This is mainly to fix a race when river is *not* running; sometimes we
ended up allocating particles for N seats in content(), but then when
iterating the seats, run() had destroyed all, or some of the seats,
causing us to feed NULL pointers to dynlist, which crashed.
* Verify int/float/bool values are that, and nothing else
* Verify tag ranges are integers
* Verify a range tag value is inside its range
* Don’t allow anything but false|true for booleans
User is expected to send ‘false’ or ‘true’. But we were parsing the
value using `strtol()`. This caused all bools to be false, since
`strtol()` would always return 0.
This module exec’s a script (or binary), specified by the ‘path’
attribute in the configuration.
It then reads tags from the script’s stdout.
The format of the output is:
tag|type|value
tag|type|value
<empty line>
I.e. the script writes N tags followed by an empty line. This
constitutes a transaction.
When a new transaction is received, its tags replaces *all* previous
tags.
We overloaded enum mpd_state to add an “offline” state. Don’t do
this. We can detect the offline state by checking if our connection
object is NULL.
Closes#16
The ‘river’ module depends on generated protocol sources. These are
normally provided, indirectly, by the Wayland bar backend. But when
building plugins as shared modules, the plugins no longer depend on
the bar, and thus ‘river’ fails to build.
Fix this by doing two things:
* **Remove** the ‘river’ protocol from the ‘generic’ Wayland protocols
* Explicitly add both the generic and the ‘river’ specific protocols
to the ‘river’ module.
When we detect an error in yml_parser_parse(), we handle that
specifically, and then jump to the generic error handler.
The generic error handler overwrites the previously formatted error
before cleaning up and returning.
This meant that a) the actual error message was lost, and replaced
with a generic “unknown error”, and b) the dynamically allocated error
string buffer was leaked.
