Associative Array potential performance pitfall?

[dayllenger]

On Friday, 13 March 2020 at 03:40:11 UTC, Adnan wrote:
> Where am I losing performance?

It is nonsensical to say without measuring. Humans are 
notoriously bad at predicting performance issues. Wrap all your 
hardworking code into a loop with like 100 iterations, compile 
and run:

$ perf record -g ./app file.txt && perf report

Measure dmd debug build first, then ldc2 -release -O.

Besides what H. S. Teoh said, these things help:

- disable GC before doing the job and enable it afterwards
- use .byLineCopy instead of lines(), it's faster
- use .byKeyValue when printing the results, it's lazy and 
returns both key and value

90ms after vs 215ms before