Add short writes in the faulty_libc
As @PThorpe92 stated in #3209, this should be implemented here instead
of the memory io in the simulator. Running this in the stress test I
caught a logic bug in the try_pwritev_raw I will create a pr for that
small fix. I will close#3209 in favor of this pr.
Closes#3569
Fixes incorrect buffer length calculation in try_pwritev_raw when
handling the first slice of the I/O vector.
Why:
With the faulty libc tests, a short write was simulated and the code
incorrectly computed the length for the first buffer, leading to more
bytes being written than expected, leading to a panic inside the
completion.
How:
Adjusted the slice calculation to ensure only the intended range of
bytes is written.
Backport: 0.2
Reviewed-by: Preston Thorpe <preston@turso.tech>
Closes#3570
Depends on #3502
sqlite has a field in index header called `iChange` that is incremented
after every commit. This commit simply adds support for it.
```c
struct WalIndexHdr {
u32 iVersion; /* Wal-index version */
u32 unused; /* Unused (padding) field */
u32 iChange; /* Counter incremented each transaction */
```
Closes#3559
Fixes a page cache staleness issue where connections could incorrectly
believe the database hasn't changed after checkpointing. This can happen
when writes following a checkpoint resulted in the same `max_frame
value`, causing connections to miss updates since they only checked
`max_frame` to detect changes.
Closes#3502
Fixes the following problems with COLLATE:
- Fix: incorrectly used e.g. `x COLLATE NOCASE = 'fOo'` as index
constraint on an index whose column was not case-insensitively collated
- Fix: various ephemeral indexes (in GROUP BY, ORDER BY, DISTINCT) and
subqueries did not retain proper collation information of columns
- Fix: collation of a given expression was not determined properly
according to SQLite's rules
Adds TCL tests and fuzz test
Closes#3476Closes#1524Closes#3305
Reviewed-by: Preston Thorpe <preston@turso.tech>
Closes#3538
`IOResult` implies we have a state machine that needs to be polled to
`Completion`, which is not the case here. We are just emitting the IO
operation in this case. This led us to never reaching the
`IOResult::Done` branch that actually fsynced the logical log in
`Checkpoint`.
I also sprinkled some
```rust
if c.is_completed() {
Ok(TransitionResult::Continue)
} else {
Ok(TransitionResult::Io(IOCompletions::Single(c)))
}
```
just to be more efficient with sync IO, but it is not strictly necessary
here.
Reviewed-by: Jussi Saurio <jussi.saurio@gmail.com>
Closes#3549
