[
https://issues.apache.org/jira/browse/NIFI-1522?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15157597#comment-15157597
]
Joe Skora commented on NIFI-1522:
---------------------------------
I think the needs of the user should outweigh developer interests on a
enterprise system like NiFi. I was impressed with the input flexibility when I
first started using NIFi, from past work I knew it wasn't trivial to satisfy
the disparate needs of flexibility and precision cleanly. I could say "1 GB",
"1024 MB", or "1048576 B" and internally they would be treated equally. I'm
all for making the system easier for the user with such flexibility, but it
can't sacrifice precision.
And after decades coding I don't see millisecond as a standard, if you run "$
man 5 sleep" on a Linux system today it will tell you that the standard C
library sleep() function takes an unsigned integer number of seconds. And
though "Epoch" seconds (since midnight, Jan 1, 1970) is commonly used for
timestamps, the Unix Time specification is in seconds, Java and JavaScript use
integer milliseconds, and other contexts use other measures including floating
point seconds, integer seconds, or nanoseconds.
Just my 2 cents. I think consensus should carry the day, but my concerns about
implied units will remain.
> Period and Unit time expressions should default to milliseconds
> ---------------------------------------------------------------
>
> Key: NIFI-1522
> URL: https://issues.apache.org/jira/browse/NIFI-1522
> Project: Apache NiFi
> Issue Type: Improvement
> Reporter: Joseph Witt
> Assignee: Oleg Zhurakousky
> Fix For: 0.6.0
>
>
> Discussion on Github about our current support for period and unit
> expressions for time
> https://github.com/apache/nifi/pull/210#discussion_r53167209
> We should support that if a unit of time is not specified that the period is
> assumed to be in milliseconds.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
