automake-patches
[Top][All Lists]
Advanced

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

Re: tracing from aclocal


From: Paul Eggert
Subject: Re: tracing from aclocal
Date: 18 Aug 2003 14:04:24 -0700
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.3

Akim Demaille <address@hidden> writes:

> Is this really all that can do a modern system?  A one second accuracy
> on time stamps?

Things are improving a bit in this area, though not as much as I'd like.

Sun patch 109933-02 for Solaris 8 sparc, released August 1, added
support to "cp -p" to set file timestamps to microsecond resolution.
A similar patch
<http://mail.gnu.org/archive/html/bug-coreutils/2003-08/msg00016.html>
has been accepted to coreutils but it's only in CVS now.
(This patch has uncovered some glibc bugs in this area: the current
glibc CVS sometimes gratuitously adds a half-second to file
timestamps, which can move them into the future!  See
<http://sources.redhat.com/ml/libc-alpha/2003-08/msg00062.html>.)

However, this only narrows the window down to 1 microsecond; it
doesn't entirely close it.  So we'd still need a 'sleep 0.000001'; but
unfortunately that syntax is only supported in recent coreutils
'sleep' implementations.  And anyway, the vast majority of systems
don't have these recent improvements, deficient as they are.

I'd really prefer being able to copy file timestamps to their full
(currently: nanosecond) resolution, but I don't know of any system
that can do that yet.

The bottom line is that I suspect we'll be stuck with that 'sleep 1'
for another decade at least.




reply via email to

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