MVCC is like the annoying younger cousin (I know because I was him) that
needs to be treated differently. MVCC requires us to use root_pages that
might not be allocated yet, and the plan is to use negative root_pages
for that case. Therefore, we need i64 in order to fit this change.
Improve serialization for DBSP views.
The serialization code was written organically, without much forward
thinking about stability as we evolved the table and operator format.
Now that this is done, we are at at point where we can actually make it
suck less and take a considerable step towards making this production
ready.
We also add a simple version check (in the table name, because that is
much easier than reading contents in parse_schema_row) to prevent views
to be used if we had to do anything to evolve the format of the circuit
(including the operators)
Closes#3351
as we make changes to the way materialized views are generated (think
adding new operators, changing the id of existing operators, etc), we
will need to persist the topology of the circuit itself. This is a
change that I believe to be premature. For now, it is enough to reserve
the first operator id for it, and add a version number to the table
name. We can just detect that something changed, and ask the user to
drop the view. We can get away with it due to the fact that the views
are experimental.
We have used i64 before because that is the size of an integer in
SQLite. However, I believe that for large enough databases, the chances
of collision here are just too high. The effect of a collision is the
database silently returning incorrect data in the materialized view.
So now that everything else is working, we should move to i128.
Our code for view needs to extract the list of columns used in the view.
We currently extract only from "the base table", but once we have joins,
we need a more complex structure, that keeps the mapping of
(tables, columns).
This actually affects both views and materialized views: for views, the
queries with joins work just fine, because views are just aliases for
a query. But the list of columns returned by pragma table_info on the
view is incorrect. We add a test to make sure it is fixed.
For materialized views, we add extensive tests to make sure that the
columns are extracted correctly.
And also change the schema of the main table. I have come to see the
current key-value schema as inadequate for non-aggregate operators.
Calculating Min/Max, for example, doesn't feat in this schema because
we have to be able to track existing values and index them.
Another alternative is to keep one table per operator type, but this
quickly leads to an explosion of tables.
indexes with the naming scheme "sqlite_autoindex_<tblname>_<number>"
are automatically created when a table is created with UNIQUE or
PRIMARY KEY definitions.
these indexes must map to the table definition SQL in definition order,
i.e. sqlite_autoindex_foo_1 must be the first instance of UNIQUE or
PRIMARY KEY and so on.
this commit fixes our autoindex creation / parsing so that this invariant
is upheld.
This fairly long commit implements persistence for materialized view.
It is hard to split because of all the interdependencies between components,
so it is a one big thing. This commit message will at least try to go into
details about the basic architecture.
Materialized Views as tables
============================
Materialized views are now a normal table - whereas before they were a virtual
table. By making a materialized view a table, we can reuse all the
infrastructure for dealing with tables (cursors, etc).
One of the advantages of doing this is that we can create indexes on view
columns. Later, we should also be able to write those views to separate files
with ATTACH write.
Materialized Views as Zsets
===========================
The contents of the table are a ZSet: rowid, values, weight. Readers will
notice that because of this, the usage of the ZSet data structure dwindles
throughout the codebase. The main difference between our materialized ZSet and
the standard DBSP ZSet, is that obviously ours is backed by a BTree, not a Hash
(since SQLite tables are BTrees)
Aggregator State
================
In DBSP, the aggregator nodes also have state. To store that state, there is a
second table. The table holds all aggregators in the view, and there is one
table per view. That is __turso_internal_dbsp_state_{view_name}. The format of
that table is similar to a ZSet: rowid, serialized_values, weight. We serialize
the values because there will be many aggregators in the table. We can't rely
on a particular format for the values.
The Materialized View Cursor
============================
Reading from a Materialized View essentially means reading from the persisted
ZSet, and enhancing that with data that exists within the transaction.
Transaction data is ephemeral, so we do not materialize this anywhere: we have
a carefully crafted implementation of seek that takes care of merging weights
and stitching the two sets together.
My goal with this patch is to be able to implement the ProjectOperator
for DBSP circuits using VDBE for expression evaluation.
*not* doing so is dangerous for the following reason: we will end up
with different, subtle, and incompatible behavior between SQLite
expressions if they are used in views versus outside of views.
In fact, even in our prototype had them: our projection tests, which
used to pass, were actually wrong =) (sqlite would return something
different if those functions were executed outside the view context)
For optimization reasons, we single out trivial expressions: they don't
have go through VDBE. Trivial expressions are expressions that only
involve Columns, Literals, and simple operators on elements of the same
type. Even type coercion takes this out of the realm of trivial.
Everything that is not trivial, is then translated with translate_expr -
in the same way SQLite will, and then compiled with VDBE.
We can, over time, make this process much better. There are essentially
infinite opportunities for optimization here. But for now, the main
warts are:
* VDBE execution needs a connection
* There is no good way in VDBE to pass parameters to a program.
* It is almost trivial to pollute the original connection. For example,
we need to issue HALT for the program to stop, but seeing that halt
will usually cause the program to try and halt the original program.
Subprograms, like the ones we use in triggers are a possible solution,
but they are much more expensive to execute, especially given that our
execution would essentially have to have a program with no other role
than to wrap the subprogram.
Therefore, what I am doing is:
* There is an in-memory database inside the projection operator (an
obvious optimization is to share it with *all* projection operators).
* We obtain a connection to that database when the operator is created
* We use that connection to execute our VDBE, which offers a clean, safe
and isolated way to execute the expression.
* We feed the values to the program manually by editing the registers
directly.
SQLite does not store the rowid alias column in the record at all
when it is a rowid alias, because the rowid is always stored anyway
in the record header.
A lot of the structures we have - like the ones under Schema, are
specific for materialized views. In preparation to adding normal views,
rename them, so things are less confusing.
This is just the bare minimum that I needed to convince myself that this
approach will work. The only views that we support are slices of the
main table: no aggregations, no joins, no projections.
drop view is implemented.
view population is implemented.
deletes, inserts and updates are implemented.
much like indexes before, a flag must be passed to enable views.
