`db_size` is `>0` in case of last frame written of a transaction. This
is necessary as we need to know -- while recovering wal contents -- that
we have read a transaction fully instead of treating every frame as its
own transaction.
Closes#1866
Closes#1888 . This PR fixes UPDATE translation by not emitting an
ephemeral plan when we are doing a `RowIdEq` search. Also, we should
delete the previous rowid when the rowid is in the set clause.
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#1891
Bundle UUID and time extensions and drop the dependency to other
extensions so that we don't have to publish every crate.
Reviewed-by: Preston Thorpe (@PThorpe92)
Closes#1886
This adds a `turso_assert` macro that is Antithesis aware when
`antithesis` feature flag is enabled. I did not yet convert any call-
sites to use it.
Closes#1880
This adds a `turso_assert` macro that is Antithesis aware when
`antithesis` feature flag is enabled. I did not yet convert any
call-sites to use it.
Co-authored-by: Nikita Sivukhin <sivukhin@turso.tech>
Stress runs have bunch of errors like this caused by reconnect in the
middle of a transaction:
```
Error executing query: Transaction error: cannot rollback - no transaction is active
```
This is fine, but let's add some logging that it's obvious why this
happens.
Closes#1878
Stress runs have bunch of errors like this caused by reconnect in the
middle of a transaction:
```
Error executing query: Transaction error: cannot rollback - no transaction is active
```
This is fine, but let's add some logging that it's obvious why this
happens.
I am not changing any package names - I'd rather Pekka do that so he can
verifies it works, or things that look like references to external
entities mentioning limbo.
All the rest is changed.
Closes#1868
I am not changing any package names - I'd rather Pekka do that so he
can verifies it works, or things that look like references to external
entities mentioning limbo.
All the rest is changed.
`db_size` is `>0` in case of last frame written of a transaction. This
is necessary as we need to know -- while recovering wal contents -- that
we have read a transaction fully instead of treating every frame as its
own transaction.
