**What are we doing?**
We are moving towards stateless callbacks. This means that when registering a
callback, you no longer pass a user data pointer. Instead, you pass a user data
pointer to tox_iterate. This pointer is threaded through the code, passed to
each callback. The callback can modify the data pointed at. An extra indirection
will be needed if the pointer itself can change.
**Why?**
Currently, callbacks are registered with a user data pointer. This means the
library has N pointers for N different callbacks. These pointers need to be
managed by the client code. Managing the lifetime of the pointee can be
difficult. In C++, it takes special effort to ensure that the lifetime of user
data extends at least beyond the lifetime of the Tox instance. For other
languages, the situation is much worse. Java and other garbage collected
languages may move objects in memory, so the pointers are not stable. Tox4j goes
through a lot of effort to make the Java/Scala user experience a pleasant one by
keeping a global array of Tox+userdata on the C++ side, and communicating via
protobufs. A Haskell FFI would have to do similarly complex tricks.
Stateless callbacks ensure that a user data pointer only needs to live during a
single function call. This means that the user code (or language runtime) can
move the data around at will, as long as it sets the new location in the
callback.
**How?**
We are doing this change one callback at a time. After each callback, we ensure
that everything still works as expected. This means the toxcore change will
require 15 Pull Requests.
Also, fix the hstox build that was taking half an hour. It now takes 5 minutes.
Also, perform distcheck on travis to ensure that make dist works. It's not
actually failing the build at the moment due to broken tests.
> increased the timeout for TCP tests because per @irungentoo the network on Travis-CI can be slow sometimes
> allowed groupchats test to restart on error until timeout This had to be done because current groupchats are fundamentally broken and 3/5 times they'll 'net-split' on connect
>> Drop group chat tests, add comment to the reason
> added some debugging information to TCP tests, and a #define to force IPV6 (Travis-CI only uses IPv4 on their containers) and decreased the itr interval
> Went crazy with timeouts for Tox network stuff on Travis. Tests on TCP will still randomly fail due to timeouts. I can't reproduce on any local system. So again per @irungentoo, Travis is slow, let's offer it a short bus.
Upon my own decision, the two tox_encryped_new convenience functions were removed due to basically needing two different sets of error codes.
At iphydf's suggestion the corresponding tox_get_encrypted_savedata convenience functions were removed as well.
A couple of minor reasons, combined warrant a PR imo:
a) fileChunkRequested is a better signal name than fileRequestChunkReceived, and I don't want to break consistency by reordering words for just this signal
b) "request chunk" is parsed by English speakers as a verb-object combination,
implying sending the request, not receiving, whereas "chunk requested" is
parsed (more correctly) as an adjective-noun combo (in particular, request is
a noun not a verb), and thus reads far more like "hey heads up we just got a request"
For instance some tests/testing code had some callbacks to *receive* chunk requests, and they were called "tox_file_request_chunk"... to receive a chunk, not request it. Now they're called "tox_file_chunk_request".
So yeah...
This function can be used to seek an incoming file tranfer right
before accepting it.
It is meant to be used to resume incomplete file tranfers by clients.
file_id is a 32byte identifier that can be used by users to identify
file tranfers across core/client restarts in order to resume broken
file tranfers.
In avatar tranfers it corresponds to the hash of the avatar.
Added tox_file_get_file_id() function to api to obtain the file_id
of an ongoing file transfer.
If not set, core will generate a random one.
Also, all data now has the magic number prepended. This is incompatible
for all but the save/load functions, but I think nothing besides the one
experimental qTox branch used any of those, which is why I feel confident
about the change.
An index for replacement candidate is searched
in one lookup cycle for all types (bad, possibly bad, good).
Sorting of items has been removed (sorting logic can be substituted
by a maximum search).
correctly.
This should fix an issue that happened when both clients got
disconnected when the file was almost finished sending. The sender
would show that the file had been sent successfully when it had not.
See the modifications to tox.h
now described as the maximum width and height of images.
This is to work around what appears to be a bug in libvpx where the
resolution of the stream can be decreased but increasing it above
its originally set value introduces memory corruption.
Moved Bunch of functions from net_crypto to crypto_core.
decrypt_data_fast and decrypt_data_symmetric were the same thing
therefore, removed decrypt_data_fast.
Replaced all the crypto_secretbox_* defines with the equivalent
crypto_box_* one.
New define: crypto_box_KEYBYTES that is equal to
crypto_box_BEFORENMBYTES.
Removed a bunch of useless space from various structs.
pack_nodes() and unpack_nodes() are now used to transmit lists of
nodes for the DHT and onion parts. They provide a way to transmit
both ipv6 and ipv4 nodes in the same packet in a portable way that
takes as little space as possible without compression.
Using pack_nodes, merged the send nodes and send nodes ipv6 packets
into one packet greatly reducing the risk of amplification attacks.
Main changes:
1. Strings no longer need to be NULL terminated.
2. tox_get_friend_id is now named tox_get_friend_number.
3. The friend request callback function is now (Tox *tox, uint8_t *,
uint8_t *, uint16_t, void *), the Tox object pointer has been added to
it.
oniondata_registerhandler(...) can be used to register different packet
types to handle them and send_onion_data(...) can be used to send these
packets.
id_hash() was not at all working as expected for very small bucket size (when (size / 4) was zero). Simplified to be trivially correct.
Also added a used flag on adding an entry, which is set by callers if they have that association in active use. Those get priority over unused entries on collision.
Fleshed out test to be at least elementary useful.
Each group chat now uses an own, small assoc (80 entries).
To allow the actual waiting to run without any locking, split it into preparing the data it uses and the execution of the wait.
The caller must provide with the buffer to store whatever data it requires to wait.
Completely eliminates any reliance on the existence of anything but that data in the actual wait routine.
Also fixed a few argument type warnings inside LOGGING.
util.c:
- fix in empty section at the end of the state, showed as bug when having an empty name
DHT.c:
- fix in saving less data than originally announced, showed as bug when not having reached any clients ever (no clients or only with timestamp of zero)
auto_test/crypto_test.c:
- test "endtoend": increased timeout from default ~5s to 15s, making it pass on slower systems
- signed/unsiged warning fixed by cast
- tox_bootstrap_ex(), DHT_bootstrap_ex() renamed to tox_bootstrap_from_address(), DHT_bootstrap_from_address()
- (handle_)sendnodes_ex() renamed to (handle_)sendnodes_ipv6()
- only sending sendnodes_ipv6() if we're actually IPv6 enabled
- changed comments to conform better
nTox.c, Messenger_text.c, DHT_test.c, DHT_bootstrap.c:
- fallout from *_ex() to *_from_address()
DHT_bootstrap.c:
- corrected a potentially wrong info message
util.c:
- fixed logfile name: now (funcptr) => now() (number)
network.c:
- addead comment about the necessity of bind() to succeed
auto_test/messenger_test.c:
- defaulting ipv6enabled to TOX_ENABLE_IPV6_DEFAULT
LAN_discovery.c:
- slight cleanup and comments for clarity
- initialisation: argument added to enable/disable ipv6 as socket
Messenger_test.c:
- initialisation: ipv4 hardcoded for now
- delegating IP resolution to DHT_bootstrap_ex()
By default libsodium is used. Only if --enable-nacl is specified, then
nacl will be used instead of libsodium.
Pass locations of nacl headers and libraries by using the following
options:
--with-nacl-headers=/home/me/somewhere/nacl-20110221/build/469/include/amd64/
--with-nacl-libs=/home/me/somewhere/nacl-20110221/build/469/lib/amd64/
This update makes sure that the build still works with automake prior to
1.12 and at the same time does not give any warnings or errors with
automake 1.14
supported options:
--with-dependency-search=DIR will tell configure to look for various
dependencies in DIR/include and DIR/lib
Alternatively you can also specify libsodium header and libs location
with --with-libsodium-headers and --with-libsodium-libs if it is
installed elsewhere.
Ncurses and libconfig are handled via the default pkg-config way, see
./configure --help=short for detailed information.
The tox library is compiled as libtoxcore in shared and static variants,
public headers are installed to ${prefix}/include/tox
A pkg-config libtoxcore.pc configuration file is provided.
Use ./configure --help for a full list of configure options or
./configure --help=short for the options that I added.
To generate the configure script after pulling from git use:
autoreconf -i
To generate a release tarball use:
make dist
Unit tests are handled by the libcheck library integration that is provided
by autotools, use:
make check
to compile and run the tests.
Unit tests are currently optional, i.e. - if the check library is not
found on the system, then tests will be disabled. Same goes for nTox and
DHT bootstrap daemon - they will be enabled or disabled depending on the
availability of ncurses (for nTox) or libconfig (for DHT bootstrap
daemon).
The above can be also tuned by:
--enable-tests / --disable-tests
--enable-ntox / --disable-ntox
--enable-dht-bootstrap-daemon / --disable-dht-bootstrap-daemon
While doing the checks configure might generate "core" files and will
then try to remove them. Having a "core" directory generates an error
while runing the configure script.
There's no workaround but to rename the core directory.
Moves static state out of Messenger.c and into a Messenger struct
Purely stylistic, no functional changes were made.
This commit also changed all the callers of Messenger as they now have
to pass an instance of the Messenger struct to messenger functions.
Also removed some uses of the 'static' keyword at the beginning of
function definitions when the function was already declared static, as
these caused gcc to whine.
