jonkeane commented on PR #12589: URL: https://github.com/apache/arrow/pull/12589#issuecomment-1118974609
> I edited the top comment. Not sure what to do about the NEWS. I'm a bit hesitant to edit NEWS since that looks ready for release. > > How in sync does the release of the R package have to be with the libarrow release candidate? Aaaah good catch — what we are adding here should not be under 8.0.0 at all, we should have that be up above in the dev section — you might need to add one to it with this PR. This PR will not get folded in to the 8.0.0 release, so we don't need to worry about adding to the top of the release notes at this point. We will release both libarrow and the R package from the release branch, so things merged in to master now will not be released for either. > We have the Jira + this comment. Not quite sure how else to call attention to it. One of the main differences between the Arrow binding an lubridate's parse_date_time() is that there is no (and it looks like there won't be, at least for a while) any guessing / inference in Arrow. In lubridate, guess_formats() takes the orders argument and builds the most likely formats. > > To me, parsing a string with no separators is a step into inference territory since we need to make a guess in order to break it up into several components. Maybe some separator-less formats are "obvious", but they still sort of rely on inference logic. We should note that not all functionality in `parse_date_time()` from lubridate is implemented. Like you mention here, we don't support guessing or inference (though some of what we can do with coalesce here might be confused as a type of inference...) Looking at the lubridate description: > *orders* a character vector of date-time formats. Each order string is a series of formatting characters as listed in [base::strptime()](https://rdrr.io/r/base/strptime.html) but might not include the "%" prefix. For example, "ymd" will match all the possible dates in year, month, day order. Formatting orders might include arbitrary separators. These are discarded. See details for the implemented formats. If multiple order strings are supplied, they are applied in turn for parse_date_time2() and fast_strptime(). For parse_date_time() the order of applied formats is determined by select_formats parameter. Which parts of this do we support with this PR? Which do we not? Let's craft some details for NEWS that is truthful about what is implemented and mention that other parts are forthcoming | not yet implemented. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
