> How is this change related to your patch description?
> -#      encoder         "vorbis"                # optional, vorbis or lame
> +#      encoder         "vorbis"                # optional, vorbis or lame or 
> any encoder (use mpd --verbose to get a list)

It's not. The comment is wrong, and is 2 lines above my change. I
fixed it since I was here, can revert it, but I'm not going to make a
patch for this one (false) line.

> Still no documentation, and still no reason why this feature is
> useful.  I don't like it at all, and the commit message needs to
> convince me.
Actually, it's useful to clients but not only. In my use case, MPD is
on a headless server. Without it, it's not possible to receive this
stream in Javascript context due to Cross Origin protection in
browsers. That prevents using HTTPD plugin as a audio source for a web
page, unless you have the Allow Origin header (ditto for
Icecast/Shoutcast mode).

The usual workaround, is to serve the music library out of MPD via the
same port as the webserver, but it's bad because:
1) It duplicates the effort (what is the point of a music daemon if we
have to use another software doing 80% of the same work ?)
2) No synchronization of "meta action". If I create a playlist in MPD
it's not in the second software so I have to redo it.
3) Need to redo in Javascript what MPD does very well (like
crossfading, replay gain, mixramp etc...)
4) Some feature will never be doable in Javascript/HTML, like input
streaming from any sources.

Also, the feature can be used for reverse proxy mode, identifying the
source MPD when numerous are being used.

mpd-devel mailing list

Reply via email to