ObjFW  History of src/OFTCPSocket.m of f37b373a95db781b

This allows using IPv4 and IPv6 for the same connection. file: [31cc0fb260] check-in: [1abe178915] user: js, branch: mptcp, size: 15414 [annotate] [blame] [check-ins using] [diff]

Unfortunately, I haven't found a way to reliably check if a connection
is using MPTCP on macOS/iOS yet, so -[usesMPTCP] always returns false
for now. In my tests with SIOCGCONNINFO, the flags were always 0. file: [2efb3cbab4] check-in: [4e763857b8] user: js, branch: mptcp, size: 14627 [annotate] [blame] [check-ins using] [diff]

It doesn't seem to be needed anymore. file: [28180f5555] check-in: [cf0effe834] user: js, branch: trunk, size: 11139 [annotate] [blame] [check-ins using] [diff]

While it works for those two protocols, it's a bad precedent as other
protocols have different types for ports. file: [74ced9f7d1] check-in: [a54730b88f] user: js, branch: trunk, size: 11410 [annotate] [blame] [check-ins using] [diff]

struct sockaddr can contain a variable length array, which breaks the
union. file: [8b6de7a0f6] check-in: [25adea9a23] user: js, branch: trunk, size: 11535 [annotate] [blame] [check-ins using] [diff]

The previous API could never work cleanly and would always require
hacks, as it needed intercepting all interactions of OFTCPSocket with
the raw socket and did not work at all if the OFTCPSocket had anything
in its read buffer before starting the TLS handshake. This also could
not be fixed easily, as it would have required the object to contain two
read buffers, one for the unencrypted connection and one for the
encrypted connection. There was also no clean way to perform the
handshake in a non-blocking way.

The new API is a lot cleaner and requires none of the hacks, but using
it requires slightly more work. But this is more than made up for by
making a fully asynchronous handshake possible. It uses the concept of a
stream wrapping another stream, meaning the entire connecting part is
being handled by OFTCPSocket and then the connected socket is passed off
to OFTLSStream to wrap it. This also makes for a lot cleaner separation
of concerns. file: [cfa63a9a83] check-in: [d30efa8bbf] user: js, branch: trunk, size: 11510 [annotate] [blame] [check-ins using] [diff]

There is a nasty bug in HP-UX: When using the proper socket APIs that
take a socklen_t (which is 64 bit in LP64 mode), it still calls into
methods that expect an int - so there is a type mismatch between the
headers and the compiled system libraries. This leads to all socklen_t
being interpreted as 0 (due to big endian). Defining
_HPUX_ALT_XOPEN_SOCKET_API makes it use the correct symbols in the
system libraries instead. file: [49e1470816] check-in: [1833194867] user: js, branch: trunk, size: 11571 [annotate] [blame] [check-ins using] [diff]

Arguments that can just be captured by the block don't need to be
repeated as an argument. A lot of arguments passed to a delegate are
therefore not needed for a block. file: [36820b9499] check-in: [d7ddb3dbc7] user: js, branch: trunk, size: 22984 [annotate] [blame] [check-ins using] [diff]

This is intended for SCTP and SPX. file: [e938092c44] check-in: [e5b2ee56ff] user: js, branch: trunk, size: 26569 [annotate] [blame] [check-ins using] [diff]

This is a new superclass for OFTCPSocket to allow sharing code with
other IP stream socket types, namely SCTP. file: [120137b646] check-in: [3dfe642dd3] user: js, branch: trunk, size: 3771 [annotate] [blame] [check-ins using] [diff]

This removes the ability to let a thread add or remove objects for
another thread. This features wasn't used at all, but caused significant
problems with kqueue and epoll, as this often resulted in removing a
file descriptor after it had already been closed. file: [6288b956d6] check-in: [24eff635db] user: js, branch: trunk, size: 26610 [annotate] [blame] [check-ins using] [diff]

These are different enough that it makes no sense to have them in one.
Having them in one means all methods are @optional and typos cannot be
caught by the compiler. file: [5389356633] check-in: [e3c911c7f5] user: js, branch: trunk, size: 26024 [annotate] [blame] [check-ins using] [diff]

This temporarily doesn't use the search domains anymore and makes CNAMEs
slightly more inefficient. The next commits will fix both. file: [21a88b2103] check-in: [ee592cf603] user: js, branch: trunk, size: 26020 [annotate] [blame] [check-ins using] [diff]

This is required for MorphOS. file: [34fdf0c261] check-in: [366ce5073a] user: js, branch: trunk, size: 26056 [annotate] [blame] [check-ins using] [diff]

This does not fully fix it yet, but at least the socket tests in the
test suite pass on 3DS now. file: [909cd558d8] check-in: [ab89c47f42] user: js, branch: trunk, size: 26095 [annotate] [blame] [check-ins using] [diff]

Because there is usually only one resolver per thread, having the
delegate as a property on the resolver would not work. It is therefore
passed as an argument.

This only changes the public API so far. Internally, the old target /
selector / context pattern is still being used. This will be changed
later. file: [b5ab8aeaeb] check-in: [0afb04d93c] user: js, branch: trunk, size: 25444 [annotate] [blame] [check-ins using] [diff]

This avoids the entire problem of keeping the buffer alive until the
write finished. file: [d516c427cd] check-in: [f218986f51] user: js, branch: trunk, size: 25724 [annotate] [blame] [check-ins using] [diff]

Otherwise, there would be two methods for every operation: One for
success and one for failure. It also makes it easy to forget about
handling failure, so it's better to always pass an optional exception. file: [dd4630f033] check-in: [064dbe5127] user: js, branch: trunk, size: 25706 [annotate] [blame] [check-ins using] [diff]

The target / selector approach had several drawbacks:

* It was inconvenient to use, as for every read or write, a target,
selector and context would need to be specified.
* It lacked any kind of type-safety and would not even warn about using
a callback method with the wrong number of parameters.
* It encouraged using a different callback method for each read or
write call, which results in code that is hard to follow and also
slower (as it needs to recreate the async operation with a new
callback every time). file: [df6639e5af] check-in: [d16ad96cbd] user: js, branch: trunk, size: 26324 [annotate] [blame] [check-ins using] [diff]

Instead, OFDNSResolver is used for everything now. file: [0378e7d3f9] check-in: [6e9ee122eb] user: js, branch: trunk, size: 28448 [annotate] [blame] [check-ins using] [diff]

This pretends that net_getsockopt() exists, which it does not yet. As
soon as it lands in libogc, this will work again. file: [3cac08541a] check-in: [bed2db0fd3] user: js, branch: trunk, size: 16854 [annotate] [blame] [check-ins using] [diff]

Instead of containing a struct sockaddr_storage - which does not exist
on all supported platforms - it now contains a union of all struct
sockaddr_* types.

Additionally, if struct sockaddr_in6 does not exist, it is defined so
that an IPv6 can be stored even if the system does not support IPv6. file: [a77bbf08e1] check-in: [8f810ecb7e] user: js, branch: trunk, size: 15412 [annotate] [blame] [check-ins using] [diff]

This has always been wrong. file: [9f8d11c4a6] check-in: [e36455441d] user: js, branch: trunk, size: 15491 [annotate] [blame] [check-ins using] [diff]

This replaces of_address_to_string_and_port(). file: [7dca3c8c74] check-in: [8aeee6680a] user: js, branch: trunk, size: 15482 [annotate] [blame] [check-ins using] [diff]

Instead, explicitly declare them, as OF_ASSUME_NONNULL_{BEGIN,END} does
not apply to implicit return types. This means that after this commit,
all init methods have a nonnull return type, as they should have. file: [f8b1b83a89] check-in: [2f4e0df8be] user: js, branch: trunk, size: 15232 [annotate] [blame] [check-ins using] [diff]

Previously, socket_helpers.h would just define close() to whatever is
needed to close a socket. However, this makes it possible to
accidentally call a socket-specific close on something that is not a
socket. Having to explicitly call closesocket() (and defining it to
close() on platforms where it's the same) prevents this. file: [16f6a959b9] check-in: [b4cd87cd81] user: js, branch: trunk, size: 14861 [annotate] [blame] [check-ins using] [diff]

This matches Apple's style. file: [b98f01fc09] check-in: [6b77a5dd8b] user: js, branch: trunk, size: 14831 [annotate] [blame] [check-ins using] [diff]

Casts are now written like types in variable declarations. file: [c7187f1cc0] check-in: [4af49a13c3] user: js, branch: trunk, size: 14852 [annotate] [blame] [check-ins using] [diff]

Forgot to add 2017, even though I already did quite some changes in
2017. file: [f234be8d33] check-in: [44f45c2e35] user: js, branch: trunk, size: 14820 [annotate] [blame] [check-ins using] [diff]

Right now, this is only useful for OFTCPSocket, as this is the only
class so far not establishing the stream in the init method. However,
this adds the general infrastructure to allow reuse to all subclasses of
OFStream. file: [2735f938b2] check-in: [1de551cb5f] user: js, branch: trunk, size: 14814 [annotate] [blame] [check-ins using] [diff]

This is required for static linking, as otherwise
+[OFStdIOStream_Win32Console load] will not be called. file: [15a3d3213a] check-in: [e53a4fd61c] user: js, branch: trunk, size: 14825 [annotate] [blame] [check-ins using] [diff]

This removes the port registry that existed for the Wii and instead
tries binding to random ports in a loop until it succeeds or errno is
not EADDRINUSE. This is done for the Wii and Nintendo 3DS now.

This also includes several other socket-related cleanups. file: [f48a1756ef] check-in: [12b8a7552e] user: js, branch: trunk, size: 14809 [annotate] [blame] [check-ins using] [diff]

While at it, also update the mail address. file: [73559345eb] check-in: [cec0f072f8] user: js, branch: 0.8, size: 14131 [annotate] [blame] [check-ins using] [diff]

While at it, also update the mail address. file: [ccb8fac7b9] check-in: [2a27cf3000] user: js, branch: trunk, size: 14433 [annotate] [blame] [check-ins using] [diff]

This increases the required GCC version from 4.0 to 4.6 (exception:
Apple GCC, which already supports this with >= 4.0 starting with OS X
10.5). Since even GCC 4.6 is really old by now, there is no point in
still supporting something even older and making the code ugly because
of that. While some hardware and OS support was dropped from GCC 4.6
compared to GCC 4.0, there is nothing in there that would be an
interesting target with the exception of BeOS maybe - but a port to BeOS
can also be achieved using the Haiku support. The other dropped OSes are
mostly old versions of OSes while newer ones are still being supported
(and those newer versions of those OSes still support the same
hardware). file: [0119b6524b] check-in: [48980f2297] user: js, branch: trunk, size: 14429 [annotate] [blame] [check-ins using] [diff]

This is necessary as the Wii does not allow picking a random free port,
and thus we need to track which ports are used. file: [a3e1049524] check-in: [ed4549ddd3] user: js, branch: trunk, size: 14655 [annotate] [blame] [check-ins using] [diff]

This also adds -[isTCPNoDelayEnabled] and -[isKeepAliveEnabled],
together with OFGetOptionFailedException to indiciate failure to
retrieve the current value of the option. file: [43965236bf] check-in: [12fbe85aab] user: js, branch: trunk, size: 13902 [annotate] [blame] [check-ins using] [diff]

of_getsockname() calls getsockname() after acquiring a mutex. This is
necessary as some implementations of getsockname() are not thread-safe. file: [421ec8b16e] check-in: [f18615e057] user: js, branch: trunk, size: 13027 [annotate] [blame] [check-ins using] [diff]

The old behaviour where the exception would access errno directly on
creation of the exception was very fragile. The two main problems with
it were that sometimes it would pick up an errno even though none had
been set and in other cases that when the exception was created errno
had already been overridden.

This also greatly increases errno handling on Win32, especially in
conjunction with sockets. It can still be improved further, though. file: [f120fb3aeb] check-in: [62e2de30b9] user: js, branch: trunk, size: 12995 [annotate] [blame] [check-ins using] [diff]

This prevents reading old data when closing and reconnecting the socket. file: [e59925afa4] check-in: [c6a4b0a95f] user: js, branch: trunk, size: 11203 [annotate] [blame] [check-ins using] [diff]

This means that OFObject.h imports macros.h now, making it unnecessary
to manually import macros.h in almost every file. And while at it, also
import autorelease.h in OFObject.h, so that this doesn't need to be
manually imported in almost every file as well. file: [dc50295dfd] check-in: [13ee56edf3] user: js, branch: trunk, size: 11063 [annotate] [blame] [check-ins using] [diff]

It caused a warning about an implicit cast on OS X/PPC64. file: [3b9e3e0a34] check-in: [3fa37632f9] user: js, branch: trunk, size: 11106 [annotate] [blame] [check-ins using] [diff]

It is now called +[getHost:andPort:forAddress:]. This makes it much
clearer from the selector alone what it actually does. file: [fb3f587e28] check-in: [fb590316ba] user: js, branch: trunk, size: 11138 [annotate] [blame] [check-ins using] [diff]

This is in preparation for UDP sockets, so that they can share the
resolving functionality with TCP sockets. file: [4b99095be9] check-in: [49b1989b63] user: js, branch: trunk, size: 11085 [annotate] [blame] [check-ins using] [diff]

The new file socket.h includes all headers required for sockets on the
used platform, while the file socket_helpers.h defines the BSD API
functions to the platform specific functions if necessary.

This cleans up the classes dealing with sockets a lot and also reduces
code duplication. file: [f0c5480103] check-in: [81d47f4398] user: js, branch: trunk, size: 15858 [annotate] [blame] [check-ins using] [diff]

This replaces OF_STRING_ENCODING_NATIVE, as a define was not flexible
enough (determining the native OS encoding at startup was not possible). file: [367b0cdd47] check-in: [e54c8c0368] user: js, branch: trunk, size: 17417 [annotate] [blame] [check-ins using] [diff]

This mostly removes the argument for the class in which the exception
occurred. As backtraces were recently added for all platforms, the
passed class does not give any extra information on where the exception
occurred anymore.

This also removes a few other arguments which were not too helpful. In
the past, the idea was to pass as many arguments as possible so that it
is easier to find the origin of the exception. However, as backtraces
are a much better way to find the origin, those are not useful anymore
and just make the exception more cumbersome to use. The rule is now to
only pass arguments that might help in recovering from the exception or
provide information that is otherwise not easily accessible. file: [3305aefb9e] check-in: [3d16a30f41] user: js, branch: trunk, size: 17438 [annotate] [blame] [check-ins using] [diff]

Instead, a decreasing variable is used now. This should be safe, as
nothing else should be binding ports while homebrew code is running. The
only problem that could arise is when code manually binds to very high
port numbers.

This also solves the problem that getsockname() is not available on the
Wii, as we know the port now anyway.

And while we're at it, this also adds struct sockaddr_storage. file: [17c2cc5542] check-in: [ca113e0145] user: js, branch: trunk, size: 18148 [annotate] [blame] [check-ins using] [diff]

gethostbyname() is not required to accept an IP address and does not do
so on the Wii. file: [73d0c39dd0] check-in: [8faa16f249] user: js, branch: trunk, size: 17988 [annotate] [blame] [check-ins using] [diff]

Not functional yet due to bugs in the SDK.
Bugs found so far:

* Binding to port 0 fails instead of choosing a free port.
* gethostbyname() does not work for IPs.
* getsockname() is missing.
* struct sockaddr_storage is missing.

I have not decided yet whether I fix those bugs in the SDK (I already
implemented getsockname() and added struct sockaddr_stroage and it seems
to work) or if I work around them in ObjFW. This will mainly depend on
how cooperative the developers of the Wii SDK are. file: [4800955ec9] check-in: [ffb91daffe] user: js, branch: trunk, size: 17114 [annotate] [blame] [check-ins using] [diff]

It's safe to always import OFThread.h now as OFThread even offers
limited functionality when threads are unavailable. file: [4199e0da2a] check-in: [d7365be372] user: js, branch: trunk, size: 16445 [annotate] [blame] [check-ins using] [diff]

The only places where BOOL is left are those where they are required by
the ABI. file: [5920a82404] check-in: [c5ef582958] user: js, branch: trunk, size: 16332 [annotate] [blame] [check-ins using] [diff]

OFStreamObserver_select: Fix missing include.
OFTCPSocket: Fix missing rename. file: [e79d51f216] check-in: [a2a48d4d29] user: js, branch: trunk, size: 16247 [annotate] [blame] [check-ins using] [diff]

This reverts 5362941.

Even though *UsingEncoding: is what Foundation uses, *WithEncoding: is
more natural and is more consistent with the rest. Also, this means this
restores it to the API of 0.7. file: [4724c73932] check-in: [9ddd33a258] user: js, branch: trunk, size: 15860 [annotate] [blame] [check-ins using] [diff]

For connect(), this sets ai_flags to:
AI_NUMERICSERV | AI_ADDRCONFIG

For bind(), this sets ai_flags to:
AI_NUMERICSERV | AI_ADDRCONFIG | AI_PASSIVE file: [b609afae72] check-in: [541b33d978] user: js, branch: trunk, size: 15841 [annotate] [blame] [check-ins using] [diff]

This removes of_http_client_tls_socket_class in favor of having one
symbol for all places where TLS would be desired. file: [d55696187d] check-in: [1abe5a111f] user: js, branch: trunk, size: 15340 [annotate] [blame] [check-ins using] [diff]

This makes it possible to handle exceptions that occurred during async
I/O in the specified handler. file: [29f21ecbe7] check-in: [6343e04dcd] user: js, branch: trunk, size: 15674 [annotate] [blame] [check-ins using] [diff]

This makes it clear that multiple evaluation of parameters is possible. file: [0f410ad5c9] check-in: [1cb8fee5c3] user: js, branch: trunk, size: 15495 [annotate] [blame] [check-ins using] [diff]

Async connecting is still missing. file: [b7640aefbf] check-in: [4d9a641d87] user: js, branch: trunk, size: 13089 [annotate] [blame] [check-ins using] [diff]

For example, this is useful if you want to use
+[OFString stringWithContentsOfURL:] with a proxy or if you just want
to use anything that uses an OFTCPSocket with a proxy (even 3rd party
libraries). file: [59c9f9facb] check-in: [a3b3333128] user: js, branch: trunk, size: 11658 [annotate] [blame] [check-ins using] [diff]

This is safe as an "exception loop" can't happen, since if allocating
an exception fails, it throws an OFAllocFailedException which is
preallocated and can always be thrown.

So, the worst case would be that an autorelease of an exception fails,
triggering an OFOutOfMemoryException for which there is no memory,
resulting in an OFAllocFailedException to be thrown. file: [6c64c65615] check-in: [e1e7ffa903] user: js, branch: trunk, size: 10195 [annotate] [blame] [check-ins using] [diff]

Without this, the headers try to declare functions that use __block as
an argument name and thus fail to compile when using -fblocks. file: [5139608894] check-in: [59e52af26d] user: js, branch: 0.5, size: 9753 [annotate] [blame] [check-ins using] [diff]

Without this, the headers try to declare functions that use __block as
an argument name and thus fail to compile when using -fblocks. file: [d0efc140d4] check-in: [da2701ad14] user: js, branch: trunk, size: 9806 [annotate] [blame] [check-ins using] [diff]

Does compile with --disable-threads --disable-shared, but does not link
yet. file: [1565c3535f] check-in: [d230ac577d] user: js, branch: trunk, size: 9686 [annotate] [blame] [check-ins using] [diff]

SIGPIPE is quite useless and contraproductive as we throw an
OFWriteFailedException anyway. file: [94deb56f45] check-in: [1404d33024] user: js, branch: trunk, size: 9273 [annotate] [blame] [check-ins using] [diff]

ivars are now named like this: thisIsAnIVar. If an accessor exists for
the ivar, the ivar is named like the accessor. This is required for
@property to work. file: [dd20905993] check-in: [37db8f0fb3] user: js, branch: trunk, size: 9032 [annotate] [blame] [check-ins using] [diff]

Changesets:
* b95fcaa6d694
* 804c68d222b4
* da8cd738da3d
* 1109d5ce3419
* d03f5c1ca95d
* 2a7017722165
* 4fccdc79eeb7
* d228149fbc04
* 8782d412a4a6 file: [709f9ba347] check-in: [b91a8283fe] user: js, branch: 0.2, size: 7813 [annotate] [blame] [check-ins using] [diff]

Only use getaddrinfo() if it's thread-safe - it does not make much
sense to allow getaddrinfo() if it's not thread-safe, as then its
biggest advantage to gethostbyname() is gone.

Also, copy the result from gethostbyname() so we can release the lock
before connecting - this should greatly improve the speed when trying
to connect to multiple hosts without a thread-safe getaddrinfo(). file: [d088ca13d0] check-in: [1f7898f109] user: js, branch: trunk, size: 7681 [annotate] [blame] [check-ins using] [diff]

OFExceptions:
* OFNoMemException to OFOutOfMemoryException.
* OFMemNotPartOfObjException to OFMemoryNotPartOfObjectException.

OFObject:
* -[addItemToMemoryPool:] to -[addMemoryToPool:].
* -[allocWithSize:] to -[allocMemoryWithSize:].
* -[allocNItems:withSize] to -[allocMemoryForNItems:withSize:].
* -[resizeMem:toSize] to -[resizeMemory:toSize:].
* -[resizeMem:toNItems:withSize:] to
-[resizeMemoryToNItems:withSize:].
* -[freeMem] to -[freeMemory:].

OFString:
* -[urlencode] to -[urlEncodedString].
* -[urldecode] to -[urlDecodedString]. file: [3f2fb0c9c6] check-in: [033054ad75] user: js, branch: trunk, size: 6709 [annotate] [blame] [check-ins using] [diff]

getMemWithSize: -> allocWithSize:
getMemForNItems:ofSize: -> allocNItems:withSize:
resizeMem:toNItems:ofSize: -> resizeMem:toNItems:withSize: file: [59fbba1365] check-in: [e959fed010] user: js, branch: trunk, size: 6112 [annotate] [blame] [check-ins using] [diff]

Since we don't use Object anymore and define isa in OFObject, we can
rely on it. file: [9dddc5248f] check-in: [139591afe1] user: js, branch: trunk, size: 6126 [annotate] [blame] [check-ins using] [diff]