Hi All, +1 to Szabi's comment. As it is a bugfix I wouldn't think that a new option should be introduced to switch it on and off. However, it should be emphasized in the release notes (in whichever release it will be included) that this is a behavior changing fix.
Thanks for bringing it up Anna! Cheers, Bogi On Fri, Jul 21, 2017 at 3:14 PM, Szabolcs Vasas <va...@cloudera.com> wrote: > Hi guys, > > I think that this is basically a bug fix and because of that I would not > add a new flag to switch it on/off. > It is true that it changes behavior but many other bug fixes do the same so > if we add a switch now we might want to do the same thing later and that > could lead to a large number of extra command line arguments. > > Regards, > Szabolcs > > On Thu, Jul 20, 2017 at 8:29 PM, Anna Szonyi <szo...@cloudera.com> wrote: > > > Hi @dev list, > > > > Jilani has created a great patch for SQOOP-3149 > > <https://issues.apache.org/jira/browse/SQOOP-3149>: Sqoop incremental > > import - NULL column updates are not pulled into HBase table. > > > > This is a patch that (in my opinion) corrects previously incorrect > behavior > > in sqoop, however it can also be seen as just changing behavior from the > > users perspective ("we did not have nulls there before, now we have"). > > > > The patch is ready to be committed, however I wanted to ask the dev list > on > > thoughts and opinions on whether we should merge the change as-is or > should > > we include a flag to turn this feature on and off? And if we need to > > include a flag, should it be in a separate commit or should we amend the > > original patch? > > > > Please find the review request below: > > https://reviews.apache.org/r/57499/ > > > > Please chime in with your thoughts - if there are no objections on this > > thread, I plan to merge the change as-is in a week or so and pick up this > > conversation again after. > > > > Thanks, > > Anna > > > > > > -- > Szabolcs Vasas > Software Engineer > <http://www.cloudera.com> >
