| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
| |
This is meant to exactly mimic the terminal UI, just without
the restriction to ugly monospace fonts.
The Curses implementation has been reworked, and may have become
more computationally expensive. In general, though, the codebase
has been significantly cleaned up.
|
| | |
|
| | |
|
| |
|
|
| |
"z" stands for VIM's "zz".
|
| | |
|
| |
|
|
| |
Make it unanchored, as well as case-insensitive.
|
| | |
|
| |
|
|
| |
The improvement is very minor in character.
|
| | |
|
| | |
|
| |
|
|
|
| |
Instead, assume that the whole previously selected range
has been removed, and try to go after or before it accordingly.
|
| | |
|
| |
|
|
| |
Also, make it apparent which value comes from where.
|
| | |
|
| |
|
|
|
|
|
| |
Mostly because I wanted to nest preprocessing.
This makes the build more complex and slightly less portable,
but the code itself is much cleaner.
|
| |
|
|
|
|
|
|
| |
I know, son, it might be hard to accept,
but you're imported. Your true parents are wmstatus
and paswitch, from the desktop-tools family.
Also, fix unnecessary linking of optional dependencies.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
The updater assumed the terminal cursor was invisible.
|
| |
|
|
|
|
|
|
| |
I was tired of using `mpv --no-video`, this is a bit better.
It's all rather quirky, but very little code is involved.
I've added a few related TODO entries.
|
| |
|
|
|
|
|
| |
Also, do not set up the spectrum visualiser before a password is sent.
It would look a bit weird to have it run but display "Disconnected",
even though technically, it would probably work.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Don't just abort() on failures, print a proper error message.
Also, set up ncurses as late as possible. This should be alright wrt.
signal handlers according to ncurses code, as well as XSI:
> Curses implementations may provide for special handling of
> the SIGINT, SIGQUIT and SIGTSTP signals if their disposition
> is SIG_DFL at the time initscr is called ...
termo blocks job control, so SIGTSTP is not a concern at all.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
nncmpp CPU usage went from 2 to 1.7 percent, a 15% improvement.
Sort of worth it, given that it's a constant load.
The assembly certainly looks nicer.
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
The statement can be eliminated, then it suggests braces.
|
| |
|
|
|
|
| |
This is really more of a demo. It's doable, just rather ugly.
It would deserve some further tuning, if anyone cared enough.
|
| |
|
|
| |
We might even depend on termo now more than is stated.
|
| |
|
|
| |
Should help prevent accidents in other windows.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
Ideally we'd make columns configurable, which isn't trivial.
This brings the "Current" and "Library" tabs closer together.
Closes #2
|
| |
|
|
| |
Instead of merging the fields into other items.
|
| |
|
|
|
|
| |
The `struct strv` was clunky, it's better to store items
directly in the format we use for all processing.
The additional memory cost is negligible.
|
| |
|
|
|
|
|
|
| |
It is less distracting.
Also use mpd_read_time() and load "duration" as well.
This value isn't rounded to whole seconds, so we load
it before "time" as a fail-safe measure.
|
| |
|
|
|
|
| |
- reject negative values, which strtoul() happily accepts
- deal with an arbitrary number of decimal digits
- don't return milliseconds when we fail to parse seconds
|
| |
|
|
| |
Taken from Windows Explorer, which abandoned the Backspace binding.
|
| |
|
|
| |
No functional change.
|
| |
|
|
|
|
|
|
| |
While still avoiding busy loops.
It works well enough to enable this by default.
Closes #1
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
Playback may sometimes stall but it won't produce any events.
This popular workaround likes to jump around, though.
It might be a good idea to use some kind of hybrid approach.
Therefore this is disabled by default so far.
Updates #1
|
| |
|
|
| |
Took time to read. Also fix an invalid comment.
|
| | |
|
| |
|
|
|
|
| |
Might be an issue specific to my bbc-on-ice, since we're not asking
for SHOUTcast by including "Icy-MetaData: 1" in request headers
but the proxy always outputs an "ICY 200 OK" header.
|
| |
|
|
| |
The behaviour is defined by the standard.
|