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
Currently you can select --rows or -F not both at the same time. What I'm thinking that could be really useful, is to have both options, but we will need to support a new file format: <database>.<table>.<PART 1>.<PART 2>.sql
So, in cases where you have an integer primary key and you use -r and -F, we could see files like this if the table has gaps in it:
test_db.test_table.00000.sql
test_db.test_table.00000.00000.sql
test_db.test_table.00000.00001.sql
test_db.test_table.00000.00002.sql
test_db.test_table.00001.sql
test_db.test_table.00003.sql
test_db.test_table.00004.sql
test_db.test_table.00008.sql
Another alternative to implement it, is to keep an autoincrement value per table, but we need a mutex to avoid having 2 files with the same number and we need call that function every time we need to. I don't like this one, as it might cause fragmentation when we insert as we will be loosing the order.
What is the goal?
We are not going to create files larger than -F. NEVER. This is useful for the streaming use cases as large files could be a bottleneck when you have low network bandwidth.
This is the other side of #422, so, if you ask why are we trying to do both things and not just keep one way? well, there are multiple use cases and I think that we could support them.
The text was updated successfully, but these errors were encountered:
Currently you can select --rows or -F not both at the same time. What I'm thinking that could be really useful, is to have both options, but we will need to support a new file format:
<database>.<table>.<PART 1>.<PART 2>.sql
So, in cases where you have an integer primary key and you use -r and -F, we could see files like this if the table has gaps in it:
Another alternative to implement it, is to keep an autoincrement value per table, but we need a mutex to avoid having 2 files with the same number and we need call that function every time we need to. I don't like this one, as it might cause fragmentation when we insert as we will be loosing the order.
What is the goal?
We are not going to create files larger than -F. NEVER. This is useful for the streaming use cases as large files could be a bottleneck when you have low network bandwidth.
This is the other side of #422, so, if you ask why are we trying to do both things and not just keep one way? well, there are multiple use cases and I think that we could support them.
The text was updated successfully, but these errors were encountered: