@cblake, sorry for the late follow up. Each object is described by a .TXT file,
containing only grid pixels (one line is a grid pixel) having a numeric value
(float). No, the grid is not the same across the .TXT files, and a pixel may
contain more than one object values. The subset I used to run the
mini-benchmark produces a CSV of 1.8 Millions of lines (so 1.8 Million pixels,
this is the table size), and median is around 3 objects per-pixel (mean is
4.3), so I would say it's more giant hash table, short lists. You're right,
your code was missing the '\n' (I didn't check the output... it was a single
very looooooong line), I implemented the two small fixes and the output now is
right. I confirm that the cligen powered variant is about 18-20% faster than
the other one... also I would say time is more "predictable" (the other one has
a larger variance... though my Windows laptop is defintely not a good / stable
reference). @treeform I did not engineered the data myself that way, however
the .TXT files are a raw format not meant to be consumed directly (a real DB is
loaded by someone else). This was my one-time (or let's say occasional)
processing with limited ambitions, it's not normal runtime operation. And it
was another chance to try to learn something more on Nim (at my "level") from
the smart guys. By the way... I tried to install cligen a while ago... and
failed... happy to have it working now. Thank you all.
