[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Add sleep builtin

From: Robert Elz
Subject: Re: Add sleep builtin
Date: Tue, 28 Aug 2018 01:54:41 +0700

    Date:        Mon, 27 Aug 2018 10:57:56 -0700
    From:        L A Walsh <address@hidden>
    Message-ID:  <address@hidden>

  | I certainly would not take the ever-changing POSIX spec as
  | a guide to what actually exists or is implemented in the field.
  | POSIX moved away from their original mission statement of being
  | "descriptive" to one that is "prescriptive" a long time ago,
  | with "prescriptive" being based on the wants and desires of
  | its current membership with little or no regard for usability
  | or compatibility.

I don't think that's true - certainly not as much as other (related)
standards (like those for C etc).   While there have been inventions
(unfortunate) mostly what is in POSIX is what is available in systems
in the field - if anything it goes too far in accomodating vartiations
by making all kinds of things unspecified just because there is a
system somewhere which does it differently than everyone else.
That is, when users might see different behaviour that the normal
on some system somewhere.

Howeven it most certainly is not a description of what you will get
from Linux or anything GNU.


ps: for sleep - things have always been somewhat vague, as most
systems run with clock ticks, at some fixed frequency - on those it
is impossible to get accurate short sleeps.   The "sleep 1 means
until the next tick" was an ancient issue, fixed a very very long time
ago I think.   But whan a process sleeps, it is giving up the CPU to
some other process - there's no guarantee that it is going to be
made runnable again instantly that its sleep period is over, and even
if it is, no guarantee that the schedueller will pick it  to run next
(there might be dozens of other processes also ready to run.)

reply via email to

[Prev in Thread] Current Thread [Next in Thread]