Prepend leading zeros

[gleeper]

rfc3339 with nanoseconds wasn't prepending leading zeros so if an object had less than 100,000,000 nanoseconds, the nanoseconds string field would be incorrectly set.

[crwilcox]

@gleeper I cannot find specifics about this in the RFM specification. I took time to compare it to the implementation in Go, time.Parse(time.RFC3339Nano, "2006-01-02T15:04:05.7010Z") which will format as 2006-01-02T15:04:05.701Z

This seems to match our existing behavior.

[gleeper]

I made a fix for prepending zeros, not appending zeros. DatetimeWithNanoseconds currently changes 2006-01-02T15:04:05.00123Z" to 2006-01-02T15:04:05.123Z". It's changing the nanoseconds field by removing leading zeros (and therefore multiplying the nanoseconds value by 10 or 100 or 1000 etc.), not trailing zeros

[crwilcox]

Sorry for the confusion @gleeper and maybe I could have been clearer. From what I could tell our behavior properly matched how things were happening in other languages.

I think I understand now the issue. We have have handling on the parsing side to shift incoming values. So when you provide .7 it becomes 700000000 nanoseconds. While this is done also with leading 0s so .007 becomes 7000000 we would return that as the string .7 and not .007 as expected.

One of your tests (the one removing trailing but adding leading) threw me a bit and sort of made me wonder what you were trying to do :)

[crwilcox]

@gleeper thanks for the contribution to the project. Your change has been merged and will ship in the next release.

[gleeper]

@crwilcox Happy to help! Do you have an estimate as to when the next release will be?