Unfortunately the grammar of the --config parameter is a bit of a
football right now; previously you were allowed to specify it anywhere
because there wasn't much in the way of a CLI, then when subcommands
were added it became required to specify it after the subcommand.
This change makes it so that it must be specified *before* the
subcommand. Ideally it would just be required *somewhere* in the
command line but clap doesn't support "global" for required arguments
(and it is indeed required).
On the whole, I'm okay with the change - the net win of having a unified
config struct makes up for the ergonomics hit in my opinion, and it
unlocks some further refactoring I will do next.
This offers more flexibility, as a camera system admin only
needs a DNS assignment rather than DHCP. The syntax is still
as expected:
address = "cameras.example.com:9000"
This is especially handy in more complex topologies, where one
or more specified "camera" may actually be pointed at a system
providing a tunnel through to multiple cameras; names make it
easier to identify said gateway machine.
All IPs that the given name resolves to are tried so that we
correctly handle, e.g., a name with both AAAA and A records.
Note that addresses are no longer validated up front, but instead
simply get resolved when the connection attempt is made.
Name resolution will be aborted after ~60s of failure.
* Add queues to the gstreamer launch
* Add audio to the stream
* Removing format option in prep for auto detect
* Adding format detect from headers
* Set format in gstreamer launch depending on video magic bytes
* Rename MaybeAppSrcs to reflect new role of managing gst outputs
* Auto detect format
* Remove all setting of stream format
* Re-add missing media_format caused by rebase
* Add audio format auto detect too
* Use more appropiate adpcm in gst
* Combined gst and media_packet format enums into a single enum StreamFormat
* Reduce adpcm header size
* Disable adpcm output
* Add oki adpcm native rust support
* Maintain state between packets for adpcm
* Use bytes in adpcm stream to determine adpcm decoding state
* Clean up adpcm code
* Use raw rtp payload for the adpcm stream
* Bug fix to treat adpcm and aac seperatly in the match
* Remove redundent second clamp of the step index in adpcm.
* ADPCM format changes.
This is a squashed commit of changes made during live testing of ADPCM
format.
Summary: Using DVI4 for the ADPCM, with our own decoder written in rust.
* Changes to comments and IMA max range in the adpcm decoder
* ADPCM magic contains block size use it for chunking the adpcm data
* Use error logs rather than asserts in adpcm decode
* Rustfmt and clippy styling
* Use static slice for adpcm step and changes arrays
* Bounds and error checks in ADPCM
* Use i/u32 in adpcm
* Rust fmted