You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The goal of this task is to allow the EE's runtime engine ignore tuples that are marked as migrated during query execution. This will allow us to delay when we actually remove a tuple from tables and update indexes. This is very similar to how the anti-caching stuff can identify when it's trying to access an evicted tuple and needs to abort the transaction.
There are only four places in the EE that access tuples directly from PersistantTables:
executors/indexscanexecutor
executors/nestloopexecutor
executors/nestloopindexexecutor
executors/seqscanexecutor
If you look at the p_execute() function for each of these classes, you will see a while loop where we iterate over some list of tuples. For IndexScan, you can see where we check TableTuple::isEvicted() for each tuple.
And then this is when I realized that @jdebrabant may not have implemented support for the other parts of the code...
The text was updated successfully, but these errors were encountered:
…d access tracking stuff into the AntiCacheEvictionManager. We now support checking for and updating access tuple accesses in NestLoopIndexScan + SeqScan. Still need to fix NestLoop #157
The goal of this task is to allow the EE's runtime engine ignore tuples that are marked as migrated during query execution. This will allow us to delay when we actually remove a tuple from tables and update indexes. This is very similar to how the anti-caching stuff can identify when it's trying to access an evicted tuple and needs to abort the transaction.
There are only four places in the EE that access tuples directly from
PersistantTables
:If you look at the
p_execute()
function for each of these classes, you will see a while loop where we iterate over some list of tuples. ForIndexScan
, you can see where we checkTableTuple::isEvicted()
for each tuple.And then this is when I realized that @jdebrabant may not have implemented support for the other parts of the code...
The text was updated successfully, but these errors were encountered: