3021k/s
ryml_inplace             14390 ns        14394 ns        49587 bytes_per_second=141.119M/s items_per_second=69.4716k/s
ryml_arena               15190 ns        15183 ns        51004 bytes_per_second=133.792M/s items_per_second=65.8644k/s
libyaml_arena            32065 ns        32052 ns        21070 bytes_per_second=63.3758M/s items_per_second=31.1992k/s
libyaml_arena_reuse      32453 ns        32465 ns        22356 bytes_per_second=62.5702M/s items_per_second=30.8026k/s
libfyaml_arena           45159 ns        45141 ns        15089 bytes_per_second=44.9992M/s items_per_second=22.1526k/s
yamlcpp_arena           235727 ns       235590 ns         3354 bytes_per_second=8.6223M/s items_per_second=4.24467k/s

Now, with the feature:

---------------------------------------------------------------------------------
Benchmark                    Time             CPU   Iterations UserCounters...
---------------------------------------------------------------------------------
ryml_inplace_reuse       13365 ns        13365 ns        50357 bytes_per_second=151.989M/s items_per_second=74.8224k/s
ryml_arena_reuse         13598 ns        13602 ns        51875 bytes_per_second=149.338M/s items_per_second=73.5175k/s
ryml_inplace             15222 ns        15229 ns        45948 bytes_per_second=133.384M/s items_per_second=65.6637k/s
ryml_arena               14721 ns        14716 ns        46684 bytes_per_second=138.032M/s items_per_second=67.9519k/s
libyaml_arena            32124 ns        32113 ns        21813 bytes_per_second=63.2553M/s items_per_second=31.1399k/s
libyaml_arena_reuse      31572 ns        31589 ns        21761 bytes_per_second=64.3045M/s items_per_second=31.6564k/s
libfyaml_arena           45563 ns        45546 ns        15376 bytes_per_second=44.5992M/s items_per_second=21.9557k/s
yamlcpp_arena           211465 ns       211388 ns         3318 bytes_per_second=9.60946M/s items_per_second=4.73063k/s