MergeTree tables do not include primary key definitions, which limits TLP’s ability to uniquely identify rows across partitions. if we add primary key support it would:
enable unique row identification
allow TLP to more reliably verify that the union of partitioned results matches the original query result
complex queries involving join or aggregations will benefit from having a clear unique key.
we can make After generating column names, pick one or more columns at random (or by some conditions we determine) that will act as our primary key.
I think we can generalize the solution in all DBMSs.
could you tell me your opinion dr @mrigger ?
MergeTree tables do not include primary key definitions, which limits TLP’s ability to uniquely identify rows across partitions. if we add primary key support it would:
enable unique row identification
allow TLP to more reliably verify that the union of partitioned results matches the original query result
complex queries involving join or aggregations will benefit from having a clear unique key.
we can make After generating column names, pick one or more columns at random (or by some conditions we determine) that will act as our primary key.
I think we can generalize the solution in all DBMSs.
could you tell me your opinion dr @mrigger ?