When launching catnip on Pipewire 1.4.0, catnip gives the following error:
```shell
$ catnip -d alsa_output.platform-es8388-sound.stereo-fallback -sf 0 -ct 0 -dt 1
sndfile: failed to open audio file "-": Error : this file format does not support pipe write.
error: open failed: Input/output error
```
Version information:
### Pipewire
```shell
$ pipewire --version
pipewire
Compiled with libpipewire 1.4.0
Linked with libpipewire 1.4.0
```
### Catnip
```shell
$ catnip --version
Version: 1.8.5
```
ooof. this is interesting, we use pw-cat to get the stream out of pipewire.
@Danct12 you did not pass a backend flag, so catnip will use the deafult backend.
can you ensure you are using the pipewire backend, passing `-b pipewire` to the catnip command? you can also run `catnip list-backends` or `catnip lb` and it should have a star next to the default backend.
if that does not work, can you try running:
`pw-cat --record --format f32 --rate 44100 --latency 1024 --channels 2 --target auto --quality 0 --media-category Capture --media-role DSP --properties '{"application.name": "pw-cat"}' -`
it should start spewing the stream data of your default input device. if not, can you confirm the error message returned?
same stuff happens to me but without the random garbage binary characters...???
@saxophone-dev @Danct12
what is the output of `pw-cat --help`?
this may be an issue with pw-cat, but lets rule out the possibility of a new parameter flag that enables printing to STOUT.
for context, this is my output:
```
> pw-cat --version
pw-cat
Compiled with libpipewire 1.2.7
Linked with libpipewire 1.2.7
> pw-cat --help
pw-cat [options] [<file>|-]
-h, --help Show this help
--version Show version
-v, --verbose Enable verbose operations
-R, --remote Remote daemon name
--media-type Set media type (default Audio)
--media-category Set media category (default Playback)
--media-role Set media role (default Music)
--target Set node target serial or name (default auto)
0 means don't link
--latency Set node latency (default 100ms)
Xunit (unit = s, ms, us, ns)
or direct samples (256)
the rate is the one of the source file
-P --properties Set node properties
--rate Sample rate (req. for rec) (default 48000)
--channels Number of channels (req. for rec) (default 2)
--channel-map Channel map
one of: "stereo", "surround-51",... or
comma separated list of channel names: eg. "FL,FR"
--format Sample format (ulaw|alaw|u8|s8|s16|s32|f32|f64) (req. for rec) (default s16)
--volume Stream volume 0-1.0 (default 1.000)
-q --quality Resampler quality (0 - 15) (default 4)
-p, --playback Playback mode
-r, --record Recording mode
-m, --midi Midi mode
-d, --dsd DSD mode
-o, --encoded Encoded mode
```
edit --
the [1.4.0 release](https://gitlab.freedesktop.org/pipewire/pipewire/-/releases/1.4.0) explicitly says pw-cat "now support[s] stdin and stdout"? unsure what that means.
@noriah
```shell-session
$ pw-cat --version
pw-cat
Compiled with libpipewire 1.4.1
Linked with libpipewire 1.4.1
$ pw-cat --help
pw-cat [options] [<file>|-]
-h, --help Show this help
--version Show version
-v, --verbose Enable verbose operations
-R, --remote Remote daemon name
--media-type Set media type (default Audio)
--media-category Set media category (default Playback)
--media-role Set media role (default Music)
--target Set node target serial or name (default auto)
0 means don't link
--latency Set node latency (default 100ms)
Xunit (unit = s, ms, us, ns)
or direct samples (256)
the rate is the one of the source file
-P --properties Set node properties
--rate Sample rate (req. for rec) (default 48000)
--channels Number of channels (req. for rec) (default 2)
--channel-map Channel map
one of: "stereo", "surround-51",... or
comma separated list of channel names: eg. "FL,FR"
--format Sample format (ulaw|alaw|u8|s8|s16|s32|f32|f64) (req. for rec) (default s16)
--volume Stream volume 0-1.0 (default 1.000)
-q --quality Resampler quality (0 - 15) (default 4)
-a, --raw RAW mode
-p, --playback Playback mode
-r, --record Recording mode
-m, --midi Midi mode
-d, --dsd DSD mode
```
@noriah
```
$ pw-cat --help
pw-cat [options] [<file>|-]
-h, --help Show this help
--version Show version
-v, --verbose Enable verbose operations
-R, --remote Remote daemon name
--media-type Set media type (default Audio)
--media-category Set media category (default Playback)
--media-role Set media role (default Music)
--target Set node target serial or name (default auto)
0 means don't link
--latency Set node latency (default 100ms)
Xunit (unit = s, ms, us, ns)
or direct samples (256)
the rate is the one of the source file
-P --properties Set node properties
--rate Sample rate (req. for rec) (default 48000)
--channels Number of channels (req. for rec) (default 2)
--channel-map Channel map
one of: "stereo", "surround-51",... or
comma separated list of channel names: eg. "FL,FR"
--format Sample format (ulaw|alaw|u8|s8|s16|s32|f32|f64) (req. for rec) (default s16)
--volume Stream volume 0-1.0 (default 1.000)
-q --quality Resampler quality (0 - 15) (default 4)
-a, --raw RAW mode
-p, --playback Playback mode
-r, --record Recording mode
-m, --midi Midi mode
-d, --dsd DSD mode
```
and version:
```
$ pw-cat --version
pw-cat
Compiled with libpipewire 1.4.1
Linked with libpipewire 1.4.1
```
@Danct12, @saxophone-dev can you test the same command as https://github.com/noriah/catnip/issues/24#issuecomment-2746031972, with `-a` or `--raw` before the `-`?
example: `pw-cat --record --format f32 --rate 44100 --latency 1024 --channels 2 --target auto --quality 0 --media-category Capture --media-role DSP --properties '{"application.name": "pw-cat"}' --raw -`