leventov commented on issue #7306: Reconcile terminology and method naming to 'used/unused segments'; Rename MetadataSegmentManager to MetadataSegments URL: https://github.com/apache/incubator-druid/pull/7306#issuecomment-477783437 @gianm I didn't question that "published in `sys.segments`" and "`used` in metadata store" is the same thing, logically. I did question whether making this same logical thing called the same would actually make positive, rather than negative impact on the codebase. However, now I also think it's higher probability that it will have positive impact on the codebase. So I will support the rename. It doesn't seem to me that this PR is conflicting with the rename. I think it's supportive to the rename because in this PR I did some nasty work at finding and sweeping old names and variants of naming that were looking completely unrelated. Now for the rename, somebody needs to make just two full-text searches in the codebase on Github (I did many more when preparing this PR): - "[used segment](https://github.com/apache/incubator-druid/search?q=unused+segment&unscoped_q=used+segment)" - "[unused segment](https://github.com/apache/incubator-druid/search?q=unused+segment&unscoped_q=unused+segment)" Then visit all result files that look relevant manually and rename "used/unused segments" to "publish/published/unpublished". (Note that this search needs to be performed after this PR is merged, because Github indexes only master codebase for full-text search.) For this reason, I tag this PR `Development Blocker` for #7233. > Maybe this is how: > - Tasks "publish" segments when they insert records into the metadata store with used = true (upon creation of the segment) > - Later on if we drop a segment, we do that by "marking it unused" or, potentially, "deactivating" it. > - If we re-enable a segment, we do that by "marking it used", or, potentially, "activating" it. Oh no, I think we must go all-in with publish/unpublish/published/unpublished terminology. So we should call these actions "mark (as) published", however awkward it sounds, and "unpublish" (I think "unpublish all segments in a data source" sounds fine), or, alternatively, "mark (as) unpublished".
---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
