Earnestly

@0X1A If the respective `XDG_*` variables are not set, they should fallback to the appropriate defaults as defined by the specification. The unfortunate legacy behaviour cargo needlessly introduced will likely...

Blegh, nevermind. Now there's a huge spike of cpu usage about 1 minute in and setting different timeouts isn't a solution. At this point I don't know what could be...

This seems related to the `nanosleep()` returning earlier (EINTR) in the `delay()` function. If nanosleep is interupted it will return early with a return value of `-1`. Since you don't...

As a follow up, I can manually set `working` to `0` in gdb and replicate this behaviour. The program never recovers from this and the cursor stays hidden indefinately.

Here is an untested potential fix for this, it basically just continues after being interupted. It does not handle `EINVAL`. It may be better to just crash instead of setting...

Hm, this may not be the cause as I suspected, although it does fix one error case. It seems to get stuck in `select()` sometimes and never returns.

This might be a `CANTFIX` issue. Either way here's some gdb output after catching it fail: ``` Attaching to process 26362 Reading symbols from /usr/bin/hhpc...done. Reading symbols from /usr/lib/libX11.so.6...(no debugging...

Okay, well I seem to have "fixed" this by simply adding a timeout on the `select()`. This is not really very satisfactory. ``` diff diff --git a/hhpc.c b/hhpc.c index f106f7f..84f6cc0...

That commit appears to fix both the demo here and the real example which segfaulted.

In my case I have `1 Feb 2022 12:00:00 GMT` where `GMT` ultimately means `UTC` and so both could be eschewed. But your suggestion of `--from-zone` is probably more appropriate,...