Closed jteresco closed 6 months ago
An unlikely series of very specific events got us here.
sprintf
up till now, I was only able to control the number of digits after the decimal point, not the overall number of sig figs.unordered_map
s. These can get stored & thus summed in non-deterministic order, meaning cumulative rounding errors can lead to slightly different data between runs.routes
table, it "rounds" alb.a002
's mileage to 17 decimal places -- even though the new file only had 16 sig figs to begin with. Meanwhile, the old file had 17 decimal places.floatfield
flag, without an understanding of what it actually did. Bad idea. This is actually the same as setting scientific | fixed
. This is where the hex strings came from.double
, uses sprintf
to "round" to 17 decimal places, compares it to the string from the old file, and sees no difference. It reports that everything's fine when in reality it's anything but.
https://forum.travelmapping.net/index.php?topic=6238.0
The floating point format introduced in #623 didn't play nice with the version of MySQL on noreaster. All mileages were being read in as 0.