By performing more pointer casting gymnastics, additional types of
pointer can be supported. For example, now const qualified pointers
can be passed through thanks to the use of constCast. Also, having
explicit ptrCast allows nested pointers (e.g. a pointer to a slice,
such as `*[]const u8`) to be passed as userdata as well (the compiler
refuses to coerce a pointer to a pointer to `?*anyopaque` for some
reason, I guess because maybe it's ambiguous somehow?) Hopefully this
extra casting does not appreciably reduce the compiler's ability to
catch real bugs (for example, on a 64-bit machine, ptrCast can convert
a *u64 into a **u64 because there is no alignment change).
Also, the `volatile` pointer specifier is still not supported.
`allowzero` pointers probably also have a problem. Those are both
extreme edge cases, however.
This was intended to work before but did not due to an oversight.
Specifically, because the userdata pointers are stored as ?*anyopaque,
which does *not* have the const qualifier, they must have their const
qualifiers also removed. This is safe because the thunk guarantees
that the consumer code never sees the non-const version of the
pointer, and the nats library itself does nothing except pass the
pointer through to the user callback.
The tests have been updated to ensure this case works. The examples
still use a mutable userdata pointer to show that that case also
works. More tests could be added for the sake of increased rigor, but
I don't think it adds much.
One of the things I have done a better job of internalizing while
working with zig over the last few months is that the less magic that
exists, the better. In the case of parameterized functions, this means
that it is much better to restrict the range of types that are
permitted to be passed than to perform type manipulation. In other
words, it's more confusing to see a function that is parameterized
with `SomeType` taking a pointer to that type than having it be
parameterized directly to take the pointer. Obviously there are
exceptions to this rule, like std.mem.eql taking slices of its
parameterized type.
In fact, this new approach fixes some edge cases. Null userdata may now
be passed in, since the user can now actually specify an optional
pointer type (e.g. `?*void` may be used to provide always-null
userdata). Additionally, a pointer to a constant value can now be
passed in, which wasn't possible before (this could have been worked
around by use of constCast and being careful, but that's an
exceedingly bad option compared to having the type system work for
you).
I was originally going to have this generate a random token by default,
but I don't think it matters. the bigger thing would perhaps be to
manage the listen port to avoid conflicts, but this isn't currently a
problem, so I will pretend it won't ever be a problem.
If you think these unit tests look crappy, that's because they are.
Actual behavioral tests are pretty much exclusively relying on the
underlying C library being well-tested, so all these tests are really
trying to do is ref the bindings in our endpoints to make sure there
aren't any compilation issues with them. Obviously this is making a
difference, since a variety of issues have been found and addressed as
a result.
Since I went to the effort of getting this to build with LibreSSL, it
makes sense to check that it works. And it seems to. There's nothing
special about the certificates added, except that they don't expire
for 100 years, which is hopefully long enough that nothing will matter
any more. They get baked into the test executable, since this was the
best way I could figure out to be able to consistently load them
during runtime, since they're stored relative to the source, not
necessarily relative to the working directory in use when the test
executable is run.
This required some augmentation to the test server launcher which will
make it easier to add additional flags to it in the future, if that
becomes necessary.
Tests are CC0/public domain because there's no reason for them not to
be. Examples are also CC0/public domain, but this may be a little bit
weird because they are largely straightforward ports of examples from
nats.c which carry the Apache license. However, I personally wrote
them against the zig bindings and I doubt anyone will end up in a
court of law due to their software containing uselessly trivial
example code.
Various copypasta and typing errors around the thunks, some other
copypasta errors, etc. Also add a test which hits a decent chunk of
the API surface of ConnectionOptions, which is how these issues were
caught in the first place. This is an ugly test, but it has clearly
served its purpose.
I created a utility class to spawn a NATS server. This assumes it is
installed and in PATH, which should be easy enough to accomplish for a
CI environment. The current approach will not work with parallel
tests, but that's not a practical reality, so this route should be
fine for the time being. It might be nice to spawn a default server at
the beginning of testing and tear it down at the end, to save waiting
for the startup/shutdown time in many individual tests. This makes me
wonder: if I initialize a server in the beginning of the `test` block
in main that imports the other modules, does that scope stay live
while the "child" tests are running? My default guess would be
probably not, but that would be very convenient, so I might try it out
and see.
These are mostly just tests to make sure the wrappers function as
written, rather than testing for any specific output. This commit
demonstrates the simple value of actually referencing the various
wrapper functions, as a variety of really basic compilation errors
were caught and addressed.
This approach seems nice, and as a bonus it makes it easier to run the
tests through the module interface rather than by importing the
sources directly, which I think is a good dog food approach.
