Both of google style or own special way are fine to me. :-) Another issue is how to make people feel the special import ordering is important? I prefer adding the order check to precommit run.
On 2017-10-02 03:37, Josh Elser <els...@apache.org> wrote: > Playing devil's advocate: > > Do we want to maintain our own "special" way of doing imports instead of > relying on something such as the Google Java style guide? [1] > > +1 to the idea of cleaning things up, but just curious if people feel > like our special import ordering is important (and not just vestigial). > Personally, I don't have much concern about import order than having > consistency -- I like the idea of letting someone hash out what is > recommended :) > > [1] > https://google.github.io/styleguide/javaguide.html#s3.3-import-statements > > On 10/1/17 11:27 AM, Chia-Ping Tsai wrote: > > hi folks, > > > > I noticed the code conflict occurs on the imports frequently. To resolve > > the conflict is a complete waste of time, so i feel it is time to update > > our imports layout and hold our attraction on it. > > > > The import layout is shown below. (see hbase_eclipse_formatter.xml) > > javax.* > > blank line > > java.* > > blank line > > import all other imports > > blank line > > import static all other imports > > > > Q1: > > As i see it, two updates should be considered. > > 1) Should we move the "import static" to the top? (It seems most of files > > have its static imports on the top) > > 2) Should we move the shaded class into new blocks? > > > > The new layout looks like this. > > import static all other imports > > blank line > > javax.* > > blank line > > java.* > > blank line > > org.* > > blank line > > org.apache.hadoop.hbase.shaded.* > > blank line > > import all other imports > > > > Q2: > > Should we check the import layout before committing? Perhaps we can address > > this in the HBASE-18438. The issue try to add the check of unused imports > > > > Any suggestions? Thanks. > > -- > > Chia-Ping > > >
