Revives the `MemorySim` PR and fixes a page cache issue where we could
have a unlocked and unloaded page in the page cache after a FaultyQuery.
The page would continue in the cache and could affect other queries as
the `page_cache` is at the `Connection` level.
Depends on #2785Closes#2693
## Changes
- Refactor sql generation to always accept a `Context` trait object so
we can query the current Generation `Opts`. This change allows us to be
more granular in generating our sql statements. It also opens
opportunities for us to add even more knobs to tweak generation as
needed. I tried to make this as generic as possible as I believe this
library can be useful for fuzz testing outside the simulator.
- Introduce `Profile` struct that aggregates the different
configurations needed to execute the simulator. With this Profile struct
we can bias sql generation towards different statements and create
predefined profiles.
`WriteHeavy` Profile:
```rust
Profile {
query: QueryProfile {
gen_opts: Opts {
// TODO: in the future tweak blob size for bigger inserts
// TODO: increase number of rows as well
table: TableOpts {
large_table: LargeTableOpts {
large_table_prob: 0.4,
..Default::default()
},
..Default::default()
},
query: QueryOpts {
insert: InsertOpts {
min_rows: NonZeroU32::new(5).unwrap(),
max_rows: NonZeroU32::new(11).unwrap(),
},
..Default::default()
},
..Default::default()
},
select_weight: 30,
insert_weight: 70,
delete_weight: 0,
update_weight: 0,
..Default::default()
},
..Default::default()
};
```
As you can see we disable the `delete` and `update` weights, decrease
`select` and increase `insert` weights. This means that we disable
updates and deletes in favor of inserting more data and checking the
validity of the database with fewer select statements.
- `Profile` and `Opts` are validated with `garde` and can generate json
schemas with `schemars` so that we can have editor integration when
creating new profiles to play with.
- Added some docs in the README explaining how you can add LSP
integration for the Json config by generating a `JsonSchema` file
Closes#2852
Use the same rusqlite version in every crate and use a bundled up-to-
date sqlite version
(the impetus for this PR is still me trying to figure out why sqlite in
the insert benchmark doesn't seem to be fsyncing, even when instructed)
Closes#2507
We should be allocator-agnostic. It is pretty limiting for us to force a
user to use a particular allocator. This is specially restricting for
`no_std` in the future.
Reviewed-by: bit-aloo (@Shourya742)
Reviewed-by: Preston Thorpe <preston@turso.tech>
Closes#2481
