We also at this point switch from using `Vec<Variable>` to
`BTreeSet<Variable>`. This allows us to guarantee no duplicates later;
we'll reject duplicates at parse time.
This adds an `:order` keyword to `:find`.
If present, the results of the query will be an ordered set, rather than
an unordered set; rows will appear in an ordered defined by each
`:order` entry.
Each can be one of three things:
- A var, `?x`, meaning "order by ?x ascending".
- A pair, `(asc ?x)`, meaning "order by ?x ascending".
- A pair, `(desc ?x)`, meaning "order by ?x descending".
Values will be ordered in this sequence for asc, and in reverse for desc:
1. Entity IDs, in ascending numerical order.
2. Booleans, false then true.
3. Timestamps, in ascending numerical order.
4. Longs and doubles, intermixed, in ascending numerical order.
5. Strings, in ascending lexicographic order.
6. Keywords, in ascending lexicographic order, considering the entire
ns/name pair as a single string separated by '/'.
Subcommits:
Pre: make bound_value public.
Pre: generalize ErrorKind::UnboundVariable for use in order.
Part 1: parse (direction, var) pairs.
Part 2: parse :order clause into FindQuery.
Part 3: include order variables in algebrized query.
We add order variables to :with, so we can reuse its type tag projection
logic, and so that we can phrase ordering in terms of variables rather
than datoms columns.
Part 4: produce SQL for order clauses.
* Pre: refactor projector code.
* Part 1: maintain 'with' variables in AlgebrizedQuery.
* Part 2: include necessary 'with' variables in SQL projection list.
The test produces projection elements for `:with`, even though there are
no aggregates in the query. This test will need to be adjusted when we
optimize this away!
This commit turns complex `or` -- `or`s in which not all variables are
unified, or in which not all arms are the same shape -- into a
computed table.
We do this by building a template CC that shares some state with the
destination CC, applying each arm of the `or` to a copy of the template
as if it were a standalone query, then building a projection list and
creating a `ComputedTable::Union`. This is pushed into the destination
CC's `computed_tables` list.
Finally, the variables projected from the UNION are bound in the
destination CC, so that unification occurs, and projection of the
outermost query can use bindings established by the `or-join`.
This commit includes projection of type codes from heterogeneous `UNION`
arms: we compute a list of variables for which a definite type is
unknown in at least one arm, and force all arms to project either a type
tag column or a fixed type. It's important that each branch of a UNION
project the same columns in the same order, hence the projection of
fixed values.
The translator is similarly extended to project the type tag column name
or the known value_type_tag to support this.
Review comment: clarify union type extraction.
This commit:
- Defines a new kind of column, distinct from the eavt columns in
`DatomsColumn`, to model the rows projected from subqueries. These
always name one of two things: a variable, or a variable's type tag.
Naturally the two cases are thus `Variable` and `VariableTypeTag`.
These are cheap to clone, given that `Variable` is an `Rc<String>`.
- Defines `Column` as a wrapper around `DatomsColumn` and
`VariableColumn`. Everywhere we used to use `DatomsColumn` we now
allow `Column`: particularly in constraints and projections.
- Broadens the definition of a table list in the intermediate
"query-sql" representation to include a SQL UNION. A UNION is
represented as a list of queries and an alias.
- Implements translation from a `ComputedTable` to the query-sql
representation. In this commit we only project vars, not type tags.
Review comment: discuss bind_column_to_var for ValueTypeTag.
Review comment: implement From<Vec<T>> for ConsumableVec<T>.
Complex `or`s are translated to SQL as a subquery -- in particular, a
subquery that's a UNION. Conceptually, that subquery is a computed
table: `all_datoms` and `datoms` yield rows of e/a/v/tx, and each
computed table yields rows of variable bindings.
The table itself is a type, `ComputedTable`. Its `Union` case contains
everything a subquery needs: a `ConjoiningClauses` and a projection
list, which together allow us to build a SQL subquery, and a list of
variables that need type code extraction. (This is discussed further in
a later commit.)
Naturally we also need a way to refer to columns in a computed table.
We model this by a new enum case in `DatomsTable`, `Computed`, which
maintains an integer value that uniquely identifies a computed table.
When we started expanding and narrowing type sets, it became impossible
to conclusively know during pattern application whether a type was
known. We now figure that out at the end: if a variable has only a
single known type, we don't need to extract its type tag.
* Pre: Expose more in edn.
* Pre: Make it easier to work with ValueAndSpan.
with_spans() is a temporary hack, needed only because I don't care to
parse the bootstrap assertions from text right now.
* Part 1a: Add `value_and_span` for parsing nested `edn::ValueAndSpan` instances.
I wasn't able to abstract over `edn::Value` and `edn::ValueAndSpan`;
there are multiple obstacles. I chose to roll with
`edn::ValueAndSpan` since it exposes the additional span information
that we will want to form good error messages in the future.
* Part 1b: Add keyword_map() parsing an `edn::Value::Vector` into an `edn::Value::map`.
* Part 1c: Add `Log`/`.log(...)` for logging parser progress.
This is a terrible hack, but it sure helps to debug complicated nested
parsers. I don't even know what a principled approach would look
like; since our parser combinators are so frequently expressed in
code, it's hard to imagine a data-driven interpreter that can help
debug things.
* Part 2: Use `value_and_span` apparatus in tx-parser/.
I break an abstraction boundary by returning a value column
`edn::ValueAndSpan` rather than just an `edn::Value`. That is, the
transaction processor shouldn't care where the `edn::Value` it is
processing arose -- even we care to track that information we should
bake it into the `Entity` type. We do this because we need to
dynamically parse the value column to support nested maps, and parsing
requires a full `edn::ValueAndSpan`. Alternately, we could cheat and
fake the spans when parsing nested maps, but that's potentially
expensive.
* Part 3: Use `value_and_span` apparatus in query-parser/.
* Part 4: Use `value_and_span` apparatus in root crate.
* Review comment: Make Span and SpanPosition Copy.
* Review comment: nits.
* Review comment: Make `or` be `or_exactly`.
I baked the eof checking directly into the parser, rather than using
the skip and eof parsers. I also took the time to restore some tests
that were mistakenly commented out.
* Review comment: Extract and use def_matches_* macros.
* Review comment: .map() as late as possible.
Part 1, core: use Rc for String and Keyword.
Part 2, query: use Rc for Variable.
Part 3, sql: use Rc for args in SQLiteQueryBuilder.
Part 4, query-algebrizer: use Rc.
Part 5, db: use Rc.
Part 6, query-parser: use Rc.
Part 7, query-projector: use Rc.
Part 8, query-translator: use Rc.
Part 9, top level: use Rc.
Part 10: intern Ident and IdentOrKeyword.
mod.rs defines the module and ConjoiningClauses itself, complete with
methods to record facts and ask it questions.
pattern.rs, predicate.rs, resolve.rs, and or.rs include particular
functionality around accumulating certain kinds of patterns.
Only `or.rs` includes significant new code; the rest is just split.
* Part 1 - Create as_edn_value function.
* Do not include defaults inside output.
* Pretty-printed by default. Do we want to make that a flag?
* Includes simple test just to make sure it works.
* Part 2 - only include ident if available.
* Part 3 - Remove spacing and newlines as unnecessary.
* Update function to build edn::Value directly rather than parsing from string
* Update test to actually test the functionality.
* Address review comments ncalexan.
* Rename `as_edn_value` to `to_edn_value`.
* Move `db/src/values.rs` to `core/src/values.rs` so we can reference inside `core/src/ib.rs`.
* Add `lazy-static` crate to core `Cargo.toml`
* Expose `values` as a public module from `core`.
* Update references to values in `db/src/bootstrap.rs` & `db/src/lib.rs`.
* Add new static vars for `DB_FULLTEXT`, `DB_INDEX` & `DB_IS_COMPONENT`.
* Use static vars exposed in `values` inside `to_edn_value`.
* Remove `db/id` as key in attribute output and use `entid` as `db/ident` if no `ident` is found for that `entid`.
* Update test to match new expected output.
* Add doc comment for function
* Address review comments ncalexan.
* Update function docstring to give clearer description of function.
* Do not all entid at all to output.
* Clean up code fetching ident (make it rustier).
* Address review comments rnewman.
* Extract out to new `to_edn_value` functions code for creating `edn::Value`\'s for `ValueType` and `Attribute`.
* Use `map()` to create schema edn value rather than a loop.
* Address review comments rnewman.
* pass cloned instance of ident to `Attribute::get_edn_value`.
* update `use` import for `edn`.
* remove unnecessary call when using ident as key on `associate_ident`.
* Fixed bug whereby we didn't differentiate between `db.index/value` and `db.index/identity` when generating `edn::Value`
* Add extra assert at the end to ensure we get the same output when we convert the same schema to edn multiple times
* Move check for type of uniqueness to `match` statement.
* Also use `iter` instead of `into_iter` when iterating schema map.
* Pre: Fix error in parser macros.
* Pre: Make test unwrapping more verbose.
* Pre: Make lookup refs be (lookup-ref a v) in the entity position.
This has the advantage of being explicit in all situations and
unambiguous at parse-time. This choice agrees with the Clojure
implementation but not with Datomic. Datomic treats [a v] as a lookup
ref, is ambiguous at parse-time, and is disambiguated in ways I do not
understand at transaction time. We mooted making lookup refs [[a v]]
and outlawing nested value vectors in transactions, but after
implementing that approach I decided it was better to handle lookup
refs at parse time and therefore outlawing nested value vectors is not
necessary.
* Handle lookup refs in the entity and value columns. Fixes#183.
* Pre 0a: Use a stack instead of into_iter.
* Pre 0b: Dedent.
* Pre 0c: Handle `e` after `v`.
This allows to use the original `e` while handling `v`.
* Explode value lists for :db.cardinality/many attributes. Fixes#284.
* Parse and accept map notation. Fixes#180.
* Pre: Modernize add() and retract() into one add_or_retract().
* Pre: Add is_collection and is_atom to edn::Value.
* Pre: Differentiate atoms from lookup-refs in value position.
Initially, I expected to accept arbitrary edn::Value instances in the
value position, and to differentiate in the transactor. However, the
implementation quickly became a two-stage parser, since we always
wanted to parse the resulting value position into some other known
thing using the tx-parser. To save calls into the parser and to allow
the parser to move forward with a smaller API surface, I push as much
of this parsing as possible into the initial parse.
* Pre: Modernize entities().
* Pre: Quote edn::Value::Text in Display.
* Review comment: Add and use edn::Value::into_atom.
* Review comment: Use skip(eof()) throughout.
* Review comment: VecDeque instead of Vec.
* Review comment: Part 0: Rename TempId to TempIdHandle.
* Review comment: Part 1: Differentiate internal and external tempids.
This breaks an abstraction boundary by pushing the Internal/External
split up to the Entity level in tx/ and tx-parser/. This just makes
it easier to explode Entity map notation instances into Entity
instances, taking an existing External tempid :db/id or generating a
new Internal tempid as appropriate. To do this without breaking the
abstraction boundary would require adding flexibility to the
transaction processor: we'd need to be able to turn Entity instances
into some internal enum and handle the two cases independently. It
wouldn't be too hard, but this reduces the combinatorial type
explosion.
