I am very interested in nanoseconds what exactly is ignored?
i use function logseverity() and macro

. as result the macro resolved with NS precision in e-mails

Good question. The cases you mentioned indeed use nanoseconds. The cases I was referring to are functions last(), prev(), diff(), change(), abschange() - the functions we have fixed recently in ZBX-1038.

frankly, I can hardly imagine how these functions should or can use a nanosecond. If you can, please describe in a few words, I'll be one of the first beta testers who will test it

In the current trunk, history tables have both "clock" and "ns" fields. Function last(#10,1d), for instance, would query a history table to find the 10th last value one day ago, so it has to sort by both clock and nanoseconds.

Seems that the problem has already been solved when we implemented the single point of entry for historical data (i.e., function DBget_history()).

Stylistic changes are available in development branch svn://svn.zabbix.com/branches/dev/ZBX-3113 . Please review.

<sasha> REVIEWED Excellent! Please review my small change in r22815.

<asaveljevs> CLOSED

We are going to make nanoseconds patch mandatory under this issue.

Available in development branch svn://svn.zabbix.com/branches/dev/ZBX-3113 .

Available in pre-1.9.8 in r23035.