#186448 time: %D %K %X %p %t formats are inoperative

Package:
time
Source:
time
Description:
GNU time program for measuring CPU resource usage
Submitter:
Peter Eckersley
Date:
2024-07-07 15:39:02 UTC
Severity:
wishlist
Tags:
#186448#5
Date:
2003-03-27 06:27:22 UTC
From:
To:
Some of time's documented memory tracking features don't appear to work:

pde@luminance:~$ /usr/bin/time -f "%M mem, %K avg, %F/%R pfs" vim /debian/Contents-i386.gz

0 mem, 0 avg, 1039/23403 pfs

Is this possibly kernel-related?  The relevant data structure is in
/usr/include/linux/resource.h, which indicates that some parts of struct rusage
don't exist yet (in which case, should these flags be taken out of the man
page?)
#186448#10
Date:
2003-03-27 06:42:02 UTC
From:
To:
This explains the problem:

http://www.uwsg.iu.edu/hypermail/linux/kernel/0207.1/1022.html

Perhaps Debian's time should implement that LD_PRELOAD hack, and warn
users through the manpage that it's not reliable?
#186448#15
Date:
2003-03-27 12:45:34 UTC
From:
To:
Patches are welcome -- time has been orphaned upstream for years, I'm
afraid.

Dirk
#186448#24
Date:
2003-04-12 21:50:48 UTC
From:
To:
Hi,

I'm Debian's maintainer for GNU time, and received the following bug report.
As time has been de facto without an upstream maintainer for years, I am
taking the liberty of forwarding the report to this list in the hope that
either someone may come forward to work on this, or better still, someone
already is and he may tell me :).

A little bit on an ensuing email exchange between Peter and myself is in the
Debian bug archive at the address http://bugs.debian.org/186448

Regards,  Dirk

----- Forwarded message from Peter Eckersley <pde@cs.mu.oz.au> -----

Envelope-to: edd@edd.debian.net
Delivery-date: Thu, 27 Mar 2003 00:38:58 -0600
Subject: Bug#186448: time: time's memory usage tracking is broken
Reply-To: Peter Eckersley <pde@cs.mu.oz.au>, 186448@bugs.debian.org
From: Peter Eckersley <pde@cs.mu.oz.au>
To: submit@bugs.debian.org

Package: time
Version: 1.7-15
Severity: normal

Some of time's documented memory tracking features don't appear to work:

pde@luminance:~$ /usr/bin/time -f "%M mem, %K avg, %F/%R pfs" vim /debian/Contents-i386.gz

0 mem, 0 avg, 1039/23403 pfs

Is this possibly kernel-related?  The relevant data structure is in
/usr/include/linux/resource.h, which indicates that some parts of struct rusage
don't exist yet (in which case, should these flags be taken out of the man
page?)



----- End forwarded message -----
#186448#27
Date:
2014-02-09 02:28:09 UTC
From:
To:
retitle 186448 time's %D %K %X %p %t formats are inoperative
tag 186448 + upstream
thanks

Peter Eckersley wrote:

Since the original reporting of this bug the ru_maxrss has been added
to the Linux kernel.  In the current program it is being reporting
correctly.  Therefore I think it is no longer accurate to say that
memory tracking is still broken.  Certainly %M works now whereas
before it did not.

However the %D %K %X %p %t formats and possibly others are still
inoperative.  These are not provided by the Linux kernel at this time.
The right place to add them is in the kernel.  It is really an
upstream issue.  In this case upstream is the Linux kernel not the GNU
time upstream.  But they are still documented in the time package.  I
am therefore setting the bug title to reflect the current status.

Bob
#186448#34
Date:
2024-07-07 15:30:36 UTC
From:
To:
Besides "186448 time's %D %K %X %p %t formats are inoperative"

you can add;
-v, --verbose
-o file, --output=file
-V, --Version
--quiet

None of these work on Debian Testing (Trixie) time command line app.

Plus, the manpage on Trixie, does match the manpage at:

https://man7.org/linux/man-pages/man1/time.1.html