Improve the build times of dyaml.node

[rikkimax]

In an attempt to improve dub's build times.

[Herringway]

What impact does this have on runtime?

[codecov]

Codecov Report

Merging #314 (efda074) into master (2c915b3) will increase coverage by 0.12%.
The diff coverage is 100.00%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #314      +/-   ##
==========================================
+ Coverage   95.20%   95.33%   +0.12%     
==========================================
  Files          29       29              
  Lines        4569     4604      +35     
==========================================
+ Hits         4350     4389      +39     
+ Misses        219      215       -4     

Impacted Files	Coverage Δ
source/dyaml/node.d	92.41% <100.00%> (+0.55%)	⬆️
source/dyaml/resolver.d	97.18% <100.00%> (+1.10%)	⬆️

---

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 2c915b3...efda074. Read the comment docs.

[rikkimax]

What impact does this have on runtime?

It's running in a module constructor, so it's a one-time additional cost. The benefit is in build times (at least with dub it helps speeds things up).

[Herringway]

What impact does this have on runtime?

It's running in a module constructor, so it's a one-time additional cost. The benefit is in build times (at least with dub it helps speeds things up).

Once-per-execution is more than once-per-compilation. This will increase the time it takes dub to start compiling. Have you factored that into your comparison?

[rikkimax]

I haven't looked at the speed at which dub itself runs. I've only been looking at ways to get the 24s build time down (got it to 22s).

Due to std.regex pulling in the Unicode tables which get built upon first use, this stuff isn't cheap and certainly can be done cheaper at runtime than compile time.

[Herringway]

Please look at that. Runtime costs are much more important than compilation costs.