The message-size itself is part of the message which
stored in the ringbuffer. If the rinbuffer overflows
the message is misinterpreted -> segfault.
Choose a more conservative ring-buffer size and take
the requested LV2 size into account.
This is like cmp_nocase(), only that it doesn't use toupper(), tolower()
and therefore is agnostic of the current locale, and attempts to compare
strings in a UTF8-aware way (or falls back to ASCII if one of the
strings isn't UTF8-encoded).
17:29 < rgareus> las: ok. I'll make this translatable, but this warning will show up every time ardour starts (and scans for plugins) which can be annoying.
17:30 < las> rgareus: yeah, well we need to do black/whitelisting of plugins anyway so ...
This caused an export bug when:
a) normalizing
b) adding silence to end
c) having more than one channel
The sound was corrupted by repeating each jack frame as many times
as there were channels.
MTC and MIDI Clock port input handling was moved into the process/RT thread(s) during audioengine work, but MMC was
left orphaned. Add it to the port(s) handled by the MIDI UI thread.
Also, remove PortChange request from MidiUI because it has no meaning anymore
Unlike all the other jack backends which allow setting the number of inputs and
outputs with -i & -o, the dummy backend uses -P & -C for this. Make the jackd
command line use these options when the dummy backend is requested with a
specified input or output channel count.
The dummy jackd backend doesn't require a device to be specified, so much
of the error checking in get_jack_command_line_string() is irrelevant, if
not actively wrong, when the dummy backend is specified.
Only perform the checks if the chosen jack backend is not the dummy.
Outdated and fuzzy translations of MTC, MTC, LTC, etc caused
various issues (duplicate jack port names, exceptions, crashes).
Functionality should not be affected by translations (for now).