| Commit message (Collapse) | Author | Age |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This represents a major change in the HTTP code base, consisting
of a complete revamp of the HTTP API. The changes here are too
numerous to mention, but the end result should be a vastly
simpler API for both server and client applications.
Many needless allocations were removed by providing fixed buffers
for various parameters and headers when possible.
A few bugs were fixed. Most especially we have fixed some bugs
around very large URIs and headers, and we have also addressed
conformance bugs to more closely conform to RFCs 9110 and 9112.
As part of this work, the APIs for WebSockets changed slightly
as well. In particular the properties available for accessing
headers have changed.
There is still documentation conversion work to do, and additional
functionality (such as proper support for chunked transfers), but
this is a big step in the right direction.
|
| |
|
| |
Fix typo of filename. Can not find filename
|
| |
|
|
|
|
|
|
|
|
| |
This simplified API lets callbacks obtain the response from the
connection objection directly, and does not require the aio to carry
it as a parameter. Further, the request and response are both
stored inline in the connection, reducing allocations.
This is at present only for the server; the client will get a similar
set of changes.
|
| |
|
|
|
| |
This is a step towards simplifying this API and ultimately simplifying
the HTTP callback API used for the server side.
|
| | |
|
| |
|
|
|
|
| |
The only thing using this was the transport lookups, but as
those transports are now fully initialized in nng_init, we
no longer need to lock that at all.
|
| | |
|
| |
|
|
| |
The CV_INITIALIZER is error prone, as it cannot use cv_until.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
The API is identical, except that some names have changed, and this is now a
header library in `nng/args.h` - so the core library does not need to carry this
code in binaries. Being a header library also means it is not necessary to
link against NNG, and it does not include any parts of NNG; it only depends on
a standard C99 or C11 environment.
|
| | |
|
| | |
|
| |
|
|
|
| |
This will ensure that we fast fail if a test cannot complete,
rather than muddling on and reporting failures elsewhere.
|
| |
|
|
| |
This was occasionally causing "sigabrt" and similar failures in the tests.
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
This aligns more closely with the nng_ctx functions.
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Also, clarification and add socket option docs for MAXRECVSZ, RECONNMINT, RECONNMAXT
|
| |
|
|
|
|
|
|
| |
This flag failed to provide real zero copy that it was intended for,
and it also involved extra allocations. Further, the API for it was
brittle and error prone.
Modern code should just work directly with nng_msg structures.
|
| |
|
|
|
| |
The CMAKE framework does this properly now by ensuring that
nng uses nng_find_package for any of its own dependencies.
|
| |
|
|
|
| |
Using nng_find_package as the helper fixes this, by ensuring that the
dependency gets properly added to incorporating projects.
|
| |
|
|
|
| |
This is a recent regression that affects any server that does not
explicitly set an HTTP status code.
|
| |
|
|
|
|
| |
There are no other consumers for this, and reasonably unlikely to
be others for now. (Other use cases are JWTs, but that would be
another whole set of functionality that we're not ready to take on.)
|
| |
|
|
|
|
| |
Nothing else uses it, and nothing else *should* use it because SHA1 is insecure.
WebSockets have to use it by definition, unfortunately. The implementation is
not very fast, but doesn't have to be for the use case of websocket keying.
|
| |
|
|
|
| |
This should simplify things for developers. Just one header to include
in most cases now.
|
| |
|
|
| |
This is another occasional test failure due to EADDRINUSE that we can avoid.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
This should help Linux platforms scale even further with NNG.
Port events and *maybe* poll are the last to do. Probably select
will remain left in the cold, because honestly select based systems
are already performance constrained.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This allows greatly increased scalability for kqueue based systems
with lots of cores (more likely FreeBSD than Darwin, as most macs
only have a smattering of cores), but even for macs we can engage
a few cores for system calls giving improvements.
The implementation here is pretty simple -- each file descriptor
gets assigned to its own kqueue by taking the numeric value of the
file descriptor modulo the number of kqueues we have opened.
The same approach will be adopted for epoll and Solaris/illumos port events.
|
| |
|
|
|
|
|
| |
This involved test-specific hacks, since connect() for UNIX domain
sockets always completes synchronously one way or the other, even though
it is documented that it might not. This found a bug, with an uninitialized
poll FD as well!
|
| |
|
|
|
| |
This was observed only in the select poller, but the behavior of having
the same file descriptor registered in the poller more than once is undefined.
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
| |
This introduces a new option "NNG_OPT_LISTEN_FD", understood by TCP, TLS,
and (on POSIX systems) IPC. This option is used to pass a file descriptor
or handle (Windows) that is already listening (ready for ACCEPT to be called).
For TCP and TLS, the socket must be of type AF_INET or AF_INET6, and for IPC
it must be of type AF_UNIX.
|
| |
|
|
|
| |
This will be slower, as each vector element has to be sent in a single
system call, but for platforms that lack sendmsg it will at least work.
|
| | |
|
| | |
|
| |
|
|
|
| |
This includes checks to determine if those functions are present,
and a test case to verify that scatter gather with UDP works.
|
| |
|
|
|
|
| |
While here, remove some code paths that do not occur by definition.
(For example, if the resolver succeeds, we will definitely have a
valid set of addresses, but if it fails, we will definitely not.)
|
