summaryrefslogtreecommitdiff
path: root/ratox.c
AgeCommit message (Collapse)Author
2017-03-02Always append FCALL_IN when in a callz3bra
2017-01-13Whitespace fix.pranomostro
2017-01-13Remove FOLDER in enum.pranomostro
2017-01-13Removed unneeded whitespace.pranomostro
2016-12-05Fix typo in error message, update nodes.hpranomostro
2016-12-05Fix possible NULL memcpy() after allocating data.pranomostro
2016-12-02Check data for being NULL instead of intermediate, we did that already.pranomostro
2016-12-01Remove the return value from friendcreate()pranomostro
It is not used anywhere and it is unlikely that this will change anytime.
2016-11-25Make ringing delay checks simplerz3bra
2016-11-25Discard sending of the avatar picturesz3bra
2016-11-25Cancel calls ringing longer than RINGINGDELAYz3bra
2016-11-25Hang up only when we're in TRANSMITTING modez3bra
When f->av.state equals TRANSMITTING, it means that we're neither receiving, nor sending data, thus, we can cancel the call. This doesn't apply to RINGING mode because this state is possible when someone calls us, and we didn't answer yet.
2016-11-25Fix accepting friend requestsz3bra
Actually passes the friend number returned by tox_friend_add_norequest() to the friendcreate() function.
2016-11-24Fix tests against friend call statez3bra
2016-11-24Let the caller hang up while call is ringingz3bra
2016-11-24Allocate resources for outgoing call when neededz3bra
If we're not initiating the call, we might be in TRANSMITTING mode even if call_in hasn't been openned by the user. In this case, we shouldn't send any data. sednfriendcalldata() should only be called when resources are allocated for an OUTGOING call, which is triggered when the friend accepts audio.
2016-11-23Make sending only a newline succeed (patch by z3bra).pranomostro
2016-11-23Adjust the types of a few variables where it makes sense, mostly in regard ↵pranomostro
to comparisons between signed and unsigned integers.
2016-11-22Fix issue that prevents correct sending of INCOMPLETE buffers.pranomostro
f->av.state & INCOMPLETE was evaluated as 8, which made the third test always false.
2016-11-22Use ternary to prevent passing negative size to fiforead()z3bra
2016-11-22Remove noisy logmsg() / Fix grammarz3bra
2016-11-21Discard buffer if size is bigger than expected sizez3bra
2016-11-21Make sure call is only canceled oncez3bra
Upon receiving a FINISHED state, the client should consider the call over, and free its local variable. It should NOT try to send a CANCEL signal, as it could try to cancel the call twice, resulting in a double free() that can crash the core.
2016-11-19Reduce time between sample sendingz3bra
Start sending the next audio frame before the previous frame is done sending. This is for preventing the peer to receive "blanks" in the audio channel.
2016-11-19Get audio call to work properlyz3bra
This commit puts back in place the settings that were previously set regarding audio encoding, preparation of the frame, buffer and so on. These settings used to work great and I modified them without having any clue of how it was working. Now I'm done playing, and it works properly. Hooray.
2016-11-18Fixed issue preventing the user to answer callsz3bra
2016-11-18Update style for if() conditionsz3bra
2016-11-18Update audio parameters for the new APIz3bra
2016-11-18Add a RINGING state for callsz3bra
This helps knowing that a call has been initiated, but not answered yet. We need it because otherwise the state of the FCALL_OUT would not even be evaluated.
2016-11-14Revert "Compile against local toxcore submodule"z3bra
This reverts commit a1d2b63e6af32685f5828c302a0d3edeeb48090f. Ratox now compiles against the latest version of the API, so there is no need for using a local submodule anymore.
2016-11-10Updated nodes.h, small fixes in loop().pranomostro
2016-11-10Retrieve/Send nospam value as little-endianz3bra
For readiness, nospam value is written in big endian in nospam/out. toxcore expect receiving it as little-endian, so we convert it upon calling the nospam get/setter.
2016-11-10Remove useless call to FD_CLR()z3bra
This snippet was introduced when the client could send too much data to the core for the transfer. The new toxcore API introduces a callback triggered when the core is ready to send more data. It means we don't need to keep track of a local buffer anymore, and simply read from the file_in FIFO when the core is ready.
2016-11-09Fix nospam value endiannessz3bra
2016-11-08Ensure calls are answered properlyz3bra
2016-11-08Get rid of 'chunksz' attribute in transfer structz3bra
2016-11-08Handle file transfer in multiple chunksz3bra
2016-11-08Various cases of code cleanup:pranomostro
-Fixed memory leak in datasave(). -Made some types more clean, such as using size_t instead of uint32_t and size_t instead of int. -Removed unneccessary length checks
2016-11-08Accomodate file transfer callbacks to the new APIz3bra
2016-11-08Mark transfer as complete as soon as chunk len is 0z3bra
2016-11-08Fix argument passing order of tox_pass_decrypt.pranomostro
2016-11-08Pass file number to tox_file_control()z3bra
2016-11-08Monitor friends connected either via TCP or UDPz3bra
2016-11-08Convert ID string using TOX_PUBLIC_KEY_SIZEz3bra
2016-11-08Use av.state to check the current call statez3bra
2016-11-08Prevent memcpy'ing to a non-allocated memory areaz3bra
2016-11-08Change cbcalldata() to match new APIz3bra
2016-11-08Merge all call state changes in cbcallstate()z3bra
2016-11-08Added cbcallstate to deal with call changesz3bra
2016-11-08Setup callback to answer audio callsz3bra