timeout: Duration parsing does not support all float formats (e.g. hex)

[drinkcat]

GNU coreutils supports things like hex formatting, and arbitrary case for infinity parsing:

timeout 0x1.1 cat
timeout 0x1.1p-1 cat
timeout -k 0x10.1 1 cat
timeout INF cat
timeout InF cat

Not a huge deal, probably a pretty rare edge case that I doubt anybody cares about, but it seems worth it to use generic float parsing functions in uucore/format after they are fixed.

[drinkcat]

One interesting thing is that GNU coreutils rounds down very small values to 0, which has a totally different meaning (0 means no timeout, a small value should be a... very short timeout).

Those 3 exit immediately:

timeout 0.0001 cat
timeout 1e-100 cat
timeout 1e-300 cat

But this never exits:

timeout 1e-3000 cat

As if we had typed:

timeout 0 cat

Similarly to the discussion in #7641, I wouldn't find it shocking if we treated 1e-3000 as a very short timeout (and not 0).

[cakebaker]

Closed by mistake, reopen it.

[tertsdiepraam]

But this never exits:

timeout 1e-3000 cat

Might be worth sending a bug report about! I can't find anything about it on their mailing lists.

[drinkcat]

But this never exits:

timeout 1e-3000 cat

Might be worth sending a bug report about! I can't find anything about it on their mailing lists.

Good point. https://debbugs.gnu.org/cgi/bugreport.cgi?bug=77535

Also interesting to see those arguments about 0 meaning never: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=9102 (it would be better to use inf IMHO...)