We've already talked about this with Martin and other contributors, but leaving my 2c here for the wider discussion.
I'm leaning towards the second option, having a separate EL function, because parsing the RFC-compliant dates does not directly map to strptime-like format strings used in toDate. In python, they also have separate functions for parsing like strptime [1] (like the EL toDate) and parsing RFC 3339 or ISO 8601 dates [2]. Most other MiNiFi C++ devs are leaning towards the first option, extending toDate with an overload that doesn't take a format string. Ultimately I think we should follow whatever interface the NiFi devs prefer, so if the NiFi Expression Language is ever extended with a similar RFC/ISO date parser facility in the future, the MiNiFi C++ Expression Language would match that. This way users have an easier time converting their flows between the two implementations. Thanks, Marton [1] https://docs.python.org/3/library/datetime.html#datetime.datetime.strptime [2] https://docs.python.org/3/library/datetime.html#datetime.datetime.fromisoformat and https://stackoverflow.com/a/15228038 On Thu, 2 Mar 2023 at 13:09, Martin Zink <martinz...@apache.org> wrote: > > When dealing with timestamps coming from third parties, sometimes we cannot > rely on exact timestamp formats, so it is difficult to parse them using > toDate > <https://nifi.apache.org/docs/nifi-docs/html/expression-language-guide.html#todate>. > (e.g. Syslog messages allow a subset of RFC3339 for timestamps). > To circumvent this I started working on an expression language function > that could parse all RFC3339 compliant timestamps with expression > language in MiNiFi C++, but I feel this functionality would also be a great > addition to NiFi. > > It would be great if the expression language would remain consistent > between MiNiFi C++ and NiFi, and I'm not quite sure where this > functionality should be accessible from. > > - We could either change the current toDate > > <https://nifi.apache.org/docs/nifi-docs/html/expression-language-guide.html#todate> > to > accept zero arguments and calling it without argument would try to parse > the string as RFC3339 compliant. > - or we could introduce a new expression language function e.g. > parseRfc3339 > > > What's the community's stance on this? > > Thanks, > Martin
