This explanation is slightly complicated by the fact that I think
that one of the calls to update_guardset_internal() is possibly
unnecessary, and that one of the calls that it makes is potentially
ill-advised.
I'm not going to make those changes right now, however, because they
are potentially a little destabilizing.
Now it is an Option, and is set to None if bridges aren't enabled.
This simplifies `replace_bridge_config` a bit, and forces us to
check for `None` in a few more places.
We do this by checking the FirstHops we're about to return, and when
they correspond to bridges, looking up an appropriate BridgeRelay
in the current BridgeSet (if we can).
The `GuardMgr` code has functionality to tell the DirMgr "Hey,
don't switch to the new NetDir yet: we still need more guard
information!" But we never want to do that if we're selecting
bridges, since they don't come from the NetDir.
Instead of duplicating the logic about which guard sample uses which
universe, we explicitly ask it, and then use that universe. This
will avoid trouble if/when we introduce more samples.
I'm using an Arc<[]> here though I think that there's a chance
that a simple Vec<> would suffice. Since it's an internal type,
nothing will break if we change it later.
Also, we now switch into and out of the Bridges guard sample
as needed. However, that selection is not (yet) built from the
list of bridges. That will come soon.
The first part changes which guard set is active based on based on
the parameters, which always come from a NetDir; the second changes
the contents of the active guard set, based on a Universe.
These arguments were used only for legacy (testing) purposes; the
tests now use `TestNetDirProvider`. This lets us simplify our
internal logic for passing a `NetDir` to our samples, and prepare
for having a `BridgeSet` to pass there instead.
This is a breaking change to `guardmgr` and `circmgr`.
0.99.[012] have a bug https://github.com/JelteF/derive_more/issues/114
which makes the Deref derive for bridgedesc::StateGuard not work
and therefore breaks minimal-versions CI.
It seems simpler to require the newer version everywhere.
Previously we always set `dir_info_missing` to `false` for new
guards, since new guards could only be taken from ones that were
present in the NetDir. But for bridges, we don't download their
info until _after_ we have chosen them as guards.
In this and the upcoming commits I'll be changing how guards related
to `NetDir` and to `Relay`. Previously, a guard could only come
from (or be updated from) a `Relay` in a `NetDir`. Soon it will be
able to be built from a bridge as well.
To do this, I'm defining a `Universe` trait (name negotiable) that
represents a set of things that may be guards. I'm going to
continue extending its functionality until there are no more
methods in guard.rs or sample.rs that take `NetDir`.
This commit removes most of the usage of `NetDir` and `Relay` in
`guard.rs`.
Nick Mathewson
Ian Jackson
Dimitris Apostolou