Despite multiple attempts on reusing the RefMut we get from the first
borrow as stored in `authenticator`, using it in the needed spot
seemed impossible despite the compiler trying to make suggestions.
Dropping `authenticator` and reborrowing is done for that reason,
fair enough.
Let's try to avoid spamming crates.io and instead keep publishing
everything as is. The patch will be present in some of the more
recent crates, and for specific crates like youtube3 and drive3
I will create specific patch releases.
They cause issues now that we can't represent them anymore via
the call-type protocol enumeration. Something went wrong there
and it's not worth fixing it on CLI side (probably nobody uses them)
even though they *were* good for debugging one previously.
…because in v3 of yup-oauth2, nobody seems to implement it with
anything else but `None`.
That's probably fair and we can revisit this once we upgrade
to a 5.1 or newer.
See 0710d310f8
In 0.2 there is only `Display` available, so we have to be wasteful
by allocating a new string.
However, this is less work than switching to 0.3 which supports
`AsRef`, which is left for another day.
This doesn't work completely yet because yup_hyper_mock is pulling in old crates: it needs to be updated or removed, but I'm not sure that I have enough context to do it.
I've switched to a new M1 laptop which couldn't build the old dependencies because the old ring library version doesn't compile on mac arm (but new version do). There's no way to update ring in isolation, so I've got to update the whole tree at once.
the top level module re-exports a few very common items like the hub
struct of the particular API and Result, Error, and Delegate.
Signed-off-by: Petros Angelatos <petrosagg@gmail.com>
In some environments, a variable-existence test in a template was
yielding a false negative which resulted in the template not generating
'-force' targets for API JSON files, even when the `FETCH_APIS`
environment variable was set during template rendering.
This commit adjusts the way the test for the variable presence is done
such that the '-force' targets now render when they should.
