On Tue, Jul 23, 2024 at 13:38:48 -0500, David Wright wrote: > On Tue 23 Jul 2024 at 09:31:36 (-0400), Greg Wooledge wrote: > > On Tue, Jul 23, 2024 at 23:22:52 +1000, Keith Bainbridge wrote: > > > The day# in my command prompt increments when I start in the morning. > > > Maybe I need to press enter. > > > > That makes it sound like you're setting the YEAR et al. variables in the > > PROMPT_COMMAND variable. > > > > If that's the case, it's *less* wrong, but only a little bit. You still > > have the issue that the date might change while you're sitting at a > > stale shell prompt from yesterday, with stale date/time variables. > > Actually, that is what I do want (the time—the date not so much). > That tells you when the last command finished. Press Return before > you start a command and you get some idea of how long it takes > to run. For the current time, press Return. > > I think putting the time±date in one's prompt is pretty popular,
Sure. That's a completely different problem from what Keith is describing, though. Putting date/time expansions into your PS1 (prompt) causes the system time to be retrieved when the prompt is displayed. The date/time strings are rendered and displayed, and then not remembered by the shell. What Keith is doing is storing components of the date/time in variables at shell startup, and then using them at some unspecified point in the future to create a directory. This causes issues when the shell survives through a day transition (midnight). Keith's workflow might work well if he rigorously logs out (or at least closes all shells) every night before midnight, and logs in (or re-opens his shells) in the morning. But even then, it's still a questionable implementation. Anything that relies on a human being to remember to do something is eventually going to fail. A safer implementation would retrieve the current date at the moment when it's needed. Your PS1 string does this.
