Before we could not observe the case where:
* thread A is blocked on B
* cycle detected in thread B
* some participants are on thread A and have to be marked
(In particular, I commented out some code that seemed necessary and
didn't see any tests fail)
If I am not in an error, all current tests uses "static" cycles -- a
cycle always present. Let's spice that up by adding conditional cycles:
cycles that appear only for specific impls
This had two unexpected consequences, one unfortunate, one "medium":
* All `salsa::Database` must be `'static`. This falls out from
`Q::DynDb` not having access to any lifetimes, but also the defaulting
rules for `dyn QueryGroup` that make it `dyn QueryGroup + 'static`. We
don't really support generic databases anyway yet so this isn't a big
deal, and we can add workarounds later (ideally via GATs).
* It is now statically impossible to invoke `snapshot` from a query,
and so we don't need to test that it panics. This is because the
signature of `snapshot` returns a `Snapshot<Self>` and that is not
accessible to a `dyn QueryGroup` type. Similarly, invoking
`Runtime::snapshot` directly is not possible becaues it is
crate-private. So I removed the test. This seems ok, but eventually I
would like to expose ways for queries to do parallel
execution (matklad and I had talked about a "speculation" primitive
for enabling that).
* This commit is 99% boilerplate I did with search-and-replace. I also
rolled in a few other changes I might have preferred to factor out,
most notably removing the `GetQueryTable` plumbing trait in favor of
free-methods, but it was awkward to factor them out and get all the
generics right (so much simpler in this version).
This was an oversight before -- the current type implies that
one introduce a synthetic write (and hence a new revision) from
a `Frozen<DB>`! Not cool.
Quickest POC I could create to get some potentially cyclic queries to
not panic and instead return a result I could act on. (gluon's module
importing need to error on cycles).
```
// Causes `db.query()` to actually return `Result<V, CycleError>`
fn query(&self, key: K, key2: K2) -> V;
```
A proper implementation of this would likely return
`Result<V, CycleError<(K, K2)>>` or maybe larger changes are needed.
cc #6
We now track the last revision in which constants were modified. When
we see a constant query result, we record the current revision as
well. Then later we can check if the result is "still" constant. This
lets us cut out a lot of intermediate work.
Previosly, panicking query would remove in-progress memos from the
table.
However, we use panic for cancellation, so it's reasonable to **not**
remove the old result after panic. This is also known as string
exception safety guarantee: not only database is in *some* consistent
state after a panic, it is in the same state it was before the panic!
Transparent queries are not really queries: they are just plain
uncached functions without any backing storage.
Making a query transparent can be useful to figure out if caching it
at all is a win
