[jira] [Commented] (CASSANDRA-9699) Make choice of sstable reader types explicit
[ https://issues.apache.org/jira/browse/CASSANDRA-9699?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14644093#comment-14644093 ] Marcus Eriksson commented on CASSANDRA-9699: +1 with a few tiny nits pushed here: https://github.com/krummas/cassandra/commit/d3c326adfee6cbbcfa3457fd365becab10401749 Make choice of sstable reader types explicit Key: CASSANDRA-9699 URL: https://issues.apache.org/jira/browse/CASSANDRA-9699 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Benedict Assignee: Benedict Fix For: 3.0.0 rc1 CASSANDRA-9580 made it apparent that the efforts to isolate unsafe users of early opened results wasn't quite fully successful. This patch attempts to impose some minor central refactors that force the (internal) user to specify if they want the live sstable set, or the canonical sstable set (or just the noncompacting set, since this is the intersection of the two). It also eliminates a lot of ugliness and some duplication around filtration, and obsoletes the CANONICAL_SSTABLES and UNREPAIRED_SSTABLES functions in ColumnFamilyStore, which were very ugly. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9699) Make choice of sstable reader types explicit
[ https://issues.apache.org/jira/browse/CASSANDRA-9699?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14642719#comment-14642719 ] Benedict commented on CASSANDRA-9699: - I've rebased against latest trunk [here|https://github.com/belliottsmith/cassandra/tree/9699] and am waiting on clean CI results. Make choice of sstable reader types explicit Key: CASSANDRA-9699 URL: https://issues.apache.org/jira/browse/CASSANDRA-9699 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Benedict Assignee: Benedict Fix For: 3.0.0 rc1 CASSANDRA-9580 made it apparent that the efforts to isolate unsafe users of early opened results wasn't quite fully successful. This patch attempts to impose some minor central refactors that force the (internal) user to specify if they want the live sstable set, or the canonical sstable set (or just the noncompacting set, since this is the intersection of the two). It also eliminates a lot of ugliness and some duplication around filtration, and obsoletes the CANONICAL_SSTABLES and UNREPAIRED_SSTABLES functions in ColumnFamilyStore, which were very ugly. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9699) Make choice of sstable reader types explicit
[ https://issues.apache.org/jira/browse/CASSANDRA-9699?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14642891#comment-14642891 ] Benedict commented on CASSANDRA-9699: - CI results look roughly equivalent Make choice of sstable reader types explicit Key: CASSANDRA-9699 URL: https://issues.apache.org/jira/browse/CASSANDRA-9699 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Benedict Assignee: Benedict Fix For: 3.0.0 rc1 CASSANDRA-9580 made it apparent that the efforts to isolate unsafe users of early opened results wasn't quite fully successful. This patch attempts to impose some minor central refactors that force the (internal) user to specify if they want the live sstable set, or the canonical sstable set (or just the noncompacting set, since this is the intersection of the two). It also eliminates a lot of ugliness and some duplication around filtration, and obsoletes the CANONICAL_SSTABLES and UNREPAIRED_SSTABLES functions in ColumnFamilyStore, which were very ugly. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9699) Make choice of sstable reader types explicit
[ https://issues.apache.org/jira/browse/CASSANDRA-9699?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14614787#comment-14614787 ] Benedict commented on CASSANDRA-9699: - Will do, as soon as CASSANDRA-9656 is in (as it also touches this) Make choice of sstable reader types explicit Key: CASSANDRA-9699 URL: https://issues.apache.org/jira/browse/CASSANDRA-9699 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Benedict Assignee: Benedict Fix For: 3.0.0 rc1 CASSANDRA-9580 made it apparent that the efforts to isolate unsafe users of early opened results wasn't quite fully successful. This patch attempts to impose some minor central refactors that force the (internal) user to specify if they want the live sstable set, or the canonical sstable set (or just the noncompacting set, since this is the intersection of the two). It also eliminates a lot of ugliness and some duplication around filtration, and obsoletes the CANONICAL_SSTABLES and UNREPAIRED_SSTABLES functions in ColumnFamilyStore, which were very ugly. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9699) Make choice of sstable reader types explicit
[ https://issues.apache.org/jira/browse/CASSANDRA-9699?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14614580#comment-14614580 ] Marcus Eriksson commented on CASSANDRA-9699: Could you rebase? Make choice of sstable reader types explicit Key: CASSANDRA-9699 URL: https://issues.apache.org/jira/browse/CASSANDRA-9699 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Benedict Assignee: Benedict Fix For: 3.0.0 rc1 CASSANDRA-9580 made it apparent that the efforts to isolate unsafe users of early opened results wasn't quite fully successful. This patch attempts to impose some minor central refactors that force the (internal) user to specify if they want the live sstable set, or the canonical sstable set (or just the noncompacting set, since this is the intersection of the two). It also eliminates a lot of ugliness and some duplication around filtration, and obsoletes the CANONICAL_SSTABLES and UNREPAIRED_SSTABLES functions in ColumnFamilyStore, which were very ugly. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9699) Make choice of sstable reader types explicit
[ https://issues.apache.org/jira/browse/CASSANDRA-9699?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14610508#comment-14610508 ] Benedict commented on CASSANDRA-9699: - Patch available [here|https://github.com/belliottsmith/cassandra/tree/explicit-sstable-selection], awaiting CI Make choice of sstable reader types explicit Key: CASSANDRA-9699 URL: https://issues.apache.org/jira/browse/CASSANDRA-9699 Project: Cassandra Issue Type: Improvement Components: Core Reporter: Benedict Assignee: Benedict Fix For: 3.0.0 rc1 CASSANDRA-9580 made it apparent that the efforts to isolate unsafe users of early opened results wasn't quite fully successful. This patch attempts to impose some minor central refactors that force the (internal) user to specify if they want the live sstable set, or the canonical sstable set (or just the noncompacting set, since this is the intersection of the two). It also eliminates a lot of ugliness and some duplication around filtration, and obsoletes the CANONICAL_SSTABLES and UNREPAIRED_SSTABLES functions in ColumnFamilyStore, which were very ugly. -- This message was sent by Atlassian JIRA (v6.3.4#6332)