get rid of crap; run audio/aacenc ourselves with a small raw pcm buffer
readme: update for mixfs
add -A option to specify audio offset (-200 for audio delayed by 200ms)
readme: aacenc: use -b
adts: fix size parsing
audio: don't write random crap (GASpecificConfig that makes no sense)
allow streaming to several URLs at the same time
move debug global to util.h
use a function to convert timestamps
better "disabled sync" fix
update readme
if initial timestamps aren't set, disable sync
don't flush on every audio/video packet
protect against audio overruns
adts: skip 'nsec' packet
filesize → fileSize
sync audio with video based on timestamps coming from the streams
ivf: fix wrong timestamp values
ivf: set biobuf
ivf: start with timestamp 0
readme: add audio information
put audio reading into a separate proc
readme: add audio example
audio works, update the readme
fix audio timestamps
readme: add ingests url
rework closing
fix misleading error message
response may come on a different chunk stream - compare the code (if set) in the response
estrndup: set malloc tag
remove -v option: always read from stdin
update the readme
use stdin if -v option not used
fix video stream packing and url parsing (can stream to Twitch directly now)
send video data
more logic (not working yet)
stream publish
[Aa]mf0 → [Aa]₀
amf: make sure type is set when parsing
forgot another eof
createStream
set object encoding explicitly to AMF0
set out chunk size to 4096 after connecting
separate in/out chunk size
rtmpdial: remove unused args
connect callback: make sure it's "_result"
move msg pretty-printing elsewhere
add amf0 stuff
rewrite according to the rtmp spec
parse server/client bw
parse control packets
loop: validate (and debug print) chunk size
implement proper amf parsing & pretty printing; rpc
print packet contents on Invoke
rework send/recv
more wip logic
amf: use short strings when possible, apparently rmtpsrv doesn't like the long ones at first
more rtmp logic
split ivf logic out
first, nothing to see yet