I agree with Pramod. Current order + org.apache.apex.* with delta change. IMO, com.datatorrent.* should be after org.apache.apex. For example, if some company xyz is developing their applications on top of apex; they would put com.xyz.* after org.apache.apex.
I understand that, com.datatorrent.* imports would be eventually removed once we have all the classes migrated to org.apache.apex.*. But, still having a explicit mention of it, in the import order will give some guidelines for the other projects dependent on apex. ~ Yogi On 7 April 2016 at 04:09, Pramod Immaneni <[email protected]> wrote: > Like original suggestion from Chandni, current order + org.apache.apex.* at > the end. > > On Wed, Apr 6, 2016 at 3:36 PM, Vlad Rozov <[email protected]> > wrote: > > > I would prefer not to change the current import order and once > > org.apache.apex is introduced to keep it together with the org.apache > > group. It will be extremely inconvenient to work on different branches > with > > different import orders. > > > > Thank you, > > > > Vlad > > > > > > On 4/6/16 13:18, Munagala Ramanath wrote: > > > >> I prefer imports to be organized in groups with the groups themselves > >> ordered by decreasing order of "scope" where I > >> consider java to have the widest scope and com.datatorrent the narrowest > >> (roughly the way we have it now) with a > >> single blank line separating groups. When "scope" is not clear we can > >> default to lexicographic. I think this helps > >> readability. > >> > >> No major project does completely lexicographic ordering as far as I > know. > >> As a couple of data points : both Hadoop and > >> Flink are doing roughly what I propose though they are not always > >> consistent. Here are some examples: > >> > >> > >> > https://github.com/apache/flink/blob/master/flink-core/src/main/java/org/apache/flink/core/fs/FSDataOutputStream.java > >> > >> -- java imports come after org.apache > >> > >> > >> > https://github.com/apache/flink/blob/master/flink-core/src/main/java/org/apache/flink/core/fs/FileSystem.java > >> > >> -- java imports come before org.apache > >> > >> > >> > https://github.com/apache/flink/blob/master/flink-core/src/main/java/org/apache/flink/core/fs/local/LocalFileSystem.java > >> > >> -- org imports not alphabetic > >> > >> > >> > hadoop-2.7.2-src/hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/service: > >> > >> AbstractService.java > >> > >> -- import groups not lexicographic: java comes before org but com comes > >> after. > >> > >> Service.java > >> > >> -- java comes after org > >> > >> Ram > >> > >> On Wed, Apr 6, 2016 at 12:24 PM, David Yan <[email protected]> > wrote: > >> > >> I think the import order should be alphabetical or lexicographical, and > no > >>> blank lines are needed. > >>> > >>> The reason is this: > >>> > >>> 1) It's simple and future proof, and there will be no need to have this > >>> discussion again in the future about what order it should be and where > >>> the > >>> blank lines should be placed > >>> 2) It's easy to follow, and there is no rule to remember other than > it's > >>> alphabetical > >>> 3) It's easy to find where a particular import lines are because it's > >>> alphabetical > >>> > >>> Whatever we come up with, we should make the change ASAP and fix all > the > >>> style violations once and for all. > >>> > >>> David > >>> > >>> > >>> On Tue, Apr 5, 2016 at 5:12 PM, Chandni Singh <[email protected] > > > >>> wrote: > >>> > >>> Hi, > >>>> > >>>> We need to include org.apache.apex in the import order that we have > >>>> right > >>>> now: > >>>> > >>>> *Current order:* > >>>> java > >>>> javax > >>>> org > >>>> org.apache > >>>> io > >>>> com > >>>> com.datatorrent > >>>> * > >>>> all static imports > >>>> > >>>> *Proposed order:* > >>>> java > >>>> javax > >>>> org > >>>> org.apache > >>>> io > >>>> com > >>>> com.datatorrent > >>>> org.apache.apex > >>>> * > >>>> all static imports > >>>> > >>>> Thanks, > >>>> Chandni > >>>> > >>>> > > >
