[SPARK-34668][SQL] Support casting of day-time intervals to strings #32070
Conversation
|
Kubernetes integration test starting |
|
Kubernetes integration test status failure |
|
Test build #136965 has finished for PR 32070 at commit
|
| Duration.ofMillis(1234) -> "0 0:0:1.234", | ||
| Duration.ofSeconds(-59).minus(999999, ChronoUnit.MICROS) -> "-0 0:0:59.999999", | ||
| Duration.ofMinutes(30).plusMillis(10) -> "0 0:30:0.01", | ||
| Duration.ofHours(-23).minusSeconds(59) -> "-0 23:0:59", |
There was a problem hiding this comment.
For times, I think it's more common to always have 2 digits. Have we checked with other databases?
There was a problem hiding this comment.
Have we checked with other databases?
For example, Oracle doesn't prepend zero for hours, see the doc:
INTERVAL '4 5:12:10.222' DAY TO SECOND
There was a problem hiding this comment.
that's the literal syntax, which is supposed to be more flexible. How about the cast behavior?
There was a problem hiding this comment.
added leading zeros for time parts
|
@cloud-fan @gengliangwang @yaooqinn @AngersZhuuuu Could you review this PR, please. |
| val hours = rest % HOURS_PER_DAY | ||
| val days = rest / HOURS_PER_DAY | ||
| s"INTERVAL '$sign$days $hours:$minutes:$secondStr' DAY TO SECOND" | ||
| val leadSecZero = if (seconds < 10000000) "0" else "" |
There was a problem hiding this comment.
nit: use MICROS_PER_SECONDS?
| var sign = "" | ||
| var rest = micros | ||
| if (micros < 0) { | ||
| if (micros == Long.MinValue) { |
|
Kubernetes integration test starting |
|
Kubernetes integration test status failure |
|
Kubernetes integration test starting |
|
Kubernetes integration test status failure |
|
Test build #137012 has finished for PR 32070 at commit
|
|
Test build #137020 has finished for PR 32070 at commit
|
|
Refer to this link for build results (access rights to CI server needed): |
|
thanks, merging to master! |
7 participants
What changes were proposed in this pull request?
toDayTimeIntervalString()toIntervalUtilswhich converts a day-time interval as a number of microseconds to a string in the form "INTERVAL '[sign]days hours:minutes:secondsWithFraction' DAY TO SECOND".Castexpression to support casting ofDayTimeIntervalTypetoStringType.Why are the changes needed?
To conform the ANSI SQL standard which requires to support such casting.
Does this PR introduce any user-facing change?
Should not because new day-time interval has not been released yet.
How was this patch tested?
Added new tests for casting: